Studia Ekonomiczne nr 1-2 2018 Economic Studies no 1-2 2018 by Instytut Nauk Ekonomicznych PAN

INSTYTUT NAUK EKONOMICZNYCH POLSKIEJ AKADEMII NAUK

STUDIA EKONOMICZNE

641008 770239 9

ISSN 0239–6416

ECONOMIC STUDIES nr 1–2 (XCVI–XCVII) 2018

WARSZAWA 2018

STUDIA EKONOMICZNE ECONOMIC STUDIES

INSTYTUT NAUK EKONOMICZNYCH POLSKIEJ AKADEMII NAUK

STUDIA EKONOMICZNE ECONOMIC STUDIES nr 1–2 (XCVI–XCVII) 2018

WARSZAWA 2018

Czasopismo Instytutu Nauk Ekonomicznych PAN

Studia Ekonomiczne RADA NAUKOWA Marek Belka, Barbara Despines, Marian Gorynia, Tamara E. Kuzniecowa, Adam Lipowski, Peter Mihályi, Krzysztof Starzec, Lew V. Nikiforow Komitet Redakcyjny Krzysztof Bartosik (Redaktor Naczelny), Urszula Grzelońska, Joanna Kotowicz-Jawor, Witold Kwaśnicki, Leszek Morawski, Jerzy Mycielski (Redaktor Statystyczny), Adam Noga, Lesław Pietrewicz, Łukasz Piętak (Sekretarz Redakcji), Andrzej Sławiński Redakcja Władysława Czech-Matuszewska Lesław Pietrewicz Opracowanie graficzne i projekt okładki Beata Gratys Wydawca Instytut Nauk Ekonomicznych PAN © Copyright by Instytut Nauk Ekonomicznych PAN, 2018 ISSN 0239–6416 Wersja elektroniczna (e-ISSN 2084–4395) jest dostępna na stronie: http://inepan.pl/wydawnictwa-studia-ekonomiczne Forma drukowana stanowi wersję pierwotną.

REALIZACJA WYDAWNICZA Wydawnictwo Key Text sp. z o.o. 01–142 Warszawa, ul. Sokołowska 9/410 tel. 22 632 11 36, 665 108 002 www.keytext.com.pl wydawnictwo@keytext.com.pl

SPIS TREŚCI EDITORIAL (Wprowadzenie do numeru) – Łukasz Hardt . . . . . . . . . . . . . . . . . .

ARTYKUŁY Jarosław BORUSZEWSKI, Procedural Semantics and Economic Models . . . Łukasz HARDT, Economic Models and Ceteris Normalibus Laws . . . . . . . . . . . Robert MRÓZ, Weberian Perspective on Value Judgements in Economic Models – an Application to Methodological Value Judgements Contained in the Austrian Business Cycle Theory and the Real Business Cycle Theory . . . . . . Krzysztof NOWAK-POSADZY, On Two Modes of Appraisal of Economic Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 41 71 93

POLEMIKA Z TEKSTEM ŁUKASZA HARDTA Marcin GORAZDA, Are the Concept of Capacities and Ceteris Normalibus Clause Redundant? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

ESEJE Marian GORYNIA, O pięknie nauk ekonomicznych . . . . . . . . . . . . . . . . . . . . . . . . . 142

CONTENTS EDITORIAL – Łukasz Hardt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 41 71 93

POLEMIC WITH THE TEXT OF ŁUKASZ HARDT Marcin GORAZDA, Are the Concept of Capacities and Ceteris Normalibus Clause Redundant? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

ESSAYS Marian GORYNIA, About the Beauty of Economic Sciences . . . . . . . . . . , . . . . . . 142

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

EDITORIAL

This special issue of “Studia Ekonomiczne” contains papers focusing on economic models from philosophy of economics perspective. Therefore, it subscribes to a growing philosophical literature on models, namely studies asking questions on how models, being based on unrealistic assumptions, explain; whether models are purely theoretical constructions with interpretations given only ex-post or are models always isolations, namely models of some empirical targets; how models relate to economic laws and how external validity of models can be assessed. One can easily multiply such questions adding more philosophically deep problems, e.g., on how model-based science relates to some more traditional accounts of scientific reasoning, for instance, the one offered by the deductive-nomological model of explanation. Since “models make economics a science” (Rodrik, 2015, 45), then the above questions are of practical importance for economists and for all those interested in what kind of insights about the world economics gives us. Also, as contemporary philosophy of science is less normative than it used to be, its focus is on research practices in particular sciences, and here the one of modelling in economics. So, emphasis is not only put on models as such but also on models’ producers and users. Papers in this special issue of “Studia Ekonomiczne” come primarily from a research effort of a group of economists and philosophers realizing a research grant within the framework of The National Programme for the Development of Humanities.1 This grant focuses on integrated humanistic as a research heuristics with special emphasis put on culture of economic modelling. Thus, J. Boruszewski discusses various issues regarding economic semantics. He suggests that an adequate version of procedural semantics, namely the one based on the pragmatic and operational theory of meaning, can provide a satisfactory method of interpreting economic models. His analysis is illustrated by a detailed study of Irving Fisher’s Cash Loop Model. Next, Ł. Hardt in his paper discusses the idea of ceteris normalibus laws in economics. He offers two understandings of such laws. First, ceteris normalibus clause can simply say that a given statement is only 1 Grant no. 2bH 15 0266 83 which is realized at the Faculty of Economic Sciences at the University of Warsaw (Ł. Hardt and R. Mróz) together with researchers from the Institute of Philosophy at the Adam Mickiewicz University in Poznań (J. Boruszewski and K. Nowak-Posadzy).

EDITORIAL

entirely true in a particular model. Second, such laws can be read not as describing regularities but as the ones referring to capacities and powers. They state what is in nature of a given factor to produce. He opts for the latter reading of ceteris normalibus laws. Subsequently, R. Mróz presents a Weberian perspective on value judgements in economic models which he illustrates by referring to some differences between the models in Austrian Business Cycle Theory and Real Business Cycle Approach. He argues that in order to compare economic models in a thorough way, one should include in such comparisons value judgements expressed or assumed in these models. Finally, K. Nowak-Posadzy discusses two modes of appraising economic models, namely the one based on formal and explicit criteria but also the second one which is more implicit and informal. The latter he describes as art of skillful navigation among models and therefore he refers to Keynes’ and Rodrik’s reflections on art as compared to science of economics. This issue of “Studia Ekonomiczne” contains not only the four papers described above but also two accompanying essays of slightly different character. First, M. Gorazda critically appraises Ł. Hardt’s insights on ceteris normalibus laws. He claims that ceteris normalibus clauses seem to be redundant in interpreting economic models and laws. Second, M. Gorynia reflects on whether economics can be beautiful. His question is in fact more general one since he wonders whether science as such can be described in terms of aesthetics. Not surprisingly, he claims that economics can be beautiful. I would like to thank both M. Gorazda and M. Gorynia for offering their insightful papers for this special issue of “Studia Ekonomiczne”. Now, let me finally thank the editorial board of this journal, and especially prof. Krzysztof Bartosik, its editor, for giving us the opportunity to present our papers to readers of this journal. It should be added that it is not the first time this journal opens its pages for philosophers of economics. In the past, issues 1/2013 and 3–4/2009 of “Studia Ekonomiczne” were entirely devoted to methodology and philosophy of economics. I hope that all six papers from this current issue will provide important insights on models and the ways they contribute to better understanding of economic world. Łukasz Hardt

REFERENCES Rodrik D. (2015), Economics Rules. Why Economics Works, When it Fails, and How to Tell the Difference, Oxford University Press, Oxford.

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

ARTYKUŁY

Jarosław Boruszewski*

PROCEDURAL SEMANTICS AND ECONOMIC MODELS1

ABSTRACT The aim of the article is to present issues regarding economic semantics. The key problem is finding a satisfactory view of semantic interpretation for economic models. The author of the article suggests that an adequate version of procedural semantics can provide such a view. This theory combines semantic interpretation with cognitive and practical activities of subjects which use economic models. Philosophical roots of procedural semantics come from the pragmatic and operational theory of meaning. The article contains a detailed discussion of operationalism in economics and a comparison of the specifics of procedural approach to semantics with the philosophical and methodological characteristics of economic models. The version of procedural semantics used in the analysis is Jan Żytkow’s semantics of operational procedures, in the light of which Irving Fisher’s Cash Loop Model is reconstructed. Directions for further studies of procedural explication of semantics economic models are also outlined, taking into account the methodological specifics of economic sciences. Keywords: philosophy of economics, economic models, economic semantics, operationalism. JEL codes: B41, B13, B00 *

Institute of Philosophy, Adam Mickiewicz University, Poznań; e-mail: borjar@amu.edu.pl This research was financed by a research grant within the framework of the The National Programme for the Development of Humanities (NPRH) (grant no. 2bH 15 0266 83). 1

Jarosław Boruszewski

It is obvious that meaning is not just something that either is or isn’t, but that meanings are generated in time and they become (Bridgman, 1949, p. 258)

1. INTRODUCTION: PROBLEMS WITH ECONOMIC SEMANTICS The semantics of theoretical constructs of economic science and of economic models in particular is a matter which cannot be thematised easily. How do economic models acquire meaning and refer to things which are external to them? Such questions are difficult to answer. However, they are one of the focuses of contemporary philosophy and methodology of economics (Mäki, 2018, p. 4). One of the sources of problems for the semantics of economic models is – as it is emphasised by Uskali Mäki – the semantic scepticism in economics. In its strong version, it is expressed in a view that “models do not even refer to actual economic reality (but rather to some imaginary fiction)” (Mäki, 1999, p. 307). However, a far-reaching inquiry is needed to overcome this semantic scepticism. Undoubtedly, one should pay special attention to the use of key terms in the economic discourse. The analyses carried out by Fritz Machlup (1991) within the field of economic semantics are especially useful in this respect. However, this does not end here. Contemporary semantics has worked out many theories of meaning. Therefore, it is essential to investigate whether it is possible to use these theories to reconstruct the semantics of economic models. Traditionally, the dispute in semantics appeared between the followers of descriptivist theory and causal theory (Barnes, 1982). In short, the descriptivists claim that the extension of the term is fixed through identifying descriptions of sets of manifest properties of objects. Conversely, the causalists presume that this extension is set through causal interactions. First, the object is ‘baptised’ with a given name and then a causal communication chain emerges and links this ‘baptism’ with another uses of the term. However, we refer to objects not because of their manifest properties but because of their essential relations which occur between the objects within the range of a given term. Both theories encounter serious problems in economics. The problem with the descriptive theory is that theoretical models in economics are not adequate descriptions of economic reality and are not constructed to perform this task. As far as the causal theory is concerned, it is impossible to define exact causal relations between objects in the economic reality and models which would be defined with respect to the essential properties of these objects. It turns out then that there is no single mode of fixing reference (Mäki, 1999, pp. 310–315). These traditional semantic theories, collectively referred to as extensionalism or referentialism, are unable to capture the semantics of economic models adequately. On the grounds of philosophy and methodology of economics, referentialism is opposed to inferentialism which is nowadays treated as a more promising view of economic semantics. Particularly, this refers to causal generalisations in economics (Claveau, Mireles-Flores, 2017), but there are also some suggestions

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

to use inferentialism in the semantics of economic models (de Donato Rodriguez, Zamora Bonilla, 2009). In short, according to inferentialism, the meaning of a given unit depends on its role which it fulfils in all the inferences where it occurs. Inferentialism will be discussed in more detail in the final part of the present article as the main purpose of this article is to propose a view of the semantics of economic models with inferentialism being its special case. The proposal is an attempt to set up a new candidate for a promising account of the semantics of economic models. Procedural semantics seems to fit this purpose and – to the best of the author’s knowledge – nobody has considered such an approach before. From a general philosophical perspective, procedural semantics is “an approach to semantics that views understanding in terms of a set of procedures for deciding whether terms apply to things, or procedures for deciding the truth-values of propositions” (Blackburn, 1996, p. 305). Several philosophical sources of the procedural view of semantics can be pointed out. One of them is the pragmatic theory of meaning by Charles S. Peirce, and especially his famous essay How to Make our Ideas Clear. A source which comes second historically and is the most important to the following considerations is the operational theory of meaning by Percy W. Bridgman. The third source can be found in Ludwig Wittgenstein’s late philosophy which was dependent on the connection between the meaning of an expression with the way it is used in a language. These three sources are philosophically related, which is often highlighted in the literature (Hands, 2004; Chang, 2017). Operationalism will be the focus of this article for the following reasons: –– this theory has been and still is considered controversial, sometimes unjustifiably; –– this is an account which closely links semantics with methodology; –– operationalism in economics has its own peculiarity and it needs to be emphasised in order to establish certain semantic postulates which should be taken into account in the proposal of the semantics of economic models; –– as compared to Peirce and Wittgenstein, Bridgman more frequently referred to research practice. There are many procedural semantic theories because “procedural semantics approach is a paradigm or a framework for developing and expressing theories of meaning, rather than being a theory of meaning itself” (Woods, 1981, p. 302). When considering the possibility of the use of procedural semantics in the development of the semantics of economic models, one has to refer to only one version of semantics. In this article, the point of reference is the semantics of operational procedures by Jan M. Żytkow (1944–2001), who was a distinguished methodologist and a long-term collaborator of Herbert A. Simon (e.g. Simon, Żytkow, 1986, 1988). This in a version of procedural semantics in the methodology of sciences,2 and in this respect Żytkow uniquely modified and developed 2 Other

versions have also been constructed in the semantics of programming languages, logical semantics, cognitive science and, the philosophy of language.

Jarosław Boruszewski

Bridgman’s operationalism. Żytkow also directly tackled the problems related to modelling in science, which obviously refers to the subject of this article. The structure of the article is as follows. The next section presents the discussion of operationalism in economics in 20th century with special attention paid to the questions of semantic nature. The third section presents Jan Żytkow’s theory of procedural semantics in the context of the philosophical and methodological characteristics of economic models. The fourth section provides an account of scientific modelling in the procedural perspective and its use to reconstruct a particular economic model – Irving Fisher’s Cash Loop Model. The fifth and final section contains selected possible perspectives of the development of procedural semantics in philosophy and the methodology of economics.

2. OPERATIONALISM IN ECONOMICS Before tackling the problem of operationalism in economics, three important issues have to be considered. The first one is of a general philosophical nature and relates to the fact that operationalism is too easily and frequently conflated with positivism despite the fact that Bridgman was not a positivist (Mirow ski, 1998, p. 347). This conflation often results from a selective treatment of Bridgman’s work, as well as from a limited presentation of his view based only on his early writings. In this respect, a binding statement for these considerations was made by Wade Hands, according to whom “making social science more ‘operational’ was, and is, a legitimate goal, but […] supporting such a statement does not require allowing positivism to define the rules of scientific engagement” (Hands, 2004, p. 965). The next issue relates to the development of operationalism in social sciences. It should be noticed that the first reaction of social sciences to Bridgman’s The Logic of Modern Physics emerged on the grounds of economics. Henry Schultz wrote an article Rational Economics in 1928, where he was enthusiastic about operationalism. Unfortunately, this article did not contribute much to the development of operationalism in the methodology of economics. However, operationalism began developing firmly in other social sciences, such as psychology (Stevens, 1935) and (to a lesser extent) sociology (Lundberg, 1939). It was mainly operationalism in psychology that shaped the discussion of operationalism in social sciences. This was a rather unfavourable situation to Bridgman because it was a source of many misinterpretations, distortions and even violations of the methodology and philosophy of operationalism (Koch, 1992). Therefore, this article will not mention any issues relating to operationalism in psychology as it is a separate subject and a problem in itself. The final issue relates to the scientific work of Paul Samuelson who is considered a representative of operationalism in economics. Strictly speaking, Samuelson was not an operationalist from a methodological perspective. He applied operationalist rhetoric only in his early career and gradually abandoned

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

it. Being unable to clarify this issue at this point (as it requires a separate discussion), let us relate to some statements from literature on Samuelson’s alleged operationalism: –– “his operationalism bears no resemblance to Bridgman’s thesis […]. It is therefore not surprising that Samuelson’s advocacy of ‘operationalism’ caused little consternation among economists” (Caldwell, 1994, p. 190); –– “[t]his invocation of operationalism is somewhat suspect” (Boland, 2008, p. 3); –– “[i]ronically enough, however, this is not operationalism as that term is usually understood” (Blaug, 1992, p. 87); –– “comparisons between Bridgman’s and Samuelson’s operationalism often conclude that the latter’s operationally meaningful theorems were only operational in spirit, but not in Bridgman’s sense” (Carvajalino, 2018, p. 2; emphasis in original). Considering the statements presented above, Samuelson’s research will not be the subject of the following considerations. Instead, Donald Gordon and Julian Simon’s operational analyses as well as Fritz Machlup and Ben Seligman’s criticism of operationalism will be paramount. Gordon’s economic operationalism is particularly crucial for the following considerations not only because – as Blaug claimed – Gordon “makes a more promising effort to pin down the meaning of operationalism in economics” (Blaug, 1992, p. 89), but also because his suggestions are useful in contemporary problems of economic models. What is meant here is the problem of informativeness of economic models (Boland, 1975) and the question of their mathematical complexity (Coelho, McClure, 2005). Gordon’s analyses concentrate on functional relationships among observable variables. However, as Gordon points out, this does not mean that either only operational propositions have scientific value or that all operational propositions should have a functional form. In this respect, Gordon raises an important matter: “What, if any, is the operational significance of such functions? The answer is that the use of such functions must be interpreted as hypothesizing that they are stable, if they are to have operational significance. […] The operational test for the stability of a function is always its ability to predict changes in the dependent variable from changes in the independent variable” (Gordon, 1955, pp. 151–152). When we deal with unpredictable shifts of function, this function is at least in a given moment deemed unstable. This can happen in two situations. The unpredictability of shifts can be caused by an unaccounted observable variable or by factors that are in principle unobservable, such as tastes. This differentiation is pragmatic and not purely operational because it refers to the investigator’s attitude. These two situations are not clearly distinguishable. The unpredictability can be eliminated when the two missing observable variables are taken into account, which can be difficult to achieve. A starting point as defined by Gordon is undoubtedly

Jarosław Boruszewski

quite strict and can make the formulation of valid propositions using functional relationships difficult or even impossible. An operationalist interpretation of economic functions emerges in this respect and this can have far-reaching methodological and semantic consequences for modeling in economics. This interpretation is not based on a presentation of functions with traditional curves but on bands of a certain width. This way of presentation is meaningful in the context of this article and was also used by, for example, Trygve Haavelmo (1939), who in doing so presented not the exact relations, but relations with errors or variations. As a result, additional factors can be taken into account in economic modeling (with its aim of defining the relations between function shifts, e.g. supply and demand), namely threshold and reaction time (Reder, 1952, pp. 185–188).3 Obviously, the relations between those additional factors are defined on the grounds of particular models. There can be a shift considerable enough to influence something else within a short period of time or a relatively small shift which will make a visible change within a longer period of time, etc. In this way, the economic function does not simply show that if x goes up by A units, then y goes up by B units, provided that w and z are constant. Instead, there is a statement that if x goes up by A units, then y goes up by something between B and D units within the T period of time, provided that w does not change by F units and z does not change by G units. The conclusion which Gordon drew from the considerations above is paramount. From an operational point of view, only a small number of statements of the same or similar type as above can be linked together. This can be summarized in the following statement: few versus too many functions. From a general methodological and philosophical perspective – the mathematical complexity of economic models is inversely related to their operationality. At this point, it may be worthwhile to quote Gordon in extenso: “the relationship between x and y may be stable long enough for a shift along that function but not stable enough for a shift along that function plus a subsequent shift along another. With more functions these difficulties multiply in combination. Given a specified band pertaining to a specified interval of time for each function, the more functions there are in the system the cruder will be the valid conclusions – that is, a ‘very large’ shift in an initial exogenous variable will be necessary to insure an unequivocal increase or decrease in each of the dependent variables. (On the other hand, it may be that the larger the shifts along the function the more difficult will it be, because of the possibly longer time lag involved, plausibly to hypothesize sufficient stability over time.) The important conclusion is that, although each function in a system may be operational by itself, the combined functions may not be” (Gordon, 1955, pp. 155–156; emphasis added). 3 In contemporary mathematical economics, function bands are modelled with, for example, fractal geometry: “In complex systems all relationships are time-dependent and exhibit only ‘demi-regularities’. […] Curve is a fractal band that takes up space not a single-valued determinants relationship. The width of the band denotes the variability of the demi-regularity characterizing different complex relationships” (Moore, 2006, p. 312).

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

On the basis of Gordon’s considerations, it is possible to formulate a hypothesis that mathematical complexity of theoretical models is inversely related to their operationality. This hypothesis is currently researched by Philip R.P. Coelho and James E. McClure. In their series of articles (2005, 2008, 2011), they presented content analyses of prestigious economic journals (among others, “American Economic Review”, “Economic Journal”, “Quarterly Journal of Economics”, “Journal of Economic Theory”), and their results support Gordon’s hypothesis (or at least they do not disconfirm it). They observed a trend of increase in professional articles including complex mathematical models at the expense of models with operational meaning. One of the indicators of this trend is a frequent use of the term ‘lemma’ in these publications, and the authors called it the market for lemmas. More pressure is put on the inner coherence of the models rather than on their operational interpretation based on current or historical empirical data. Coelho and McClure provide important statements which develop Gordon’s hypothesis: –– complex economic models rarely operate in empirical world (Coelho, McClure, 2008, p. 78), –– there is a need for “the ‘appropriate’ balance between mathematical complexity and operationalism” (Coelho, McClure, 2011, p. 213; emphasis in original). It has to be clearly stated that these analyses are not aimed against the mathematization of economics. The problem relates to the way of using mathematical theories in economic modelling, particularly whether economists intend to confront mathematical models which they have constructed with economic reality. The mathematization of economics per se is not a problem as long as it does not lead to theoretical emptiness and separation of economics from real economic problems. On the other hand, theoretical insights produced by economic models should be applied to these problems in a modest way (Hardt, 2016a, p. 284). Tradeoff between operationalism and the mathematical complexity of economic models is also crucial for their semantics. As Jan Żytkow puts it in his article with a meaningful title Scientific Modeling: Round Trips to Many Destinations: “the process of model creation […] oscillates between solvability of equations and adequacy of description. An acceptable model is simple enough so that the equations of the model can be solved and complex enough to provide an adequate description of the investigated phenomenon” (Żytkow, 1995, p. 179). The above oscillating character of modeling shows the need to take into account two goals which will not be easily achievable in practice. In Gordon’s view of operationalism, this is related to the width of economic function band: “If economic functions are interpreted as bands, the widths must be specified to make them operational […]. [I]ts usefulness will depend upon how narrow the range is, while its accuracy is likely to depend upon how wide it is” (Gordon, 1955, p. 154). An increase of empirical accuracy by taking into account numerous

Jarosław Boruszewski

additional factors influencing changes of an economic size (‘expanding the band’) can make an analysis with the use of a given function useless (Reder, 1952, pp. 188–189). On the other hand, limiting oneself to the condition of solvability leads to the emergence of uninformative economic models with no empirical content (Boland, 1975, p. 27). Let us proceed to Fritz Machlup’s well-known criticism of operationalism. In retrospect, one can say that his criticism did not undermine the validity of operationalism in economics although it uncovered some of its considerable limitations in its original version. Machlup considered a situation outlined with the following question: ‘What consequences could be expected from the imposition of a tax on imports?’ The concept of tax on import is an operational concept, which is clearly highlighted by Machlup. However, this is not enough to say that the theoretical arguments meant to answer the aforementioned question are operational in nature. This happens because some of the concepts presented in arguments have operational counterparts, and others have not. And so, according to Machlup, terms such as ‘domestic prices of imports’, ‘physical volume of imports’, or ‘quantity of export demanded’ can be operationally defined. Whereas terms such as ‘foreign supply of import’ or ‘domestic demand for imports’ cannot be operationally defined because they are purely theoretical. Therefore, the following conclusion can be drawn: “[w]hat this exercise has shown is that for some of the concepts used in the theoretical argument operational counterparts are available; for others they could be obtained if it were really necessary; for a third group they could not be obtained even with the greatest expense and ingenuity; but that in the theoretical argument itself all concepts were pure constructs, not operationally but nominalistically defined” (Machlup, 1966, p. 65; emphasis in original). The use of quantifiers is undoubtedly crucial to the meaning of the above statement. However, this refers to the Machlup’s general attitude towards operationalism which considerably influenced further development of operationalism in economics. The quantifiers can be summed up in the following way: each/some or – equivalently – only/not only. Of course, this is all about the firm standpoint of classic operationalism, according to which only operational terms are legitimate in science or that every scientific term should have an operational meaning. However, such a standpoint can no longer be sustained and even such a declared economic operationalist as Gordon did not support it. Still, there seems to be a consensus that “for some of the constructs empirical counterparts will have to be suggested” (Machlup, 1960, p. 572; emphasis in original). If so, then is it not a bit premature to state that in the theoretical argument all concepts were pure constructs? This was suggested by Bruce Caldwell (1994, p. 192) who claimed that “such a conclusion is not a necessary consequence of his argument; other conclusions are possible.” In time, Machlup’s criticism does not question the validity of operationalism but indicates the limitations of its early versions.

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

Moreover, Machlup at least partially agreed with Bridgman’s basic and controversial semantic postulate: “the rule ‘different operations – different concepts’ would indeed be a nuisance. […] The trouble, I submit, lies in the failure to distinguish metric concepts – measurable physical or statistical quantities – from concepts of sensory or imagined objects. In the case of numerically determinate quantities it is perfectly proper to insist that different metric operations yield or imply different concepts. But in all other cases different operations may point to, or identify, the same object” (Machlup, 1960, p. 560). Machlup did not question the rule of ‘different operations – different concepts’, although he limited its use. This rule is to be used in measurement and statistical procedures, which is undoubtedly meaningful to the semantics of economic models. Moreover, Machlup expands on this topic and comes up with another statement of paramount importance in economic semantics: “the constructs of the model which have any operational counterparts usually have several such counterparts, each deficient in some way, deviating from the exact (ideal) construct for which it can be only a poor analogue” (Machlup, 1960, p. 573; emphasis in original). Another piece of evidence which proves the importance of Machlup’s criticism in the development of operationalism comes from the fact that Julian L. Simon used operational analysis in his semantic considerations on economics. Simon not only paid scientific homage to Machlup but also incorporated Machlup’s criticism in his own version of economic operationalism. Simon applied operational analysis to study the meaning of many economic terms, such as ‘utility’, ‘causality’ and ‘product differentiation’. In the course of the analysis of the last term, Simon reached some interesting conclusions pointing out to the fact that the use of this term in economic discourse is often confused and misleading. He states that the definition of product differentiation that leads to a preference for one variety of the product over another is defective from an operational point of view. Such a definition of product differentiation shows only one way to check operationally if the product is differentiated: “is a test of whether there is consumer preference (i.e., whether a change in the seller’s offering is accompanied by a change in preference). If differentiation and preference are equivalent in the sense that there is only one measurement for the two concepts, then preference and differentiation are, for all scientific purposes, equivalent. […] the same measurement constitutes the definition of both” (Simon, 1982, pp. 675–677; emphasis in original). Thus, J. Simon inversely applies Bridgman’s dictum: ‘different operations – different concepts’, hence ‘the same operation – the same concept’. Therefore, this is yet another variation of Peirce’s pragmatic maxim: “there is no distinction

Jarosław Boruszewski

of meaning so fine as to consist in anything but a possible difference of practice” (Peirce, 1878/1955, p. 30). Precisely speaking, Bridgman’s line of reasoning was as follows: if there are different operations defining the same term (e.g. ‘length’), then there are different concepts and different terms should be used. However, according to Simon’s analysis, if there is one operational definition applied to two different terms, then the concept is the same and, as a result, one of the terms is redundant. Due to operational equivalence and synonymity of the terms ‘product differentiation’ and ‘consumer preference’, the former should no longer be used as there is less basic one. The use of the term ‘product differentiation’ may lead to confusion (Simon, 1982, p. 676). However, it is not yet clear whether Simon fully agreed with Bridgman’s dictum ‘different operations – different concepts’. Such a conclusion can be drawn from the following statement: “there are several possible measures of price or cost, and it is not obvious which is the best one to look at – cost in man hours, total expenditure as a proportion of GNP, price relative to wages and price relative to other goods. The choice should depend upon one’s purpose, but luckily all of them tend to show much the same result” (Simon, 1982, p. 696; emphasis added). Obviously, the last fragment of the statement above is intriguing as the use of the word ‘luckily’ is slightly enigmatic. But if the above statement is juxtaposed with Machlup’s one in the perspective of economic semantics, then important semantic postulates can be obtained: –– the elements of an economic model which have any operational counterparts usually have several ones (Machlup); –– (luckily) all of them tend to bring much the same results (Simon). As it will be presented in the next section of this article, these postulates are realized in procedural semantics. However, before this happens at the end of this section, it would be worthwhile to mention some crucial points of Ben Seligman’s criticism of operationalism. The key idea of his criticism is the reference to Niels Bohr’s principle of complementarity as “perhaps the most useful antidote to the self-assurance of operationalism” (Seligman, 1967, p. 155). The aim of complementarity is to oppose bringing economics down to the field of what is measureable. Apart from the operational dimension, there is also the complementary interpretational dimension which is equally important. Operational analysis has to be complemented with understanding and comprehension. Methodological complementarity postulated by Seligman is quite important in the discussion of operationalism in economics and can be applied to the aforementioned Machlup’s premature conclusion. From the complementary perspective, one can say that Machlup made economic concepts too theoretical, and the conclusion regarding the complementarity of operational concepts and theoretical constructs would be consistent with the arguments of the author of Economic Semantics. Moreover, this type of complementarity is postulated on the grounds of Żytkow’s procedural semantics, which will be discussed in the next part of this article.

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

However, Seligman investigates complementarity not only in the methodological aspect but also in the epistemological and semantic one and he calls this ‘ambiguity of complementariness’. Seemingly contradictory explanations of economic phenomena are then allowed. In this respect, Seligman refers to Nicholas Georgescu-Roegen’s dialectic concepts: “[a] vast number of concepts belong to this very category; among them are the most vital concepts for human judgements, like ‘good’, ‘justice’, ‘likehood’, ‘want’, etc. […] they are surrounded by a penumbra within which they overlap with their opposites. […] we must accept that in certain instances at least, ‘B is both A and non-A’ is the case” (Georgescu-Roegen, 1966, p. 23; emphasis in original). By referring to the argument above, we could ask whether we deal with ‘B is both A and non-A’, or ‘B is neither A, nor non-A’? The latter option is closer to the operationalist perspective. As far as empirical predicates are concerned, there will always be a certain region of indeterminateness (Carnap, 1936, p. 445) as there are cases when neither a given predicate nor its negation can be attributed. These can be situations relating to the past, where there is a lack of accessible data concerning certain events. We can therefore talk about semantic complementarity without having to engage in contradictory concepts.

3. PROCEDURAL SEMANTICS: AN OUTLINE OF BASIC IDEAS Jan Żytkow’s intention behind his construction of procedural semantics was to overcome the notorious ambiguity which appears on the grounds of classic operationalism – different operational procedures define different concepts and one should use separate terms. In Żytkow’s semantic construction, an empirical term is defined by a collection of operational procedures instead of a single one. If different operational procedures are related to a given scientific concept, then it is possible to use the same term as long as the collection of these procedures is coherent. Otherwise, the defined concept becomes ambiguous (Żytkow, 1984, p. 488). In this way, one can avoid the problematic ‘concept proliferation’ which can lead to an excess of mutually unrelated concepts and make the language of science too complicated. In the context of economic semantics, it would be valuable to compare the starting point of Żytkow’s methodological-semantic strategy with Julian Reiss's diagnosis of the possibility to apply operationalism to the index-number problem. According to Reiss, operationalism: “contains an important grain of truth relevant to the index-number problem. Different measurement methods tend to have idiosyncrasies each of their own, and it is a fallacy to think methods measure the same concept just because we attach the same name to them […]. Operationalism overshoots its target because it tells us to regard concepts associated with dif-

Jarosław Boruszewski

ferent measurement procedures as different independently of whether or not the different procedures yield the same results. But it is right to warn us not to unwittingly use conflicting procedures unless we have investigated the behaviour of these procedures” (Reiss, 2008, p. 71). Thus, Reiss appreciates the critical potential of operationalism, and the aforementioned ‘overshooting of the target’ definitely relates to classic operationalism without touching upon Żytkow’s semantics. Let us recall Bridgman’s controversial remark included in his famous considerations regarding different measure procedures interpreting separate concepts of length (tactile length and optical length). When two procedures of measuring length are available, then using identical names in both cases is practically justified as long as it is not discovered that these procedures do not bring different results within the known scope of their application. Within the margin of error, two different procedures should bring the same results in their common range of application. Yet Bridgman claimed that principally when procedures are changed, a conceptual change also takes place and the use of identical terms for different concepts over the entire range is required for the sake of convenience that can be achieved at cost too high in terms of unambiguity (Bridgman, 1927/1948, pp. 16, 23). In this respect, Reiss was right in saying that classic operationalism overshot the target. A similar thought was expressed by Carl Gustav Hempel, according to whom classic operationalism directly “overemphasizes the need for an unequivocal empirical interpretation of scientific terms” (Hempel, 1966, p. 93). This remark can also be applied to the index-number problem. As Reiss puts it: “We may, of course, investigate the empirical behaviour of COLI and HICP measures and find that they do not systematically differ. Or we may find that they do differ but not enough to matter in the context of specific inquiry. But this is a question regarding a contingent fact about the world and hence requires empirical investigation” (Reiss, 2008, p. 71; emphasis added). In this respect, Reiss and Żytkow’s views somehow overlap. A limine, one can declare neither accordance nor discordance of results of different operational procedures, let alone the degree of accordance or discordance. Semantics cannot a priori settle the identities of concepts. Some more ‘empirical job’ is required in this case. Żytkow expresses this matter in the following way: “not putting up with the limitations of a set of considered concepts, not giving up too soon the search for hidden connections between subranges” (Żytkow, 1984, p. 492). Before we move on to detailed analyses of semantics of economic models from a procedural perspective, it is important to mention some of Mary Morgan’s general philosophical and methodological features of economic models which advocate the accuracy of their interpretation. Żytkow’s accomplishments are important in this respect, as well as the work of other followers of procedural semantics. In her famous study of economic models, Morgan distinguishes their several specific features. First, it is always important to remember that economic models

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

exist in the world and as so for the users they are modes of action. In this respect, economists work with models and even “they move to a point where they no longer use those models to interpret the world, but they see those models at work in the world – the point at which model-designed interventions seem natural” (Morgan, 2012, p. 406). In such a way, a proposition of semantics of economic models should take into account this activistic aspect of economic models. Procedural semantics fulfils this task. Jan Żytkow and Andrzej Lewenstam highlight this matter strongly in an article on modeling in analytical chemistry. In this respect, they oppose semantic reductionism according to which all concepts should be brought down to primitive concepts, and primitive concepts should be reduced to basic laws which act as axioms. However, this leads to a basic problem of the interpretation of formalism – the question of setting the intended interpretation and eliminating the unintended ones. Żytkow and Lewenstam offer an alternative view in the form of defining scientific concepts in terms of operational definitions based on actions and observations: “[t]hese types of definitions, which are usually called operational procedures, are equally important and complementary to theoretical definitions. Rather than reducing other terms to the primitive terms of basic theories, operational definitions reduce scientific concepts to the basic classes of actions and observations that the cognitive agent, the experimenter, is able to carry out” (Żytkow, Lewenstam, 1990, pp. 226–227; emphasis added). The advocates of semantics of operational procedures do not treat procedural interpretation as competitive to theoretical one as they are both complementary. Undoubtedly, this remains in line with the interpretation of operationalism in economics set forward by Seligman. Another important feature of economic models pointed out by Morgan refers to their degree of abstraction or detail. In this respect, she assigns an intermediate status to economic models. “Modelling […] provides economic science with lots of ‘middle level stuff’: in-between, generic-level accounts of what economists take to be typical in economic life rather than descriptions of particulars or very general accounts. Models result both from dividing general accounts and gathering particular empirical cases together” (Morgan, 2012, p. 394). This important feature of economic models has a crucial influence on the construction of their semantics. This means then that this semantics can neither be a formal abstract account nor can it be a fully detailed description of particulars. In this respect, procedural semantics turn out to be promising because operational procedures have an intermediate and hybrid character. The key claim of the advocates of procedural semantics is that referential semantics is too general and abstract to provide satisfactory view of the mechanism which links the language with perceptive activities and non-verbal actions. A semantic interpretation which is founded upon and abstract function of denoting (as in referential semantics) is treated as a ‘black box’. There is no access to the internal

Jarosław Boruszewski

mechanism of semantic interpretation and only the number of inputs and outputs is available. The procedural approach is meant to remove the limits of the referential approach because procedures have a fixed structure. Therefore, semantic interpretation is based on the concept of procedure instead of the concept of function. According to the set-theoretic approach, function is a mapping, and the same mapping can be obtained through a potentially unlimited number of procedures (Woods, 1981, pp. 329–331; Duží, Materna, 2010, p. 218). In this respect, the abstract referential approach does not fulfil the tasks of semantics for models. On the other hand, an overdetailed approach can also be undermined. As far as procedural issues are concerned, the problem refers to the question of the specification of procedures. This problem is also present in classic operationalism. One of Bridgman’s most important postulates described in The Logic of Modern Physics was a statement that every procedure should in principle be uniquely specified (Bridgman, 1927/1948, p. 10). This postulate has become controversial both in the past and more recent literature on this subject. The main concern is that by realizing this postulate, one can get involved in an infinite process of specification of a given procedure, which can be unacceptable in practice. Thus, it is necessary to stick to one general characteristics of procedures (Chang, 2004, pp. 222–223). In his later writings, Bridgman paid attention to the fact that no procedure can be unambiguously specified in every detail – it is not possible to come up with an absolutely precise and unambiguous specification of procedures. One should assume that apart from a certain degree of refinement, a further specification of procedure is not required. Some details are treated as insignificant (Bridgman 1950, p. 10). Therefore, an important conclusion can be drawn for the development of procedural semantics. Resignation from an interpretation based on an abstract denoting function – a certain marginal view of semantic interpretation – cannot lead to a completely opposite situation, i.e. a statement that each detail of the procedure of interpretation is an element of the meaning of an interpreted term. Fully specified procedures are idiosyncratic – private and unique. As a result, special attention is paid to the fact that on the grounds of procedural semantics procedures are abstract, which means some of their elements from the lower level are treated as insignificant to the meaning of a given term (Woods, 1986, p. 59). In Żytkow’s procedural semantics, the matter discussed in this article is handled in a very precise way. He distinguishes two types of procedures: –– type-? – their use to a given number of objects implies a yes/no answer or a certain numeric result; –– type-! – their use creates an object or some state of affairs. On the other hand, the concept of procedure is understood as a finite string of instructions. More specifically, it is a set of commands and questions which end with a terminal instruction determining the result of the procedure. Procedures are carried out on data – the realization of procedures is a transition from initial data to the final ones and it is possible due to specific instructions. Therefore,

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

one can distinguish a set of initial data D0, whereas realized procedures bring sets of final data Dt. An important point for this discussion is distinction: –– procedure-scheme; –– procedure; –– realization of the procedure (Żytkow, 1982, p. 173). The scheme of procedure defines types of instructions (questions, commands, conditional and terminal instructions) and their order. Procedure contains particular functions and operations, but there are also variables, while in the realization of procedure is supplied with constant and particular numeric values in place of variables present in the procedures. As it can be seen, operational procedures are intermediate and stretched between abstract schemes of procedures and their detailed realizations. Naturally, it is possible to go beyond the scheme of procedure in the process of abstraction up to the point where an abstract, ‘bare’ denoting function with no internal structure is obtained – a function which is a pure mathematical mapping. Similarly, it is possible to reach such a degree of detail in the specification of procedure realization where this becomes absurd (e.g. researcher’s shoe size or culinary preferences). All in all, there is no preferable and definite solution to this problem. When ‘middle level stuff’ is handled, then the localization of this level between the outlined extremes will frequently be determined by pragmatic factors. As it has already been mentioned, a set of operational procedures defines the meaning of interpreted terms on the grounds of Żytkow’s procedural semantics. Żytkow assigns this set with the following conditions: –– consistency (or empirical equivalence) – two procedures are consistent if and only if, they apply to the same objects and they give the same results within the limits of error; –– coherence (or nondisjunctiveness) – set of procedures is coherent if and only if, all its subranges are connected with one another by sequences of overlapping subranges (Żytkow, 1984, p. 482). In such a way, basic concepts of procedural semantics are obtained: procedures, data and conditions assigned upon sets of procedures. The fact that a set of procedures interprets a given concept agrees with Machlup’s aforementioned statement: if some, then several. However, it is crucial here to define the relations between the results of different procedures. But this is already an empirical issue and although it may turn out that – as J. Simon claimed – those results will appear luckily consistent, where the word ‘luckily’ can be understood as a contingent of circumstances. What if ‘unluckily’ there is a situation of inconsistency or incoherence? Let us return to the question of indices. The problem of inconsistency emerged in axiomatic index theory. It turns out that Irving Fisher’s criteria called ‘tests’ and set upon indices cannot be fulfilled simultaneously because this would lead to inconsistency. A circular test which lead to inconsistency appeared to be crucial,

Jarosław Boruszewski

so in order to obtain consistency one had to stop using it (Boumans, 2012, pp. 399–401). But even when faced with inconsistency, Fisher did not reject this test and explained this in the following way: “the important question is: How near is the circular test to fulfilment in actual cases? If very near, then practically we may make some use of the circular test as an approximation even if it is not strictly valid” (Fisher, 1922, p. 276, emphasis in original). “In short, while theoretically the circular test ought not to be fulfilled, and shifting the base ought to yield inconsistencies, the inconsistencies yielded are so slight as practically to be negligible” (ibidem, p. 303, emphasis in original). Thus, Fisher’s attitude to index numbers is different from the axiomatic one, and therefore Marcel Boumans calls it instrumental. It is about “finding the best balance between theoretical and empirical requirements, even if these requirements are incompatible” (Boumans, 2001, p. 316). This points out to a quite crucial aspect as to what extent empirical inconsistencies are acceptable – to what number and degree? Perhaps we are simply doomed to inconsistencies, as Bridgman suggested: “I personally do not believe that there is any consistent method for dealing with the complete situation, but that we are forced to a spiralling approximation or to operation on different levels” (Bridgman, 1959, p. 77; emphasis added). This issue is definitely very important for methodology of economics. However, it goes beyond the scope of this article.

4. ECONOMIC MODELLING IN PROCEDURAL PERSPECTIVE: THE CASE OF FISHER’S CASH LOOP MODEL Żytkow was particularly interested in scientific modelling. He also simultaneously worked on the development of his version of procedural semantics and its applications. He considered modelling and construction of operational definitions as distinct scientific activities yet not independent from each other. The construction of operational definitions along with experimenting supports scientific modelling (Żytkow, 1999, p. 311). In order to illustrate this fact, Żytkow used Galileo’s reflections upon the movement of a ball on an inclined plane. The key element of Galileo’s experiment was the fact that he was unable to measure the exact time when the ball would touch the ground. However, he could indirectly measure the final velocity of the ball. To be able to do this, Galileo attached a ‘launching pad’ to the bottom of the inclined plane. The ball would reach the edge of the pad and next fall to the ground and reach the p point. One could then measure the distance between the p point and the q point – the bottom of the slope. Galileo used this result to calculate the velocity of the ball at the bottom of the slope and came up with a model in the shape of an adequate empirical equation. For Żytkow, this is an example of an operational definition (in this case used for the

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

concept of velocity) and at the same time an instance of the application of the model (ibid., p. 315). Similar situations can be found in economic sciences. To show this, it will be useful to outline Żytkow’s explication of modelling perceived as a multilevel feedback process. He distinguished five steps in modelling. When dealing with an object, a process or a phenomenon O, one should: (i) “make a listing of objects, properties and processes present in O; decide which empirical parameters P of O we want to explain and which we can measure in O; a modelling task depends on how much of the modelled objects we want to represent in the model” (ibid., p. 317); (ii) create a model-diagram that captures interaction in O; (iii) construct a model-formalism corresponding to the model-diagram; (iv) simplify and augment the equations until solvable; (v) verify the solution against empirical data for P measured for O. It is important to mention that one does not deal here with a simple one-way sequence of stages. Modelling goes through many feedback loops. This happens when a solution to a problem specific for a given stage requires corrections on a preceding one. Obviously, failures at the last stage need corrections at the earlier stages. It is important to mention that there are models which do not contain formal components and instead come (only) as diagrams. They can be then referred to as informal qualitative models (Gordon, Sleeman, Edwards, 1995). Let us use Żytkow’s explication of modelling to reconstruct Irving Fisher’s Cash Loop Model (CLM). In Fisher’s original article from 1909, three stages of modelling are distinguished: –– first approximation; –– the complete formula; –– statistical application. In the first (nomen omen) approximation, we can say that Fisher’s first stage is equivalent to (i) and (ii) in the above explication; the second stage corresponds to (iii) and (iv), and the last two stages are mutually equivalent. What is important is that there are many loops in CLM construction, which will be presented further. At that moment, two essential remarks have to be made and they refer to the outermost stages of modelling in the context of the methodology of economics. Let us begin with stage (v), where a well-known and difficult problem of testing economic models emerges. However, when the last stage of CLM construction is taken into account, then in the context of the present problematics the pressure is put not on testing the model itself (which is frequently infeasible), but on the test of its application. As Francesco Guala claimed: “[w]hat you can do, though, is to test an application of a model, a hypothesis stating that certain elements of a model are approximately accurate or

Jarosław Boruszewski

good enough representations of what goes on in a given empirical situation. […] A model is always useful to a degree, as long as it is applicable to some situation […] The fact that a model turns out not to work under certain circumstances does not count as a refutation of the model but only as a (failed) test of its applicability in a given domain” (Guala, 2005, pp. 219–220, emphasis in original). If we consider the fact that in the procedural approach there is both application and operational definition, then the above statement is in line with Bridgman’s approach. He claims that if operations are used for a particular purpose, then instead of talking about good or bad operations, one should describe them as either useful or non-useful (Bridgman, 1945, p. 248). When it comes to Żytkow’s first level, important observation should also be made. When researchers make inquiries into empirical phenomena, the phenomena are not directly given to the researchers as parts of the empirical reality, but they are given as objects of inquiry and as such they are always investigated with the help of a certain conceptual apparatus. A prior conceptualization of the field of research appears (Wójcicki, 2002, p. 33). The result of conceptualization is conceptualized empirical system which cover measurement points, time lapses and parameters used to describe a given phenomenon. There is range, period and the characteristics of an empirical system. The parameters chosen to describe a phenomenon can be both quantitative and qualitative. As Ryszard Wójcicki points out, the conceptualization of a phenomenon: –– restricts area of interest and makes it definite; –– suggests certain way of conceiving the phenomenon (Wójcicki, 1979, p. 38). Strictly speaking, it is the conceptualization which defines the object of research from the formal point of view. The same object of interest can be conceptualized in different ways. On the other hand, two separate phenomena conceptualized in exactly the same way are considered identical. In the methodological perspective, conceptualized empirical system is the target and it is modelled. One could ask a question: what is the point of constructing theoretical models if systems already conceptualized are their targets? An important answer comes from Paweł Zeidler: “When the empirical system is conceptualized, then the construction of its model consists in a determination of the relationships between the parameters used for the conceptualization” (Zeidler, 2013, p. 61, emphasis added). This determination can take up a form of adequate equations, but it can also become iconic and presented through diagrams. Let us proceed directly to the CLM reconstruction. The aim of Fisher’s modelling was “tracing the circulation of money, and measuring it by bank records” (Fisher, 1909, p. 605). On the conceptual level, he divides people into three groups: commercial depositors C (e.g. firms, companies), other depositors (e.g. proprietors) O and non-depositors N (wage earners). However, he also admits that this division is not comprehensive as it does not include the subjects whose part in money circulation is negligible (e.g. street traders). Fisher also

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

introduces a demarcation line between C’s and O’s, which separates business self from personal self (e.g. John Smith Shop, John Smith). Transactions happening between these two ‘selves’ are treated the same as normal transactions taking place between companies and different people. As Fisher clearly admits: “Where such a person withdraws money from his till and puts it in his pocket, we may say his business self has paid his personal self some dividends of the business” (ibid., pp. 606–607). Therefore, this is a clear conceptualization of the target of research. The last element of conceptualisation are banks which are observation posts – places where payment flows are registered (Morgan, 2007, p. 117). Thus, it is possible to say that banks become measuring points.4 Next, Fisher moves on to stage (ii) and introduces Cash Loop Diagram. Figure 1. Cash Loop Model Diagram BO

B C

Source: Figure 1 from I. Fisher A Practical Method of Estimating the Velocity of Circulation of Money (1909), p. 608. 4 For the sake of completeness, it is necessary to mention the metaphorical aspect of Fisher’s conceptualization. Banks are homes for money and its circulation is a temporary excursion beyond. To learn more about metaphorizing in economic modelling, see: Hardt, 2016b.

Jarosław Boruszewski

In the next step, there is first feedback and first precise clarification of conceptualization: “As already indicated, money may be said to circulate only when it passes in exchange for goods. Its entrance into and exit from banks is a flow but not circulation. In the diagram the horizontal arrows represent such mere banking operations, not true circulation. The arrows along the sides of the triangle, on the other hand, represent actual circulation” (Fisher, 1909, p. 608). As a result, there are four types of circulation, i.e. exchanges of money against of goods or services: OC, ON, CN, NC. Together with the remaining non-circulative (BO and CB) they create three circuits: –– BO, OC, CB, BO; –– BO, ON, NC, CB; –– BO, CN, NC, CB. An important issue here is the fact that money circulates in the first circuit once out of bank, then circulation OC is essential and it is not present in the remaining circuits, and the whole cycle is closed with CB. Whereas in the two remaining circuits money circulates twice out of bank, because it is paid for wages. The role of intermediaries in the form of non-depositors is crucial here. Payments to N’s flow straight to commercial depositors, so money circulates twice before it returns to bank. It is very important for Fisher’s first approximation: “the total circulation exceeds the total flow from and to banks by the amount flowing through ‘Non-depositors’. In other words, the total circulation in the diagram is simply the sum of the annual money flowing from and to banks and the money handled by ‘Non-depositors’. The quotient of this sum divided by the amount of money in circulation will give approximately the velocity of circulation of money” (Fisher, 1909, p. 609). At this point, there is a transfer to the third stage of Żytkow’s modelling, which is creating an equation for the time being presented in an ideational form. At the same time, there is feedback towards final conceptualization. There are two types of measuring points: the number of B’s transactions and N’s transactions. In the part called ‘complete formula’, one can find Żytkow’s stages (iii) and (iv), which are perforce connected with feedback. There is also feedback to stage (ii) because Fisher introduces a certain additional variant of an initial diagram. This stage of Fisher’s modelling has three main components: the formulation of complete and exact formula of total monetary flow (F) in exchange of goods (all possible transfer within given community), the formulation of an algebraic form of the first approximation and – which is crucial – their mutual comparison which “will express the error of the first approximation, and will suggest a method of transforming the exact formula into a shape more suitable for sta-

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

tistical application” (p. 610). This is stage (iv) – ‘simplify and augment the equations’. Formulas for F and the first approximation (F') have got the following shape:

F = Oc + CO + Nc + CN + ON + NO + Cc + OO + NN,

F' = CB + OB + NB + NC + NO. After adequate transformations and amplifications, Fisher establishes the re mainder r = F – F', which first and foremost sums up the following: –– [(CO + CN) – BC], C’s till-paid commercial expenditures, not withdrawn from bank; –– [(CO + NO) – OB], O’s money receipts pocketed, not deposited in bank; –– (Cc + OO + NN), intraclass monetary flaws. Fisher estimates that in his times the share of remainder r in the USA is rather insignificant and it contributes less than 10% of the total. In this way, an equation for the velocity of circulation V is obtained, where M stands for the amount of money in circulation: V=

) (

)

+ OB + N B + NC + NO + r

. M It is crucial that the part of numerator (denoting bank deposits) is, as compared to others, measureable in the greatest extent. The second part denoting expenditures of non-depositors is also measureable but to a lesser extent, whereas the remainder r is conjectural. At the last stage (v) there is an application of the formula for V. Fisher used statistical data from 1896 available in the USA and calculated that velocity of circulation was 18.6 times a year. Whereas V for the year 1909 was 21.5 times a year. In other words, it was once in 17 days (Fisher, 1911, p. 289).

Figure 2. Fisher’s modelling as a multilevel feedback process clarifying conceptualization modifying diagram transforming formalism

calculation

application

Source: own elaboration.

Jarosław Boruszewski

As it can be seen, Fisher’s modelling can be reconstructed as a multilevel feedback process. Moreover, there is also an application of a model together with its operational definition as it was in Galileo’s case. Galileo was unable to measure the final velocity of the ball because exact and direct procedures of measuring time were not available. He measured the velocity indirectly on the bases of the distance between the point where the ball fell and the bottom of the slope. Obviously, Fisher was also unable to measure the velocity of currency payments directly. Of course, this is a very basic problem. When Md (demand deposits) is available, then estimating velocity Vd is possible, and Md and Vd are treated as variables which are in principle observable. As far as currency is concerned, only Mc (stock of currency) is documented well, whereas the estimation of velocity of currency circulation – Vd – poses difficult and serious problems (Cramer, 1989, pp. 328–330). Fisher’s model provides an operational procedure used to estimate the value of velocity of circulation of currency, although it is rough and indirect (Boeschoten, Fase, 1989, p. 319). In this respect, Morgan synoptically describes Fisher’s way of research and his modeling as: “concerned with classifying all the relevant payments that he wanted to make measurable and then relating them, mapping them, in whatever ways possible, to the payments that he could measure using the banking accounts. He used the visual model to create the mathematical equation for the calculation using the banking statistics, and this in turn used the flows that were observed (and could be measured) in order to bootstrap a measurement of the unobservable payments and thus calculate a velocity of circulation” (Morgan, 2007, pp. 117–118). The four types of circulation are: OC, ON, CN, NC. In order to estimate the value of velocity, one had to “add the amount of money annually withdrawn from bank to the annual money wages” (Fisher, 1911, p. 474) and correct this sum with the conjectural value r. Wages are a sum (ON + CN), while the first component is (CB + OB + NB) which is a sum of flows which are not circulations. This is the most measureable part of the formula for V thanks to which unobservable payments, i.e. (OC + NC), are estimated. Therefore, Fisher point out to the sums of these circulations indirectly – through measureable flows which are not circulations. If we compare this with Galileo’s example which was analyzed by Żytkow as a benchmark, we will see considerable similarities because there is an application of the model together with an operational definition. In Fisher’s own words: “entrance and exit of money at banks, being a matter of record, may be made to reveal its circulation outside” (Fisher, 1909, p. 604; emphasis added). A model alternative to CLM was proposed by Robert D. Laurent in 1970s and was based on using the redemption rates of worn-out banknotes of different denominations. The basic principle behind this model is that “all notes of currency are redeemed if, and only if, they have performed a constant number of transfers” (Laurent, 1972, p. 1173). The obtained results then point out to the

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

fact that the velocity of circulation increases with the decrease of denomination. Several years later, Fisher’s much older model was reused. Interestingly enough, the combination of CLM model and Laurent’s model used by different researchers brought results for the same country which were “surely comparable” (Boeschoten, 1992, p. 37). The following estimations of currency velocity were obtained in the Netherlands: 15.7 in 1977 (J.S. Cramer–Fisher’s model), whereas in 1980 W.C. Boeschoten and M.M.G. Fase (combination of Laurent’s and Fisher’s models) obtained 15.2. Thus, this can be interpreted as obtaining a consistent set of operational procedures interpreting the concept of velocity of circulation. Moreover, results obtained by Fisher and his followers may lead a quite optimistic conclusion: “These results suggest that currency velocity is a constant, as if it were set by physical limitations to the speed of currency circulation, and that it lies between 15 and 20” (Cramer, 1989, p. 331). Obviously, on account of the aforementioned problems, this does not eliminate the difficulties with estimating the value of Vc. However, this provides grounds for Fisher’s results to be treated as partial definitions of the velocity of circulation. An obvious reason for this is the fact that pioneering research of the velocity of circulation has established certain canonical ways how to use this term. However, the meaning of this theoretically and practically intriguing term does not end in the operational procedures of estimating its value. Fisher was fully aware of this fact. The theoretical context is crucial here and it is set by the equation of exchange.5 The complementarity of the theoretical and operational meaning emerges here and, in this respect, Hasok Chang’s perspective of interpretation of operationalism can be applied. “Instrumental measurement operations always need to be placed in the larger context of the assortment of various operations which give to each concept its full meaning. […] We may choose to use instrumental measurement operations to define a concept, thereby privileging them over other kinds of operations, but even then we must keep in mind that a definition is only a criterion by which we regulate the uses of a concept, not the expression of the whole meaning of the concept” (Chang, 2017, p. 29, emphasis in original). Procedural semantics does not fully exploit the meaning of interpreted terms. The range of concepts is always open. However, operational procedures can serve as partial regulatory definitions, which offer an undoubtedly valuable support for the economic modelling. Semantic theories which postulate uniquely defined meaning can easily be undermined from the perspective of research practice. Additionally, the criticism of such maximalist theories can easily lead to a false dilemma: either there is a uniquely defined meaning or not: “That there 5 “To

those who have faith in the a priori proof of the equation of exchange the real significance of the remarkable agreement in our statistical results should be understood as a confirmation, not of the equation by the figures, but of the figures by the equation” (Fisher, 1911, p. 298).

Jarosław Boruszewski

is not one unique determinate meaning to any given proposition or rule does not entail that it can mean anything, but that it has many meanings, and ‘many’ does not equal ‘any’” (Tyfield, 2008, p. 75; emphasis added). Procedural semantics as outlined in this article is resistant to such a dilemma.

5. CONCLUSIONS: TOWARDS REGIONAL SEMANTICS OF ECONOMIC MODELS The aim of the article was to juxtapose referential semantic interpretation with procedural interpretation. The former is expected to define denotation of a term, whereas the latter is expected to tell how the term is to be used. This distinction is very important in philosophy and methodology of economics. Let us take into account the idealizations which often occur in the economic discourse (e.g. perfect competition, complete information, zero transaction cost). How can idealizational concepts be semantically interpreted? As Wójcicki emphasises (using the physical point as an example): “[i]f the interpretation is not merely an abstract one, it may not be available” (Wójcicki, 1995/96, p. 510; emphasis added). If we want to treat idealizational concepts literally, we must state that they denote nothing. As Wójcicki claims further: “As long as we are not able to define the denotations of the term of the analysed object language the interpretation offered is not referential in the right sense of the word. One way or another it involves some procedural elements” (ibid., p. 512; emphasis added). This proves the necessity of at least some elements of procedural interpretation of economic models. If idealizational concepts denote nothing in the empirical world, then one should point out the situations when, for example, information can be treated as complete, i.e. how this term can be used. Let us repeat the following: if semantic interpretation is not merely abstract as the advocates of procedural semantics point out, abstract interpretation is a pure projection with no inner structure and cannot as such be operationally analysed. As it has been mentioned before, we do not come down to fully detailed procedures. There is just a partial analysis of interpretation: “‘level of operation’ may be roughly characterized by the things we leave unanalyzed” (Bridg man, 1959, p. 7). From the foregoing considerations we know that procedures operate on data. But what can be done in a situation when required data are inaccessible? There are many such situations in the real world (e.g. unregistered data referring to past events, confidential or destroyed data). This does not mean that semantics is then useless. Because procedures are represented as a structured entity, they can be used as bases for the inference of what its outcomes would have been in certain circumstances. “As a consequence of these accessibility limitations, it is clear that if procedures are to be taken as explications of meanings, one cannot expect to

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

just blindly execute them. Rather, in some (most?) cases, an intelligent inference component is required in order to deduce useful information from the procedural specification. This in turn dictates that the procedural specifications must be useful for more than just execution as ‘black box’ procedures with input-output conditions. They must have internal structure that is accessible to inferential procedures” (Woods, 1981, p. 325, emphasis added). At this point, procedural semantics meets inferentialism. Interestingly, the followers of inferentialist semantics in economics (in relation to economic causal generalisations) underline the fact that meaning does not diminish in inferentialist relations. Entries and departures transitions are also crucial: –– evidential connection – what the data are expected to say when the statement is accepted; –– policy and research implications – connections with recommendations, imperatives and eventually actions (Claveau, Mireles-Flores, 2017, p. 387). A semantic proceduralist will solve this matter easily. The first situation refers to the application of type-? procedures – measuring or diagnostic, in other words data which can be expected through the realisation of a specific procedure on data initially available. The second situation relates to the use of type-! procedures which command or recommend the realization of a specified state of affairs. Interestingly, inferentialism refers to data and actions only when the ‘semantic power’ of inference runs out. Conversely, procedural semantics refers to inference (inferential procedures specifically) only when data are inaccessible. In a sense, inferentialism and procedural semantics are two sides of the same coin. However, procedural semantics has a considerable advantage and, in this sense, inferentialism is its unique variety (in a situation when data cannot be reached). Particularly, this refers to transfers between targets and models. These transfers are described in the inferential view of economic models in the following way: “what the model has allowed us to do in the end is to derive some conclusions about the empirical system, starting from information extracted from this same system” (de Donato Rodriguez, Zamora Bonilla, 2009, p. 103; emphasis in original). Let us remind that empirical systems are subject to prior conceptualization and only then it constitutes the target. As Zeidler observes: “only a previous conceptualization of the investigated empirical systems allows to identify individual – theoretically interpreted – empirical data (measurement results) as representing specific properties of this object” (Zeidler, 2013, p. 81). We do not begin extracting information from an empirical system because information here is the point of arrival and not the starting point. This happens only when both the applications of the model and its operational definition are successful (Galileo and Fisher’s models). The construction of a model includes feedback loops and also those which can influence initial conceptualization. Then the statement

Jarosław Boruszewski

that this is ‘this same system’ becomes senseless from the procedural point of view. The problem of initial conceptualization, therefore, appears to be crucial to modelling in economics because no conceptualization can be taken for granted: “note that there may not be any obvious way to associate P [phenomenon] with a specific singular empirical system” (Wójcicki, 1994, p. 133; emphasis added). Summing up, it is necessary to mention a problem which requires separate and thorough studies – the specifics of operational procedures in social sciences and particularly in economics. This specification is crucial for the extraction of semantic and methodological peculiarities in economic sciences. Hasok Chang and Nancy Cartwright have come up with some statements which are paramount in the context of the measurement specification in social sciences. “For the purpose of comparisons, measures and measurement procedures are required that can be applied across locations, populations, economies, and cultures. This often results in measures that lose information – measures that are far from the best procedures that could be devised in the separate groups – and the more local measures often give dramatically different results from the more universal ones. Also, for theory-testing we need separate procedures that measure the same univocal concept, but for practical use we generally need a variety of purpose-specific concepts, each with measurement procedures appropriate to it” (Chang, Cartwright, 2008, pp. 373–374; emphasis added). Semantic interpretation is universal when there are no limits set upon the particular uses of the interpreted terms. In other words, this interpretation is not limited to any particular application. Whereas semantic interpretation which is narrowed down to particular uses becomes a local interpretation and as a result terms will be interpreted differently in various applications. The terms will refer to different objects, although as the author of this important distinction points out, “the difference does not necessarily mean that the general ideas underlying such interpretation must be different” (Wójcicki, 1995/96a, pp. 392–393). Universal procedural semantics could prove to be remote from modellers’ purposes. However, this would be against the relevance of its application in economic modelling because – as Boumans observes – assessment of economic models as measuring instruments depends on their validation. The validity of the model is understood as its usefulness with respect to some purpose (Boumans, 2012, p. 420). Moreover, in empirical sciences other than physics fundamental theories are rarely at disposal, so models are built on different theories from different fields of science. Due to the lack of fundamental theoretical background, it is not possible to create a universal semantic interpretation: “interpretation of at least key terms formulating a given model […] is strictly determined by the procedures related to the application of this model” (Zeidler, 2013, p. 45; emphasis added). As it can be seen, the close relation between the application of models and operational procedures comes to the foreground.

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

Does this mean that procedural semantics should have a purely local character? It can be local, although if we consider the middle-level status of economic models and also Żytkow’s semantic construction together with statements by Simon, Machlup and Reiss as well as methodological reconstruction CLM, we can conclude that there is a third possibility. Analogously to regionalist studies, we can call it regional interpretation. The ranges of particular operational procedures can be their own subsets, or they can overlap or also separate. The analogy may be as follows: –– the ranges of particular procedures are regions6; –– the union of these ranges (regions) gives a generalized region; –– intersections of ranges (regions) create subregions (Golledge, Amadeo, 1966, pp. 15–17). The pertinence of this analogy is supported by the fact that Żytkow uses the term ‘distant’ when comparing subranges of concepts. “Subranges are often distant, and their belonging to a range of one concept is determined by scientific theories in which this concept plays an important role. Exactness of formulation and a large degree of empirical conformation of these theories justify a cognation of distant subranges. […] An equivalent of scientific knowledge in a natural language can be defined as common knowledge, social experience, or social beliefs […]. So much weaker common knowledge can connect distant subranges” (Żytkow, 1984, p. 489; emphasis added). The regional approach makes it possible to come up with a double insight into procedural semantics. On the one hand, it has an integrative character – it gives the possibility to look for connections between particular realizations of procedures, establishing relations between ranges and comparing measurement results. On the other hand, procedural semantics has a differentiating character which relates to the comparison of operational procedures in applications across cultures. Then it is necessary to look into some ‘constants’ specific to particular cultures. This was highlighted by Gordon who claimed that, for example, ‘taste’ cannot just be a variable because it would be difficult to imagine a society with no stability in this respect. Taste can then be part of the common social experience. Some parameters are treated as relatively stable, and as a result the number of available functions is limited in a given theoretical model. According to Gordon, mathematical complexity is inversely related to operationalization. Although this may prima facie seem counterintuitive, the necessity of taking into account certain ‘cultural constants’ still supports and does not limit the operationalization of models. In this respect, Gordon was critical of the economists who ‘are apt to take varying constants for meaningful functional variables’ (Gordon, 1955, p. 161; emphasis added). By engulfing these 6 “two regions A and B may have the following relationships to each other: (1) they may have no common territory; (2) they may intersect; (3) A may be part of B; (4) B may be part of A; (5) they may be identical” (Rodoman, 1968, p. 45).

Jarosław Boruszewski

‘varying constants’ – which cannot be treated as ‘fixed variables’ – one makes procedural semantics even less universal. However, it can be regional, and this argument speaks for and not against its soundness.

REFERENCES Barnes B. (1982), On the Extensions of Concepts and the Growth of Knowledge, “Sociological Review”, Vol. 30, No. 1, pp. 23–44. Blackburn S. (1996), The Oxford Dictionary of Philosophy, Oxford University Press. Blaug M. (1992), The Methodology of Economics or How Economists Explain, 2nd Edition, Cambridge University Press. Boeschoten W.C. (1992), Currency Use and Payment Patterns, Kluwer Academic Publishers. Boeschoten W.C., Fase M.M.G. (1989), The Way We Pay with Money, “Journal of Business & Economic Statistics”, Vol. 7, No. 3, pp. 319–326. Boland L.A. (1975), Uninformative Economic Models, “Atlantic Economic Journal”, Vol. 3, No. 2, pp. 27–32. Boland L.A. (2008), Instrumentalism and Operationalism, In: The New Palgrave Dictionary of Economics, The 2nd Edition, S.N. Durlauf, L.E. Blume (eds.), Palgrave Macmillan, pp. 1–4. Boumans M. (2001), Fisher’s Instrumental Approach to Index Numbers, in: Age of Economic Measurement, M.S. Morgan, J.L. Klein (eds.), Duke University Press, pp. 313–344. Boumans M. (2012), Measurement in Economics, in: Philosophy of Economics, U. Mäki (ed.), Elsevier, pp. 395–423. Bridgman P.W. (1927/1948), The Logic of Modern Physics, The Macmillan Company. Bridgman P.W. (1945), Some General Principles of Operational Analysis, “Psychological Review”, Vol. 52, No. 5, pp. 246–249. Bridgman P.W. (1949), The Operational Aspect of Meaning, “Synthese”, Vol. 8, No. 1, pp. 251–259. Bridgman P.W. (1950), Reflections of a Physicist, Arno Press. Bridgman P.W. (1959), The Way Things Are, Harvard University Press. Caldwell B.J. (1994), Beyond Positivism. Economic Methodology in the Twentieth Century, Revised Edition, Routledge. Carnap R. (1936), Testability and Meaning, “Philosophy of Science”, Vol. 3, No. 4, pp. 419–471. Carvajalino J. (2018), Samuelson’s Operationally Meaningful Theorems: Reflections of E.B. Wilson’s Methodological Attitude, “Journal of Economic Methodology”, Vol. 25, No. 2, pp. 143–159. Chang H. (2004), Inventing Temperature. Measurement and Scientific Progress, Oxford University Press. Chang H. (2017), Operationalism: Old Lessons and New Challenges, in: Reasoning in Measurement, N. Mößner, A. Nordmann (eds.), Routledge, pp. 25–38. Chang H., Cartwright N., (2008), Measurement, in: The Routledge Companion to the Philosophy of Science, S. Psillos, M. Curd (eds.), Routledge, pp. 367–375.

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

Claveau F., Mireles-Flores L. (2017), Causal Generalisations in Policy-oriented Economic Research: An Inferentialist Analysis, “International Studies in the Philosophy of Science”, Vol. 30, No. 4, pp. 383–398. Coelho P.R.P., McClure J. (2005), Theory versus Application: Does Complexity Crowd Out Evidence? “Southern Economic Journal”, Vol. 71, No. 3, pp. 556–565. Coelho P.R.P., McClure J. (2008), The Market for Lemmas: Evidence that Complex Models Rarely Operate in Our World, “Econ Journal Watch”, Vol. 5, No. 1, pp. 78–90. Coelho P.R.P., McClure J. (2011), Lotta Lemmata, “Social Science History”, Vol. 35, No. 2, pp. 213–236. Cramer J.S. (1989), Velocity of Circulation, in: Money, J. Eatwell, M. Milgate, P. Newman (eds.), The Macmillan Press, pp. 328–332. de Donato Rodriguez X., Zamora Bonilla J., (2009), Credibility, Idealisation, and Model Building: An Inferential Approach, “Erkenntnis”, Vol. 70, No. 1, pp. 101–118. Duží M., Materna P. (2010), Can Concepts be Defined in Terms of Sets?, “Logic and Logical Philosophy”, Vol. 19, pp. 195–242. Fisher I. (1909), A Practical Method of Estimating the Velocity of Circulation of Money, “Journal of the Royal Statistical Society”, Vol. 72, No. 3, pp. 604–618. Fisher I. (1911), The Purchasing Power of Money: It’s Determination and Relation to Credit, Interest and Crises, Kelley. Fisher I. (1922), The Making of Index Numbers: A Study of Their Varieties, Tests, and Reliability, Kelley. Georgescu-Roegen N. (1966), Analytical Economics. Issues and Problems, Harvard University Press. Golledge R.G., Amaeo D.M., (1966), Some Introductory Notes in Regional Division and Set Theory, “The Professional Geographer”, Vol. 18, No. 1, pp. 14–19. Gordon A., Sleeman D., Edwards P., (1995), Informal Qualitative Models: A Systematic Approach to their Generation, in: AAAI Technical Report SS-95-03, pp. 18–22. Gordon D.F. (1955), Operational Propositions in Economic Theory, “Journal of Political Economy”, Vol. 63, No. 2, pp. 150–161. Guala F. (2005), The Methodology of Experimental Economics, Cambridge University Press. Haavelmo T. (1939), On Statistical Testing of Hypothesis in Economic Theory, www. sv.uio.no/econ/english/about/facts-and-figures/nobel-prize-winners/haavelmo/ published-works/files/th1939e-english.pdf [accessed: 12.10. 2108]. Hands D.H. (2004), On Operationalisms and Economics, “Journal of Economic Issues”, Vol. 38, No. 4, pp. 953–968. Hardt Ł. (2016a), The Recent Critique of Theoretical Economics: A Methodologically Informed Investigation, “Journal of Economic Issues”, Vol. 50, No. 1, pp. 269–288. Hardt Ł. (2016b), On Similarities in Modelling and Metaphorizing Economic Phenomena, “Studia Metodologiczne. Dissertationes Methodologicae”, Vol. 36, pp. 147–174. Hempel C.G. (1966), Philosophy of Natural Science, Prentice-Hall, Inc. Koch S. (1992), Psychology’s Bridgman vs Bridgman’s Bridgman: An Essay in Recon struction, “Theory & Psychology”, Vol. 2, No. 3, pp. 261–290.

Jarosław Boruszewski

Laurent R.D. (1972), Currency Transfer by Denomination, “The Journal of Finance”, Vol. 27, No. 5, pp. 1173–1174. Lundberg G.A. (1939), Contemporary Positivism in Sociology, “American Sociological Review”, Vol. 4, pp. 42–52. Machlup F. (1960), Operational Concepts and Mental Constructs in Model and Theory Formation, “Giornale degli Economisti”, Vol. 19, pp. 553–582. Machlup F. (1966), Operationalism and Pure Theory in Economics, in: The Structure of Economic Science, S.R. Krupp (ed.), Engelewood Cliffs, pp. 53–67. Machlup F. (1991), Economic Semantics, Transaction Publishers. Mäki U. (1999), Representation Repressed: Two Kinds of Semantic Scepticism in Economics, in: Incommensurability and Translation: Kuhnian Perspectives on Scientific Communication and Theory Change, R.R. Favretti, G. Sandri, R. Scazzier (eds.), Bath: Edward Elgar, pp. 307–321. Mäki U. (2018), Rights and Wrongs of Economic Modelling: Refining Rodrik, “Journal of Economic Methodology”, Vol. 35, No. 3, pp. 1–19. Mirowski P. (1998), Operationalism, in: The Handbook of Economic Methodology, J.B. Davis, D.W. Hands, U. Mäki (eds.), Cheltenham: Edward Elgar, pp. 346–349. Moore B.J. (2006), Shaking the Invisible Hand. Complexity, Endogenous Money and Exogenous Interest Rates, Palgrave Macmillan. Morgan M.S. (2007), An Analytical History of Measuring Practices: The Case of Velocities of Money, in: Measurement in Economics: A Handbook, M. Boumans (ed.), Elsevier, pp. 105–133. Morgan M.S. (2012), The World in the Model. How Economists Work and Think, Cambridge University Press. Peirce Ch.S. (1878/1955), How to Make Our Ideas Clear, in: Philosophical Writings of Peirce, Dover Publications, pp. 23–41. Reiss J. (2008), Error in Economics. Towards a More Evidence-based Methodology, Routledge. Reder M.W. (1952), Rehabilitation of Partial Equilibrium Theory, “The American Economic Review”, Vol. 42, No. 2, pp. 182–197. Schultz H. (1928), Rational Economics, “The American Economic Review”, Vol. 18, No. 4, pp. 643–648. Rodoman B.B. (1968), Mathematical Aspects of the Formalization of Regional Geographic Characteristics, in: Economic Regionalization and Numerical Methods, B.J.L. Berry, A. Wróbel (eds.), PWN – Polish Scientific Publishers, pp. 37–57. Seligman B.B. (1967), On the Question of Operationalism, “The American Economic Review”, Vol. 57, No. 1, pp. 146–161. Simon H.A., Żytkow J.M. (1986), A Theory of Historical Discovery: The Construction of Componential Models, “Machine Learning”, Vol. 1, No. 1, pp. 107–137. Simon H.A., Żytkow J.M. (1988), Normative Systems of Discovery and Logic of Search, “Synthese”, Vol. 74, No. 1, pp. 65–90. Simon J.L. (1982), Linguistic Confusion in Economics. Utility, Causality, Product Differentiation, and The Supply of Natural Resources, “Kyklos”, Vol. 35, No. 4, pp. 673–701. Stevens S.S. (1935), The Operational Definition of Psychological Concepts, “Psychological Review”, Vol. 42, pp. 517–527.

PROCEDURAL SEMANTICS AND ECONOMIC MODELS

Tyfield D. (2008), The Impossibility of Finitism: from SSK to ESK?, “Erasmus Journal for Philosophy and Economics”, Vol. 1, No. 1, pp. 61–86. Wójcicki R. (1979), Topics in the Formal Methodology of Empirical Sciences, Reidel Publishing Company. Wójcicki R. (1994), Theories and Theoretical Models, in: Patrick Suppes: Scientific Philosopher, Vol. 2, P. Humphreys (ed.), Kluwer, pp. 125–149. Wójcicki R. (1995/96a), Theories, Theoretical Models, Truth. Part I: Popperian and Non-Popperian Theories in Science, “Foundations of Science”, Vol. 1, No. 3, pp. 337–406. Wójcicki R. (1995/96b), Theories, Theoretical Models, Truth. Part II: Tarski’s Theory of Truth and It’s Relevance for the Theory of Science, “Foundations of Science”, Vol. 1, No. 4, pp. 471–516. Wójcicki R. (2002), An Introduction to Logic. With Elements of Theory of Knowledge, www.jtb.umcs.pl/jtb/papers/rw_aintl.html [accessed: 12.10.2018]. Woods W.A. (1981), Procedural Semantics as a Theory of Meaning, in: Elements of Discourse Understanding, A.K. Joshi, B.L. Webber, I.A. Sag (eds.), Cambridge University Press, pp. 300–334. Woods W.A. (1986), Problems in Procedural Semantics, in: Meaning and Cognitive Structure: Issues in the Computational Theory of Mind, Z. Pylyshyn, W. Demopoulos (eds.), Ablex Publishing Corporation, pp. 55–85. Zeidler P. (2013), Models and Metaphors as Research Tools in Science. A Philosophical, Methodological and Semiotic Study of Science, LIT Verlag. Żytkow J.M. (1982), An Interpretation of a Concept in Science by a Set of Operational Procedures, in: Polish Essays in the Philosophy of Natural Sciences, W. Krajewski (ed.), Reidel, pp. 169–185. Żytkow J.M. (1984), Partial Definitions in Science Compared to Meaning Families in Natural Language, in: Sign, System and Function, J. Pelc, T.A. Sebeok, E. Stankiewicz, T.G. Winner (eds.), Mouton, pp. 479–492. Żytkow J.M. (1995), Scientific Modeling: Round Trips to Many Destinations, in: AAAI-95 Spring Symposium on Systematic Methods of Scientific Discovery, pp. 175–180. Żytkow J.M. (1999), Scientific Modeling: Multilevel Feedback Process, in: Model-Based Reasoning in Scientific Discovery, L. Magnani, N. Nersessian, P. Thagard (eds.), Kluwer, pp. 311–325. Żytkow J.M., Lewenstam A. (1990), Analytical Chemistry: The Science of Many Models, “Fresenius Journal for Analytical Chemistry”, Vol. 338, pp. 225–233.

SEMANTYKA PROCEDURALNA A MODELE EKONOMICZNE STRESZCZENIE Artykuł dotyczy zagadnień z zakresu semantyki ekonomicznej. Głównym problemem jest dostarczenie zadowalającego ujęcia interpretacji semantycznej dla modeli ekonomicznych. W artykule wysuwa się propozycję, że teorią, która może dostarczyć takiego ujęcia jest odpowiednia wersja semantyki proceduralnej. Teo-

Jarosław Boruszewski

ria ta wiąże interpretację semantyczną z poznawczo-praktyczną aktywnością użytkowników modeli ekonomicznych. Semantyka proceduralna jest filozoficznie zakorzeniona w pragmatycznej i operacyjnej koncepcji znaczenia. W artykule szczegółowo przedstawiono dyskusje nad operacjonalizmem w ekonomii oraz porównano specyfikę proceduralnego podejścia do semantyki z filozoficzno-metodologiczną charakterystyką modeli ekonomicznych. Konkretną wersją semantyki proceduralnej, która jest wykorzystana do szczegółowych analiz, jest semantyka procedur operacyjnych Jana Żytkowa. W świetle tej teorii zrekonstruowano wybrany model ekonomiczny – model transakcyjnego obiegu pieniądza (Cash Loop Model) Irvinga Fishera. Wskazano także na dalsze kierunki badań w zakresie proceduralnej eksplikacji semantyki modeli ekonomicznych, które uwzględniają specyfikę metodologiczną nauk ekonomicznych. Słowa kluczowe: filozofia ekonomii, modele ekonomiczne, semantyka ekonomiczna, operacjonalizm. Klasyfikacja JEL: B41, B13, B00

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

Łukasz Hardt*

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS1

ABSTRACT This paper focuses on the nature of economic laws. Rather than conceptualizing such laws in ceteris paribus terms, it claims that economic laws should be read using ceteris normalibus clause, namely that they are only valid in normal conditions. Two understandings of such conditions are proposed. First, economic laws are always true in appropriate theoretical models. Also, the closer a given empirical domain to the model’s structure is, the higher probability that the model’s insights (i.e., economic laws) are to correctly explain the workings of such a domain. Nevertheless, isomorphism between models and empirical domains is never perfect and thus economic laws only describe tendencies in economic realm. Here comes second understanding of ceteris normalibus laws in economics, precisely they do not describe regularities, but they refer to capacities and powers. They state what is in nature of a given factor to produce. Thus, such economic laws are normic laws. While investigating the nature of economic laws this paper also offers a brief study of the history of ceteris paribus clause in economics as well as it refers to an interesting debate on the nature *

Faculty of Economic Sciences, University of Warsaw; e-mail: lhardt@wne.uw.edu.pl This research was financed by a research grant within the framework of the The National Programme for the Development of Humanities (NPRH) (grant no. 2bH 15 0266 83). I would like to thank referees for insightful comments that helped me significantly improve this paper. However, all errors that remain are my own. 1

Łukasz Hardt

of economic models and laws which is offered in D. Rodrik’s 2015 book Economics Rules. Keywords: philosophy of economics, models and laws in economics, ceteris paribus, ceteris normalibus, ontology of social realm. JEL codes: B41, B10, B50 Science, or research activity, never takes place in a philosophical vacuum. T. Lawson (1997, p. 50)

INTRODUCTION Economics is a modeling science, however, on the other hand, economists often refer to laws while accounting for real world phenomena. Therefore, it is worth investigating the interplay between models and laws. What matters also is the kind of entities models and laws are. Since “[…] models make economics a science” (Rodrik, 2015, p. 45), then acquiring knowledge about models’ characteristics can offer us important insights about the kind of science economics is. But not only models and laws matter, but the ways we apply them to study empirical worlds are also essential. If crafting models is a science of economics, then choosing the right model for particular circumstances is an art of economics. Referring to wise words by Keynes is in order here: “Economics is a science of thinking in terms of models joined to the art of choosing models which are relevant to the contemporary world” (Keynes, 1938/1978, pp. 296–297). Nevertheless, the focus of this paper is more on laws than the ways they are used in crafting economic policies. We are to show, however, that discussing laws without referring to models and empirical phenomena is simply impossible. When one discusses economic laws then nearly immediately she is to consider ceteris paribus laws (cp-laws, henceforth). This is nothing new, since “the literal meaning of ‘ceteris paribus’ was dominant in theoretical economics, which is historically the most important place in science where ceteris paribus laws have been used” (Schurz, 2014, p. 1802). Therefore, in economics we find plenty of statements such as, for instance, “ceteris paribus an increase in demand leads to an increase in price” (ibid.); or “all else being equal, lower interest rates tend to raise equity prices” (FED, 2018). However, even being so widely used the meaning of this clause is open to debate, e.g., whether ceteris paribus means ‘all else being equal’ or just ‘other things being absent’ (ceteris absentibus) or even ‘other things being right’ (ceteris rectis). And more fundamentally, one may ask whether such laws describe the very meaning of statements we call laws of economics (Hardt, 2017). We are very skeptical and thus in what follows another interpretation of economic laws is offered, namely the one emerging from dispositional accounts and normality theories. Or, to put it as simple as possible, instead of

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

claiming that, for instance, ceteris paribus lowering interest rates is to lead to higher investments, one should better claim the following: ceteris normalibus lowering interest rates is to lead to higher investments. Such normic interpretation of ceteris normalibus laws will be supplemented by the one treating ceteris normalibus clause as just synonymous to ‘in a model’ restriction and hence, for instance, the above example with interest rates should be read as follows: only in a given model lowering interest rates is to lead to higher investments. This paper is organized as follows. First, some comments regarding history of ceteris paribus clause in economics are offered. We do some history of economic thought here since we just want to better understand what ceteris paribus means for economists using this term. Second, in section 3, we put emphasis on cp-laws in economics. Third, in section 4, the idea of ceteris normalibus laws (cn-laws, henceforth), with its various interpretations, is introduced. Then, in section 5, the role of models in producing economic laws is put under scrutiny and we scrupulously use Rodrik’s (2015) accounts of models in economics. Conclusions follow.

1. HISTORY OF CETERIS PARIBUS CLAUSES IN ECONOMICS Since ceteris paribus laws are so central to economics, we begin below by looking into the very history of cp-clauses. And only later we move forward in order to understand how such laws should be comprehended. In doing so, we refer to various ideas taken from philosophy of science that discuss cp-laws. Finally, we are to show that one cannot successfully defend the usefulness of cp-laws in economics and thus has to go beyond a vision equalizing laws to regularities. And hence our focus on capacities, natures, possibilities, and Aristotelian dynameis. We are not to offer an in-depth study into the long past of cp-clauses but rather we would like to sketch a very brief history of cp-laws.2 In doing so we are to follow historians of ideas writing histories of concepts. Generally, they can be divided into two camps. Firstly, we have intellectual histories, mentalistic ones, or broadly speaking Platonist interpretations of concepts’ past. In such an approach particular expressions (in our case ‘ceteris paribus’) “act as secondary manifestations of an underlying conceptual development” (Klaes and Sent 2005, p. 27). Secondly, in institutional approach to conceptual history we have a greater emphasis put on the discontinuous elements of historical developments. Therefore, we take form as given in order to trace content. So, we are to search for certain classes of words and expressions. Klaes and Sent (2005, p. 28) explain it as follows: “In the institutional approach, a particular trajectory of conceptual development is deﬁned in terms of the (material) continuity of a particular word or expression, rather than the (semantic) continuity of an underlying idea”. 2 Readers

interested in history of cp-clauses can consult, for instance, Persky (1990).

Łukasz Hardt

To put it clear: we claim that it is better to analyze the past in terms of its own words. Such an approach follows from German tradition of Brunner (e.g., 1939) and Koselleck (e.g., 1972a). According to Koselleck, “Concepts are thus the concentrate of several substantial meanings” (1972b/1979, p. 85). In his understanding, concepts function as condensates of historical experiences that are put into single words. Let us hence refer to what Koselleck claims in this respect: “Sense and reference can be thought separately. However, in the case of concepts, sense and reference coincide insofar as the diversity of historical reality and historical experience enters the ambiguity of a word in such a way that it can only receive its meaning in this one word, can only be grasped by this word” (Koselleck, 1972a/1979, p. 120; translated by Klaes, 2001). Nevertheless, concepts are always somehow ambiguous. Also, a given word may change its meaning between different historical époques. Therefore, as historians of economic thought we must resist the temptation to impose contemporary understandings on concepts of the past. In his reading of Koselleck’s ideas Klaes (2001) refers to F. Nietzsche’s phrase from Genealogy of Morality (1981, p. 820), namely that “All concepts which semiotically comprise an entire process escape definition; only that which has no history is definable”. So, we may somehow empirically draw evolving conceptual schemes of given concepts, and in this paper we are to apply this method to ‘ceteris paribus’. In other words, we try to trace changing meanings of ceteris paribus. We are however to do it in two steps. First, we show the process of institutionalization of ceteris paribus, namely the “process by which an individual expression achieves the status of an institution, and hence can be regarded as a Koselleckian concept” (Klaes, 2001, p. 161). And only then we are to study systematic ambiguity of ceteris paribus, namely its various meanings. The fact that this concept is somehow vague is obvious to many historians and philosophers of economics (see e.g., Persky, 1990; Mäki and Piimies, 1998). In order to construct a conceptual field of ‘ceteris paribus’, we have to formulate a list of key words and phrases related to historical and current interpretations of ‘ceteris paribus’. In the next step our goal will be to study the frequencies of these sentences/words in economic literature. And only then we have to explain why some of them were on the rise in particular decades while others are now simply forgotten. What should be the key for choosing our list of phrases serving as interpretations of ceteris paribus? Here we can just start by referring to A. Marshall’s propositions regarding the usage of ceteris paribus, so let us cite two passages from his Principles of Economics, i.e.: “It is sometimes said that the laws of economics are ‘hypothetical’. Of course, like every other science, it undertakes to study the effects which will be produced by certain causes, not absolutely, but subject to the con-

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

dition that other things are equal, and that the causes are able to work out their effects undisturbed. Almost every scientific doctrine, when carefully and formally stated, will be found to contain some proviso to the effect that other things are equal: the action of the causes in question is supposed to be isolated; certain effects are attributed to them, but only on the hypothesis that no cause is permitted to enter except those distinctly allowed for” (Marshall, 1920/2013, p. 30). “Corresponding to the substantive ‘law’ is the adjective ‘legal’. But this term is used only in connection with ‘law’ in the sense of an ordinance of government; not in connection with ‘law’ the sense of a statement of relation between cause and effect. The adjective used for this purpose is derived from ‘norma’, a term which is nearly equivalent to ‘law’ and might perhaps with advantage be substituted for it in scientific discussions. And following our definition of an economic law, we may say that the course of action which may be expected under certain conditions from the members of an industrial group is the normal action of the members of that group relatively to those conditions” (ibid., p. 28). The very first above citation refers to constancy of causes and the second one to normality of factors influencing what we try to explain. So, we have ceteris paribus and ceteris normalibus clauses. Also, Marshall’s ‘other things being equal’ can give rise to ceteris absentibus assumptions. Cn-clauses mean that a given statement, e.g., when you rise cost of money, then you are to have less investments, is only true in normal conditions. One interpretation of such conditions may be to say that the above statement about money and investments is only true in a particular economic model. In the table below we therefore propose some words and sentences related to the three above proposed understandings of ceteris paribus conditions. Table 1. Conceptual field of ceteris paribus and number of papers in “The American Economic Review” a given phrase appears (till 2009) Ceteris paribus (983) other things being equal (285) other things being constant (3)

Source: own research.

Ceteris absentibus (0) being absent (14)

Ceteris normalibus (0) in normal conditions (3) in normal circumstances (13) in normal times (177) being normal (4) natural setting (18) in natural conditions (3) in model (including in a/the model) (3023) model conditions (6)

Łukasz Hardt

Now our task should be to check to what extent the above presented sentences forming the conceptual field of ceteris paribus are present in economic literature in different decades. Here, for the sake of simplicity, we just put emphasis on papers from “The American Economic Review”. This particular journal can serve as a good proxy of what economics is (cf. Anderson et al., 1986)3. In our sample we include all papers from “The AER”, containing the ones from its Papers and Proceedings section as well as books’ reviews. Also, we consult the Oxford English Dictionary for meanings of words and their changes. In the very first step we put in the brackets in the above table number of papers in “The AER” a given phrase appears. Next, we are interested in how fast a given phrase proliferates in economics. We show it below, but we restrict our presentation only to sentences with the highest number of appearances. Figure 1. Number of papers in “The AER” consisting of elements from the conceptual field of ceteris paribus 800 700 600 500

other things being equal

400

in normal times

300

ceteris paribus

200

in model

100 0 till 1919 20–29 30–39 40–49 50–59 60–69 70–79 80–89 90–99 00–09

Source: own research.

We have a number of insightful observations. First, we did not notice any appearances of ceteris absentibus and ceteris normalibus clauses as such. However, 3 Historians of economics often look into discipline’s most prestigious journals in order to discover important patters in development of economics. “The AER” is the official journal of the American Economic Association. Nevertheless, it is important to know that its place in the profession has changed substantially. First, as Backhouse (1998) claims, now papers from this journal have much greater prestige than in the 20’s or the 30’s. They play a crucial role in economists’ promotion procedures. What can be simply difficult to imagine now, the editor of “The AER” in the 20’s was always afraid of not having enough good material for publication. Second, before the Second World War economists used to publish their work in many general, non-economic journals (e.g., “The Annals of the American Academy”). Third, from the 30’s on some more technical journals started to emerge, e.g., “Econometrica” (in 1933) and “Review of Economic Studies” (also in 1933). And fourth, now we have many journals in economics specializing in particular areas of economic research. Nonetheless, “The AER” can still serve as a credible measure of what economists do and what methods they use.

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

we have a growing number of papers consisting at least one occurrence of ceteris paribus. And, as Mäki and Piimies (1998) claim, ceteris paribus may mean also ceteris normalibus and ceteris absentibus. So, what ceteris paribus means in its various usages? The best way to answer this very question is to investigate the reasons for an explosion in its use starting in the 40’s (8 appearances in the 30’s; 52 in the 40’s, and 92 in the 50’s) and then a relative stagnation in the 80’s and next a sudden drop in the 90’s followed by a relative stagnation at the very beginning of the new century. Knowing that correlation does not equal causation it is however interesting to know with what kind of changes ceteris paribus proliferation is correlated. Even from the above graph it is clear that an upsurge in the number of papers consisting ceteris paribus was accompanied by a similar rise in a number of papers containing ‘in model’ sentences (4 appearances in the 30’s; 23 in the 40’s, and 76 in the 50’s).4 Thus, maybe ceteris paribus just meant ‘in model’ at least in the three mentioned decades? Before answering this question, it is worth offering some insights on why such an upsurge in references to models in economic theories occurred in the 40’s and the 50’s. It is first of all worth noticing that the above-mentioned years witnessed a profound change in method and language of economics. Blaug explains it in the following way: “The metamorphosis of economics in the late 1940s and 1950s is aptly called a ‘formalist revolution’ because it was marked, not just by a preference, but by an absolute preference for the form of an economic argument over its content. This frequently, but not necessarily, implied reliance on mathematical modeling because its ultimate objective was to emulate the notorious turn-of-the-century Hilbert program in mathematics by achieving the complete axiomatization of economic theories” (Blaug, 2003, p. 145; emphasis added). So, mathematics entered economics and transformed it into a modeling science. Such a development was strongly supported by publication of P. Samuelson’s Foundations (1947) as well as by the works of K. Arrow, R. Debreu, J. Neumann, O. Morgenstern, J. Hicks, and many others. Formalism in economics in the 40’ and the 50’s was generally understood as “a methodological requirement to set up any theory as a formal system” (Kesting, Vilks, 2004, p. 286), namely as a particular model but how models and theories relate to one another? Here we have two general approaches. First, according to the syntactic view of theories, a model is an interpretation of a given calculus, e.g., the billiard balls are a model of the kinetic theory of gases.5 Supporters of such an approach treat models as being useless to science (e.g., Carnap, 1938). Second, we have the semantic view of theories which “declares that we should dispense with a formal calculus altogether and view a theory as a family of 4 Including

‘in the model’ and ‘in a model’. put it more precisely one has to reinterpret the terms in mathematical calculus of kinetic theory of gases and make them refer to billiard balls. 5 To

Łukasz Hardt

models” (Frigg and Hartman, 2018). It seems that in the very first years of the formalist revolution economists’ practice can be described by the syntactic view of theories. For instance, Debreu (1959, x) claims the following: “It […] makes possible immediate extensions of the analysis without modification of the theory by simple reinterpretation of concepts” (emphasis added) and Alchian (1955, p. 942) offers the following remark: “Fisher’s theory is a model for deriving propositions about how the economic system operates”. Thus, a given economic theory after reinterpretation serves as a model of a particular economic realm. So no one should wonder why formalism in economics is so much connected to references to models. Is it now right to claim that also ceteris paribus in the years of the formalistic revolution meant just ‘in model’? Let us thus look into some sentences from “The AER” in the 50’s containing such expressions, i.e.: “Although the use of ceteris paribus assumptions is thus necessary and useful part of analytic method, it must be admitted that it can, in certain circumstances, limit the application of the theory to such a degree that no meaningful empirical analysis can be made. When the concepts of value theory are defined in such a manner that they are observable only in very special circumstances, the error term may be so large that empirical testing can only be inconclusive […]” (Ruggles, 1954, pp. 145–146; emphasis added). “[…] we cannot in full logical consistency draw up a demand curve for investment by varying only the rate of interest (holding all other prices in the impound of ceteris paribus)” (emphasis added); and in the footnote to this sentence he adds: “This is exactly analogous to the distinction between the Marshallian partial equilibrium demand and the Walrasian general equilibrium demand discussed by M. Friedman (1949). In the present context the partial analysis curve misses the essence of capital theory, the relationship between interest rates and the price structure” (Alchian, 1955, p. 942). “Ceteris paribus, an «economic unit» will ordinarily purchase and hold a larger quantity of a durable good for future use , the lower is its current price in relation to the prices of other assets. Hence, the durable good demand curve of a single economic unity will normally slope downwards from left to right […]” (Clower, 1954, p. 65; italics in original; emphasis added). The last citation from Clower (1954) clearly links ceteris paribus with normality assumption. Also, later in his text reference is made to a very specific model, namely a graph presenting shapes of demand and supply curves.6 But what 6 Graphs are generally seen as illustrations of models but since “[…] geometry is a branch of mathematics” (Samuelson, 1952, p. 59, while discussing the relative merits of algebra and geometry in economics) we may simplify a bit and treat graphs as models.

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

about our two preceding citations? Let us start with the one from Ruggles. Here we have a very clear reference to what philosophers of science describe as Hempel paradox, namely that cp-laws are either false (because disturbing factors occur) or they are trivially true (if they are understood as purely analytical statements being true only in models, precisely special theoretical worlds constructed in such ways as to satisfy these laws). Ruggles’ “very special circumstances” may be treated as a clear reference to model conditions and his remark about inconclusiveness of empirical testing as his acknowledgment of general problems with empirical testing of cp-sentences. Now, what Alchian’s “holding all other prices in the impound of ceteris paribus” means? Here he does not simply claim that all other prices are constant (ceteris paribus) but that they are “in the impound of ceteris paribus” and it makes an important difference, since “in the impound” may be read as in a place where prices do not change, and if such a place is empirically impossible then it should be a theoretical one, namely a model. Although the above analysis suggests that cp-laws are just laws referring to particular models (i.e., to normal/model conditions), it is not legitimate to generalize and claim that all cp-clauses in economics during the first stage of formalistic revolution had such characteristics. Many authors used cp-restrictions in informal ways not referring to any special models or graphs. In the same vein, many papers while pointing out to normal conditions did it in informal ways, namely by restricting a domain of a given theory to non-extraordinary times. Coming back to cp-laws, it should be stressed that even those using them were conscious that there is a problem of applying cp-laws in empirical domains. How, then, empirical usefulness of cp-laws may be defended? We try to answer this question in the section below.

2. UNDERSTANDING CETERIS PARIBUS LAWS IN ECONOMICS History of science knows many attempts at defending cp-laws. Let us just refer to the two most popular ones. First, the method of completers may be used, precisely one has to add the missing conditions into the antecedent of the law statement and thus the aim is to have a strict law. Second, cp-laws may be understood as statements about tendencies. Even intuitively it is rather obvious that listing all factors that are claimed to be constant in antecedents of cp-laws is simply impossible and thus the method of completers is not very useful in defending such laws.7 Now what about tendencies? Here, for instance, saying that ceteris paribus lower interest rates should lead to higher investments may be reformulated in such a way: lower interest rates should produce a tendency for investments to rise. Such interpretation of economic laws, including cp-ones, has a long 7 For

more philosophical arguments, see e.g., Reutlinger et al. (2017).

Łukasz Hardt

history in economics. It was John S. Mill who strongly supported it. He claims, for instance, the following: “With regard to exceptions; in any tolerably advanced science there is properly no such thing as an exception. What is thought to be an exception to a principle is always some other and distinct principle cutting into the former: some other force which impinges against the first force and deflects it from its direction. There are not a law and an exception to that law—the law acting in ninety-nine cases, and the exception in one. There are two laws, each possibly acting in the whole hundred cases, and bringing about a common effect by their conjunct operation. […] Thus if it were stated to be a law of nature, that all heavy bodies fall to the ground, it would probably be said that the resistance of the atmosphere, which prevents a balloon from falling, constitutes the balloon as an exception to that pretended law of nature. But the real law is that all heavy bodies tend to fall […]” (Mill, 1836/2008, p. 56; emphasis added). So, one can still claim that there is a tendency of X to produce Y even though in a given case X gave rise to Z. As Mill put it: “All laws of causation, in consequence of their liability to be counteracted, require to be stated in words affirmative of tendencies only, and not of actual results” (Mill, 1843, p. 523). Reiss (2013, p. 93) adds that “A tendency claim is a claim about a regularity that would hold if disturbing factors were absent”. Speaking less formally one can say that generally X produces Y but in particular circumstances it may not be the case. So, we can now introduce notions of types as generic causal facts and tokens as singular causal facts. According to Cartwright (1989), for instance, in traditional Humean interpretation tokens are true in virtue of types; and types are regularities. On the other hand, in structural account of causation types are first and tokens only later, as it is in Hume’s theories, however, now contrary to Hume, types are understood as causal structures. Cartwright proposes even a further depart from Hume and for her tokens are crucial and thus her support for singular causation. We are to come back to Cartwright in the next section. But now let us comment on whether probabilistic approach to causation somehow solves the problem of how tokens and types are interrelated. In his seminal book on causation in macroeconomics K. Hoover offers such a formalization: “[…] an advocate of a probabilistic account might hold that a (token-)cau ses b, when a occurs and b occurs and A (type-)causes B. By “type A causes type B”, probabilistic accounts mean P(B=b|A =a) ! P(B=b)” (Hoover, 2001, p. 72). But what if one has a occurring but she is not seeing b? According to the above Hoover’s claim it is not problematic since failure of tokens to illustrate the relationship at the level of types does not threaten the causal relation at the generic level. Nevertheless, taking type level causation as primary and token level

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

as secondary leads us to some counterintuitive cases. Take, for instance, the following one due to D. Rosen and cited by Suppes (1970, p. 41): “[…] suppose a golfer makes a shot that hits a limb of a tree close to the green and is thereby deflected directly into the hole, for a spectacular birdie. If we know something about Mr. [sic] Jones’ golf we can estimate the probability of his making a birdie on this particular hole. The probability will be low, but the seemingly disturbing thing is that if we estimate the conditional probability of his making a birdie, given that the ball hit the branch, we would ordinarily estimate the probability as being still lower. Yet when we see the event happen, we recognize immediately that hitting the branch in exactly the way it did was essential to the ball’s going into the cup”. Hoover (2001) while commenting the above example stresses the fact that striking the limb lowers the chance of the birdie and at the same time it is the limb which caused the birdie. Here we see that introduction of the distinction between types and tokens helps us to cope with situations where intuitively a given factor causes something but at the same time it lowers its probability. What is also important to stress here is that we may have type causation without any instances of tokens, say lowering interest rates (type-)causes higher investments, and this very statement can be true even without any empirical cases of lower interest rates making investments higher. We can now try to reformulate cp-statements as follows: X (type-) produces Y. And to give a very simple example: diminishing interest rates (type-) produces inflation. It is somehow similar to the above Mill’s understanding of laws as statements of tendencies only. As a given tendency may be dormant and thus it is not to produce a particular (or anticipated) result, in a similar vein we may have type-causation between A and B without any manifestations of tokens a and b. Before going further to Cartwright’s ideas placing tokens at the very first place, we have to refer firstly to Hausman’s objections to the above presented types and tokens thinking and secondly (but very briefly) to other (usually failed) attempts at understanding cp-clauses. Hausman’s claim is relatively simply stated: generic-level causation cannot be primarily since causation does not reduce to relations between variables, because they are not situated in space and time. For him, causation relates only to particular aspects of events, namely tropes as he calls them. In his own words: “tropes are particulars located in space and time” (Hausman, 1998, p. 26; emphasis in original), and next he adds: “Causal relations among events and explanatory relations among facts obtain in virtue of the relations that obtain among simple tropes” (ibid., p. 26). In his interpretation laws of nature are not causal but they only link/relate variables or proprieties. In Hausman’s approach, according to Hoover (2001, p. 83), “Type-level causality generalizes the individual instances of token-level causality”. And token-level causality is possible because tokens are sets of tropes. Now one can try to understand ceteris paribus laws in the following way: X causes Y under condition that

Łukasz Hardt

a given trope x' of x causes y' of y. However, we have a problem here because causation between x' and y' is done by the existence of law of nature linking proprieties X' and Y ' instantiated by x' and y'. It is not a very attractive way of understanding cp-laws since here laws of nature are necessary and if they do not exist, as it is claimed in Hardt (2017), then we have a problem. Here it is even difficult to give a real life example of such understanding of cp-laws in economics. What is however important is that Hausman proposed an approach denying the central role of type-level causation. How it refers to Cartwright’s ideas of singular causation is analyzed in the forthcoming section. Now let us comment on another promising way of understanding cp-laws, namely the one originating from invariance and stability theories. Here, for instance, Woodward (2000, p. 197) claims the following: “According to this alternative, whether or not a generalization can be used to explain has to do with whether it is invariant rather than with whether it is lawful […]. Unlike lawfulness, invariance comes in degrees […]”. So, if one has the following generalization: once you cut interest rates, then you have a rise in investments; then the greater the range of values the ones present in this generalization can take, the higher degree of its lawhood. But not only variables in a given generalization matter for its stability but also the ones describing background conditions. So, interventions can be carried out with respect to these two kinds of variables. To illustrate his point Woodward gives an example from economics: “In microeconomics, individual economic agents are often assumed to conform to the behavioral generalizations constituting rational choice theory (RCT). […] Even if we assume, for the sake of the argument, that these generalizations are roughly accurate descriptions of the behavior of many participants in markets, it is clear that there are many changes and interventions over which the generalizations will fail to be invariant. For example, there are many pharmaceutical interventions and surgically produced changes in brain structure that will lead previously selfish agents to act in non-self-interested ways […]. However, economists have not generally regarded these sorts of failures of invariance as interesting or important, at least if […] they occur relatively rarely in the population” (Woodward, 2003, p. 263; emphasis in original). It seems that the above Woodward’s approach correctly describes research practice in economics which often comes down to formulating generalizations but the ones not excluding exceptions. However, on the other hand, we know that the majority of generalizations in economics fail not due to some “surgically produced changes” but they do not hold due to some even minor modifications in background conditions. As Reutlinger et al. (2017) comment: “By being nonstrict the generalizations in the special sciences do not satisfy a condition that is traditionally associated with laws of nature, namely the condition of universality. Nonetheless being invariant for a limited range of values is enough for a proposition to play a lawlike role in the sciences”. So, we come back to some

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

well-known problems with cp-laws, since calling them lawlike statements does not help us much. In philosophy of science we find also a very similar approach to cp-laws as the one described above. Here I have in my mind Lange’s framework referring to counterfactuals where “Some proposition L is a law iff its truth is preserved under all those counterfactual suppositions that are consistent with every physical necessity, i.e., under all physically possible counterfactual suppositions” (2009, p. 20). In economics we do not have such laws or to put it differently, laws in economics are true only in artificial words (models) where everything is under control and the modeler knows every “possible counterfactual suppositions”. Beyond models we have only cp-laws having different degrees of lawhood. But here Lange suggests that scientists formulating cp-laws in special sciences, including economics, are not obliged to describe all interfering factors. They should however list all factors they are interested in, or to be precise, their scientific discipline put emphasis on. In his own words: “[factors] that arise sufficiently often, and can cause sufficiently great deviations from [a given generalization]” (Lange, 2002, p. 411). Interestingly, he refers here to Haavelmo’s discussion on “the degree of permanence of economic laws”. A rationale for looking into the ways T. Haavelmo understood economic laws, including the ones with cp-clauses, stems from the fact that he was one of the most important figures in popularizing mathematical modeling in economics. So, in his seminal paper on The Probability Approach in Econometrics (1944) he offers us the following insight: “No matter how much we try and fail, we should never be able to establish such a conclusion as ‘In economic life there are no constant laws’. We shall consider a much more restricted problem, namely this: How far do the hypothetical ‘laws’ of economic theory in its present stage apply to such data as we get by passive observations?” (Haavelmo, 1944, p. 16). His skepticism towards the existence of laws of nature (or constant laws) is due to his claim that laws in sciences are constructed rather than discovered. Or, to put it differently, one formulates such laws using, for instance, mathematical models and only then they are empirically tested. Therefore, the fact that in physics (contrary to economics) we have constant laws “means not much more and not much less than this: The natural sciences have chosen very fruitful ways of looking upon physical reality” (ibid., p. 12). So, laws, including the ones in economics, are only statements that imperfectly describe empirical phenomena and hence Haavelmo’s interest in treating them as cp-laws. And here his focus is on analyzing the very possibility that “simple laws in economics rests upon the assumption that we may proceed as if such natural limitations of the number of relevant factors exist” (ibid., p. 24). By irrelevant factors he treats the ones having limited impact on what we are explaining as well as factors that can potentially matter but that are constant. And now we know why M. Lange in his papers on cp-laws refers to Haavelmo’s observations.

Łukasz Hardt

Before moving further to some more detailed insights into the ways ceteris normalibus laws can be conceptualized, let us here just add that Haavelmo’s treatment of economic models strongly supports my claim that a rise in the number of papers using ceteris paribus clauses in the 50’s and the 60’s is due to popularization of mathematical modeling in economics. Therefore, ceteris paribus, at least in these years, meant just “in a model”. The following beginning of Haavelmo’s first chapter of his The Probability Approach in Econometrics is illuminative in this respect: “Theoretical models are necessary tools in our attempts to understand and ‘explain’ events in real life. In fact, even a simple description and classification of real phenomena would probably not be possible or feasible without viewing reality through the framework of some scheme conceived a priori. Within such theoretical models we draw conclusions of the type, ‘if A is true, then B is true’. Also, we may decide whether a particular statement or a link in the theory is right or wrong, i.e., whether it does or does not violate the requirements as to inner consistency of our model. As long as we remain in the world of abstractions and simplifications there is no limit to what we might choose to prove or to disprove” (Haavelmo, 1944, p. 1). And after offering us such important insights he referred to worth citing passage from V. Pareto: “There is no proposition that cannot be verified under certain specific conditions. The conditions of a theorem are an integral part of the theorem and cannot be separated from it” (Pareto, 1906/2014, p. 5). So, Haavelmo simply claims that in models “there is no limit” to our imagination; or, in other words, for every theoretical claim we can offer a model in which such a statement is to be true. Therefore, ceteris paribus stands for ‘in a model’ clause. The same is in Pareto’s case where he uses the term “specific conditions” as synonymous to models. Nevertheless, the above second sentence from Pareto’s citation is worth commenting on. He claims that one cannot separate “the conditions of a theorem” from a theorem as such. It should be read as a claim that a given theorem is perfectly true only in very specific circumstances, namely in “its” normal conditions.8 8 As

Hardt (2017) documents it, in a rather anecdotal way this issue was nicely portrayed by Mises saying the following: “Once, during a speech which he was making at a statistical congress in Bern, Pareto spoke of ‘natural economic law’, whereupon [Gustav] Schmoller, who was present, said that there was no such thing. Pareto said nothing, but smiled and bowed. Afterward he asked Schmoller, through one of his neighbors, whether he knew of an inn where one could eat for nothing. The elegant Schmoller is supposed to have looked half pityingly and half disdainfully at the modestly dressed Pareto – although he was known to be well off – and to have answered that were plenty of cheap restaurants, but one had to pay something everywhere. At which Pareto said: ‘So there are natural laws of political economy” (Mises as quoted in: Rothbard, 2006, p. 459).

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

Now, let us recapitulate the main findings of our study into the very meaning of ceteris paribus clause. It seems that the only uncontroversial way to successfully defend such laws is just to claim that a given cp-statement is only true in a model used for its “production” or, in a second case, if one has a cp-law, no matter of its origin, then one can always construct a model in which such a statement is to be true. But problems arise once we try to use cp-laws in order to describe some empirical facts. We have just shown above that various ways of understanding such claims have various problems. Thus, in what follows we are to continue our search for proper understanding of cp-laws used to describe economic real worlds.

3. CETERIS NORMALIBUS LAWS IN ECONOMICS Having in mind what has been said above, one should agree that economic laws (usually stated with ceteris paribus clauses) can be understood as the laws always true in economic models, and hence ceteris normalibus is just to be conceptualized as being synonymous to “in a model” phrase. So, for instance, saying that ceteris paribus lower interest rates are to stimulate investments can be rephrased that lower interest rates always stimulate investments only in theoretical models where such a relation holds or, in other words, ceteris normalibus, lower interest rates stimulates investments. However, we have a problem once we try to describe situation, say, in a Polish economy. If our economy is to be the same as our theoretical model is, then for sure lowering interest rates is to stimulate investment. Conversely, we do not have such a perfect isomorphism between the economy and its model. So, how we should understand such claims when they are directed towards our empirical domains? We are not to repeat our insights from the preceding section but rather we are to try offering you a different and, in a sense, more metaphysically rich way of interpreting ceteris normalibus clause. However, such a reading is not to falsify the above presented simple claim that cn clauses can be understood as referring particular statements to models where they are true. If we would like to describe economic realm, then we should ask how its constituting parts exist and more fundamentally a study into the nature of specific existents is needed. It is claimed here that “all features of reality can be viewed under the aspect of their being” (Lawson, 2014, p. 19). And scientific ontology can be understood as primarily interested in investigating the natures of particular existents. It differs from philosophical ontology which deals with general aspects of being.9 In what follows we are particularly interested in the nature of 9 M. Bunge, for instance, describes the difference between scientific and philosophical ontology as follows: “Ontology can be classed into general and special (or regional). General ontology studies all existents, whereas each special ontology studies one genus of thing or process physical, chemical, biological, social, etc. Thus, whereas general ontology studies the concepts of space, time, and event, the ontology of the social investigates such general sociological concepts as those of social system, social structure, and social change” (Bunge, 1999, p. 200). Our scientific ontology is an example of Bunge’s special ontology.

Łukasz Hardt

economic being and more generally in the very nature of economic world. Therefore, such a perspective is relatively skeptical towards deductivist mode of explanation in economics where to explain is to provide an account of how explanandum must be deduced from explanans and where initial conditions are supplemented by invocations to universal laws such as whenever A then B. As it is shown in Hardt (2017), economic world is not governed by such laws and hence we should not be surprised that our attempts at understanding cp-laws in economics in terms of some regularities usually fail. What we should focus on is the real domain of economic reality where such entities as the following ones are present, namely powers, mechanisms, tendencies, and structures.10 Let us refer here again to T. Lawson, one of the leading proponents of rediscovering economic ontology: “[…] science aims at uncovering causal factors, that is, it is concerned with identifying structures, mechanisms and the tendencies they ground, which produce, govern or facilitate phenomena at a different level. And if the aim of science is to illuminate structures that govern surface phenomena then laws or law-statements are neither empirical statements (statements about experiences) nor statements about events or their regularities (whether unqualified or subject to ceteris paribus restrictions), but precisely statements elucidating structures and their characteristic modes of activity” (Lawson, 1997, p. 24; italics in original; emphasis added). So, according to Lawson, every law-like statement in economics, including the ones with cp-clauses, do not describe regularities but rather “modes of activity” of particular economic entities. But how such “modes” should be understood? Definitely we do not have here any references to probabilities but rather to prototypical characteristics of broadly understood economic entities. Let us give the floor again to Lawson: “[…] these or related notions [law-like statements in economics] must be conceived in terms of potentials; as potentials that may or may not be expressed, and if expressed that may or may not be actualized because of countervailing tendencies […]. The fundamental error of orthodox theory here […] is not its focus upon such conceptions as rationality or profit seeking per se, the problem is its presumption that such matters, at some level at least, are always expressed in terms of actualities rather than capacities” (ibid., p. 106). Thus, Lawson goes further than Mill in stressing that economic laws are just statements about tendencies. His claim is that there is something real beyond appearances of economic phenomena and thus one may have a given economic entity containing a capacity to act but at the same time this very power may be dormant. In such a framework, economics can be treated as a science despite the 10 The

real layer of economic reality is accompanied by the empirical one (experience) and the actual one (actual events and states of affaires) (Lawson, 1997, p. 21).

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

fact that economic events could always have been different. We agree with N. Cartwright saying that “Our most wide-ranging scientific knowledge is not knowledge of laws but knowledge of the natures of things” (Cartwright, 1999, p. 4).11 Therefore, to know is to refer to powers, capacities, natures, mechanisms, and structures. So, a given economic entity, say, an economic agent, has, for instance, a capacity to act rationally. She can still have such a capacity even without any instances of its actualization. Thus, one may say the following: it is in nature of an economic agent to act rationally. It means something different than ceteris normalibus statement making reference to a given model in which a particular statement is true. Here, by employing in nature of something it is to produce something else condition one just refers to inner composition of economic entities.12 So, there is deep in the world and not everything is on the surface. By referring to capacities we definitely move towards Aristotelian approach to explanation since he sees “natures as primary and behaviors, even very regular behavior, as derivative” (ibid., p. 149). In Physics (II, 1, 192b22) Aristotle writes that the goal of investigating the way the world works is a search for “[…] the factor which initiates movement and rest within that thing in which it is itself immediately, not incidentally, present” (Aristotle, 1961, p. 23). So, we talk here about internal forces , or inner causes, of changes. Similarly, Aristotle offers us a parallel idea to the one of nature, namely capacities or dynameis that are “powers to do” (Crespo, 2009, p. 124). They are defined in Metaphysics (V, 12, 11 This

part of paper is heavily based on Lawson’s and Cartwright’s insights. Although they are very similar, we are conscious that one should notice also some differences, e.g., Cartwright is more sympathetic to neoclassical economics than Lawson is. She claims that laws in economics hold only in highly organized environments, or, in her own words, in nomological machines. Lawson however has some doubts whether even in model conditions a given thing’s capacity is to be always activated since “a statement of a tendency […] is an unconditional statement about something non-actual and non-empirical” (Lawson, 1997, p. 23) and hence his transcendental realism (interesting comparison of Cartwright and Lawson is offered in Hoover 2002). For the sake of this paper, it is not necessary however to offer more comments on the differences between these two authors. However, if we would be pressed to choose whose ideas are closer to our way of thinking about capacities and powers, we would choose Cartwright, since our claim is that ceteris normalibus laws are true in economic models, or, to use Cartwright’s terms, in blueprints of nomological machines. 12 Such laws are called normic laws, since a given A normally produces B, and to put it more formally Ax Bx (for “As are normally Bs”, ‘ ’ is a variable-binding conditional). We do not want to offer here more formalized treatment of such laws (it is not necessary for this analysis), however, the following words by Schurz (2004, 186) are worth referring to: “The ‘normality’ of a normic law Ax Bx is relative to both the antecedent predicate Ax and the consequent predicate Bx. For example, ‘birds normally can fly’ speaks about what is normal for birds and not about what is normal for arbitrary animals, e.g., it is normal for fishes to be able to swim but not to fly. Moreover, ‘birds normally can fly’ tells us what it means for a bird to be normal with respect to its way of locomotion, but not necessarily with respect to other property families, for example, a bird which can fly but is infertile is normal with respect to its way of locomotion but abnormal with respect its reproduction ability. This demonstrates that ‘ ’ is a genuine conditional operator which cannot be adequately understood either as a special ‘predicate’ “Ax /\ Norm(x) Bx”, or as an unary ‘normality-operator’ attached to an ordinary material implication “Norm(Ax Bx)” (Schurz, 2004, p. 186; cf. Boruszewski in this issue of “Studia Ekonomiczne” for a more formalized treatment of interplays between models and theories; see also Boruszewski, 2014).

" "

Łukasz Hardt

1019a, pp. 14–16) as follows: “Something is said to be a capacity [potentiality, power] when it is a starting-point of movement or change either in another thing or in a thing insofar as it other” (Aristotle, 2016, p. 83). Why it matters for economics? Simply because in economic realm we have capacities. In his discussion of Cartwright’s insights D. Hands claims the following: “[…] she argues repeatedly that real practicing scientists actually do presuppose that capacities and causal powers exist in systems they study” (2001, p. 313). In her 1989 book she is straightforward in claiming that capacities are also present in social sciences, including economics, however, the ones in economics are less powerful (cf. Crespo, 2009, pp. 127–128). In her own words: “Social science is hard, but not impossible. Nor should that be surprising; natural science is exceedingly hard and it does not confront so many problems as social science – problems of complexity, of reflexivity, of lack of control. Moreover the natural sciences more or less choose the problems they will solve but the social sciences are asked to solve the problems that policy throws up” (Cartwright, 2007b, p. 42). And referring to economics she adds: “The natural thought about the difference between the most fundamental capacities studied in physics and the capacities studied in economics is that the economic capacities are derived whereas those of fundamental physics are basic. Economic features have the capacities they do because of some underlying social, institutional, legal and psychological arrangements that give rise to them. So the strengths of economic capacities can be changed, unlike many in physics, because the underlying structures from which they derive can be altered” (Cartwright, 2007a, p. 54). Nevertheless, capacities in economics are real and should play an important role in explaining economic facts. If now causation can be conceptualized in terms of manifestation of power, then we should be able to explain singular events without any need for some general laws. Thus singular causation. She explains it as follows: “The generic causal claims of science are not reports of regularities but rather ascriptions of capacities, capacities to make things happen, case by case. ‘Aspirins relieve headaches.’ This does not say that aspirins always relieve headaches, or always do so if the rest of the world is arranged in a particularly felicitous way, or that they relieve headaches most of the time, or more often than not. Rather it says that aspirins have the capacity to relieve headaches, a relatively enduring and stable capacity that may if circumstances are right reveal itself by producing a regularity, but which is just as surely seen in one good single case. The best sign that aspirins can relieve headaches is that on occasion some of them do” (Cartwright, 1989, pp. 2–3).

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

Now, it is obvious that while referring to inner composition of things, or capacities, then ceteris normalibus gains a metaphysically rich imprint. For instance, one can say that ceteris normalibus people self-socialize, namely that it is in the very nature of men to be with others. As Aristotle (1984, p. 37) puts it in Politics (I, 2, 1253a, pp. 29−30): “[…] there is in everyone by nature an impulse towards this sort of partnership”. And in the same vein A. Smith famously proclaims that humans are characterized by “the propensity to truck, barter, and exchange one thing for another” (Smith, 1994, p. 14). So, such a capacity is not derived but it is situated deeply in human nature.13 But, on the other hand, Cartwright is right that majority of capacities in economics are derived. It simply means that they are based on a very small number of basic ones, e.g., as the one mentioned by Smith or Millian claim that the most important principle of human nature is to acquire wealth.14 However, as Lawson claims, we may have also powers embodied in social or economic systems. He gives the following example: “Community life, then, is organised; it is so by way of emergent collective practices and their inherent rights and obligations that structure human interaction. The result is a social totality or set of totalities. And the latter have causal powers. A motorway system for example, structured by various inter-connecting collective practices, has powers of co-ordinating that are irreducible to any of its various motoring components; and a language system has powers to facilitate communication that are irreducible to those of any individual communicator” (Lawson, 2014, p. 36; emphasis added). Therefore, although this issue is very complicated and for sure beyond the scope of this short paper, we may albeit with some reservations refer to natures in economic domain of social reality. After the above insights on what capacities are (ontology), it is now time to offer some comments dealing with how they are known (epistemology). And here one can rightly ask how it is possible to identify capacities if they can be present, but at the same time do not manifest themselves. So, three points are worth making here. First, we need special arrangements where capacities can show up. Second, measurement of their effects is necessary. Third, capacities can be deduced from probabilities, or, to say more precisely, probabilities can offer us hypothesis concerning capacities’ existence. All three issues are nicely analyzed 13 Here Smith follows clearly D. Hume’s insights concerning human natures, for instance, “It is universally acknowledged, that there is a great uniformity among the actions of men, in all nations and ages, and that human nature remains still the same, in its principles and operations” (Hume, 1748/1993, p. 55). 14 In Utilitarianism he writes: „If human nature is so constituted as […] happiness is the sole end of human action […]. It necessarily follows that is must be the criterion of morality” (Mill, 1863, p. 57).

Łukasz Hardt

in Cartwright’s writings. So, what she proposes is to build nomological machines where we can observe capacities in action. As she puts it: “[nomological machine] is a fixed (enough) arrangement of components, or factors, with stable (enough) capacities that in the right sort of stable (enough) environment will, with repeated operation, give rise to the kind of regular behaviour that we represent in our scientific laws […]. Laws of nature (in this necessary regular association sense of “law”) hold only ceteris paribus – they hold only relative to the successful repeated operation of a nomological machine” (Cartwright, 1999, p. 50). And here she is straightforward in treating cp-clauses as the ones refereeing to a given nomological machine, or, to normal conditions. Also, in such perfect arrangements capacities are to be active, however, once we move from ideal conditions to some real world settings these capacities are still to be present but they may not manifest themselves. Therefore, ceteris normalibus clause understood as the one refereeing to entities’ capacities or powers. In The Dappled World (1999) she gives a concrete example of a theoretical model, a kind of a blueprint of nomological machine, namely the one by Hart and Moore (1991) where they try to analyze how optimal contracts between banks and entrepreneurs should look like. We are not to present this case here, but we would rather give the floor again to Cartwright studying the issue of how this model’s insights should be understood.15 She says the following: “There must be a machine like the one modelled by Hart and Moore […] to give rise to it. There are no law-like regularities without a machine to generate them. Thus, ceteris paribus conditions have a very special role to play in economic laws like [Corollary 1 in Hart’s and Moore’s model]. They describe the structure of the machine that makes the laws true” (Cartwright, 1999, p. 148). So, again she supports the view treating cp-clauses as ceteris normalibus ones. Before moving further one reservation is in order here. Understanding ceteris normalibus clauses in a metaphysically rich manner does not mean that they cannot be just treated as somehow synonymous to claiming that our laws hold only in model conditions. Such an understanding of cn clauses does not require us to refer to capacities and powers. Or, to put it as simply as possible, cn restriction may be synonymous to ‘in a model’ condition. Therefore, we have two possible treatments of cn conditions with the one quite Aristotelian in nature.16 Now, after the above philosophically rich analyses of ceteris normalibus laws we would like to offer our readers a kind of a more popular treatment of such 15 This

Cartwright’s case-study is put under scrutiny in Hardt (2017, p. 117). is also possible to use ceteris normalibus in a quite informal way as a clause simply meaning that our generalizations are to hold in non-exceptional times. However, assessing whether times are non-exceptional is made by checking how far from normal conditions we are and thus again even here we come back to normality assumptions. 16 It

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

statements. Therefore, in the following brief section we are still to focus on ceteris normalibus laws but we will use insight’s from D. Rodrik’s book Economics Rules: The Rights and Wrongs of the Dismal Science (2015) and an interesting debate on his book among philosophers of economics. Such insights should nicely support our treatment of economic laws.

4. THE ROLE OF MODELS IN PRODUCING (CETERIS NORMALIBUS) ECONOMIC LAWS In discussing the role models play in producing our knowledge about the economic world it is useful not only to use some philosophical insights but also to look into the way practicing economists treat models. Here, as A. Rubinstein claims, D. Rodrik’s 2015 book “can serve as an ideal platform for discussing what economics can and should accomplish” (Rodrik, 2015, p. 162). Let us start by referring to three points from Rodrik’s work, i.e.: “[…] models enable the accumulation of knowledge, by expanding the set of plausible explanations for, and our understanding of, a variety of social phenomena. In this way, economic science advances as a library would expand: by adding to its collection” (ibid., p. 46). “Models are never true: but there is truth in models” (ibid., p. 44). “In economics, context is all. What is true of one setting need not be true of another” (ibid., p. 164). So, according to Rodrik, once you have a given context, then you need an appropriate model. And since we have an unlimited number of different contexts, then the more models you have, the higher probability that you are to find a right model is. So, there is no such thing as the model, but always it is a model (ibid., p. 43). Therefore, we do not have unconditionally true models, but models can be only relatively true, namely they can be true in virtue of a context one is to use them in. Such a vision of economic models is perfectly in line with the following one: “The fact that a model turns out not to work under certain circumstances does not count as a refutation of the model but only as a failed test of its applicability in a given domain” (Guala, 2005, p. 220). Thus, “[…] the closer a given empirical domain to the model’s structure is, the higher probability that the model’s insights are to correctly explain the workings of such a domain” (Hardt, 2017, p. 152). Nevertheless, another way of understanding Rodrik’s arguments is possible in which “[…] explanation requires finding and utilizing the right set of models for the explanatory task at hand” (Aydinonat, 2018, p. 237) and not having the right model. In such a reading models offer us a set of explanations and only later we empirically verify their plausibility. Although the above two

Łukasz Hardt

ways of treating Rodrik’s insights are different, they merge in one important aspect, namely in both models produce claims about empirical world that are always true in models generating them but only partially true once referred to domains beyond the models.17 Having said the above, we should ask a very simple and fundamental question: what are ingredients of models? As models can be viewed as pragmatically and ontologically constrained representations, we can state the following: “Agent A uses object M as a representative of some target system R for purpose P, addressing audience E, prompting genuine issues of resemblance to arise; and applies commentary C to identify and align these components” (Mäki, 2009, p. 32). Economists usually put emphasis on M rather than on conditions for using a particular model in a given context. Therefore, economists disregard models’ commentaries. They are good in producing models but they are often unable to rightly select an appropriate one. As Rodrik comments “Freshly minted PhDs come out of graduate school with a large inventory of models but virtually no formal training – no course work, no assignments, no problem sets – in how one chooses among them” (Rodrik, 2015, pp. 83−84). It reminds us strong words from The American Economic Association’s 1991 Commission on the state of graduate education in economics in the USA, namely that ”[…] graduate programs may be turning out a generation with too many idiot savants skilled in technique but innocent of real economic issues” (Krueger, 1991, pp. 1044–1045; italics in original). In other words, they even do not know that models once referred to empirical domains need commentaries. Also, it is misleading to treat models’ descriptions as commentaries. The absence of commentaries in economic models leads economists to overconfidence in the statements they offer in public debates and theories they produce. Therefore, they often seem to claim that their models can be used regardless of the context. Or, in other words, they do not inform the public that a given claim is only true in a particular model but once referred to the outside model world it is usually not perfectly correct. Believing in universality of economic laws (or theories) is just a form of scientific fundamentalism. Here it is again worth to give a floor to N. Cartwright stating the following: “Return to my rough division of law-like items of knowledge into two categories: (1) those that are legitimately regimented into theoretical 17 In his comments on Aydinonat’s paper Rodrik (2018) writes in this context: “I would also put great weight on Occam’s razor: use the least number of models as possible” (p. 278).

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

schemes, these generally, though not always, being facts about behaviour in highly structured, manufactured environments like a spark chamber; (2) those that are not. There is a tendency to think that all facts must belong to one grand scheme, and, moreover, that this is a scheme in which the facts in the first category have a special and privileged status. They are exemplary of the way nature is supposed to work. The others must be made to conform to them. This is the kind of fundamentalist doctrine that I think we must resist” (Cartwright, 1994, p. 316; emphasis added). Fundamentalism, by its very nature, is an unscientific doctrine. And it also did and it still is doing a lot of damage to economics and economists. As D. Colander claims in his well-known paper on the state of economics “Neoclassical economists made a fatal mistake that classical economists had avoided and had strongly warned against: They drew policy conclusions directly from their models and theory” (Colander, 2011, p. 8) and next he adds “Professional economists have been unwilling to admit that the economy is far too complex to be captured by any unified model. In private discussions among ourselves we recognize this complexity, but we don’t add the appropriate warning labels to our models when they are discussed in public. There, we pretend we understand more than we do” (ibid., p. 20). In his comments regarding Rodrik’s book U. Mäki writes the following: “This may result in difficulties in developing adequate model commentaries that would incorporate appropriate degrees of humility reflecting the uncertainties that are involved” (Mäki, 2018, p. 225). Coming back now to Colander’s 2009 paper, published during the Great Recession, we find even a stronger statement: “Defining away the most prevalent economic problems of modern economies and failing to communicate the limitations and assumptions of its popular models, the economics profession bears some responsibility for the financial and economic crisis” (Colander, 2009, p. 264). So, here Rodrik and Colander agree: we have more problems with the ways economic insights are used and treated than with theoretical economics as such (cf. Hardt, 2016).18 Let us now come back to our discussion on ceteris normalibus laws and here one may ask how does it fit with Rodrik’s insights on economics. To answer it very quickly we would just say that they correspond only partially. Or, in more 18 Such a claim is strongly supported, for instance, by B. Bernanke stressing the following: “[…] the recent financial crisis was more a failure of economic engineering and economic management than of what I have called economic science” (Bernanke, 2010, p. 3). To put it in more Keynesian terms, we do not have problems with science of economics but definitely we have problems with art of economics. To quote Keynes, “Good economists are scarce because the gift for using ‘vigilant observation’ to choose good models, although it does not require a highly specialized intellectual technique, appears to be a very rare one” (Keynes, 1938/1978, pp. 296–297). In his 2018 comments on U. Mäki’s discussion of his work he said the following referring to the above quote from Keynes: “Had I been familiar with this quote from Keynes before I wrote the book, I might have chosen not to spend the effort!” (Rodrik, 2018, p. 277). Luckily, he was not familiar.

Łukasz Hardt

precise terms, Rodrik’s vision agrees with the one seeing economists’ claims about economic reality as only ceteris normalibus in a sense of being true only in particular models. So, his statement that “Models are never true: but there is truth in models” (Rodrik, 2015, p. 44) can be rephrased in the following way: Economic truths (i.e., statements about real economic systems) are only cateris normalibus truths and such truths are true in particular models. However, Rodrik’s treatment of economics does not support ceteris normalibus clauses as the ones referring to economic entities’ capacities and natures. He does not refer to such metaphysically rich categories. One reason for this may be the fact that his “goal was to sketch a middle line between hardcore falsiﬁcationism (which gets us nowhere) and empirical nihilism (which presumes there is no there there)” (Rodrik, 2018, p. 278). Nevertheless, in his book we can find a claim that “At best, we can talk in terms of tendencies, context-specific regularities, and likely consequences” (Rodrik, 2015, p. 45). However, on the other hand, he somehow cannot give up his dream of finding such a model that being strongly isomorphic to its empirical target would give us specific-truths about this very target. So, he writes the following: “They [theories] are specific rather than universal theories. They aim to shed some light on particular historical episodes and do not describe general laws and tendencies” (ibid., 2015, p. 115). Here we disagree since even in a very specific empirical context our theories are to be at best statements about capacities or tendencies rather than strict (although only context specific) laws. This is so because economic world is not governed by laws but by natures and powers. Or, in Lawson’s words, economic world is a world of potentialities rather than actualities. However, and here we should be grateful to Rodrik, he takes us not far away from such more metaphysically rich treatment of economic realm. Summing up what have been just said about Rodrik, it is definitely more appropriate to treat his insights as the ones supporting interpretation of economic laws in ceteris normalibus terms rather than in ceteris paribus manner. He even explicitly uses in his books sentences treated by us earlier in second section as being nearly synonymous to ceteris normalibus clause, for instance, “Normally, broad technological progress that increases labor productivity is expected to improve everyone’s living standards” (Rodrik, 2015, p. 141), or “[…] a huge injection of money by central bank will produce inflation in normal times (ibid., p. 185). Normally, or in normal times, mean here just in a particular model. However, as Cartwright observes, “[…] the literal translation [of ceteris paribus] is ‘other things being equal’; but it would be more apt to read ‘ceteris paribus’ as ‘other things being right’” (Cartwright, 1983, p. 45). And for Rodrik being right means being in an appropriate model.19 So, again, we find more arguments for treating ceteris paribus in ceteris normalibus terms. 19 Cartwright’s ‘other things being right’ statement can be called a ceteris rectis clause. Here we understand being right as being normal (cf. Schurz, 2014; Hardt, 2018).

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

5. CONCLUSIONS The issue of what kind of statements economics laws are is of profound importance both for economics as such and economic policies based on various theoretical assumptions. Our claim in this paper is that the most appropriate way of treating economic laws is to conceptualize them in ceteris normalibus terms, namely as statements only true in normal circumstances. However, perfectly normal circumstances can be found only in theoretical models. Once we move beyond such models our statements degenerate into the ones about tendencies that are due to capacities present in economic entities. Therefore, for instance, a popular claim that a growth in money supply is to rise inflation should not be treated in terms of being true under assumption of all else being equal (ceteris paribus) but better as the one true in normal conditions (ceteris normalibus). And, simplifying a bit, the closer a given empirical domain to such normal conditions is, the higher probability that relation described by a particular law is to hold. Since such closeness or isomorphism is never perfect thus our claim that economic laws describe tendencies. Moreover, such tendencies are due to the fact that economic processes are not governed by universal laws but rather by powers, capacities, and natures. Or, in other words, economic world is the one of potentialities rather that actualities. So, refereeing to the above example, and again simplifying a bit, one can say that in the nature of growing money supply is to make inflation higher. However, since capacities may be dormant one may have growing money supply without expected effect in higher inflation. Yet, in such a case, it is still true that ceteris normalibus rising money supply makes inflation higher since “scientific knowledge is not knowledge of laws but knowledge of the natures of things” (Cartwright, 1999, p. 4). However, those denying the existence of capacities and natures can still use cn clauses as simply the ones restricting the validity of given statements to theoretical models. So, in a sense, we have two understandings of such a restriction: the one which is metaphysically rich, and the one denying capacities or being agnostic about their existence. Last but not least, my reading of economic laws does not make traditional understanding of ceteris paribus clause obsolete in economics. Still, one may say that his generalization is only true once other factors are constant. Nevertheless, it is usually not sufficient since such constancy must be supplemented with the conditions under which a given regularity holds and thus ceteris normalibus clause is needed. So, to be precise, we may have mixed laws containing both ceteris paribus clauses and ceteris normalibus ones. This is precisely what Marshall proposed in his Principle of Economics were he offers both “the condition that other things are equal”, however, he did it only after writing that economic laws describe “the course of action which may be expected under certain conditions from the members of an industrial group is the normal action of the members of that group relatively to those conditions” (see, section 2). Thus, as Schurz (2014) conceptualizes it, we may have so-called mixed cn-cp laws, for instance, the fol-

Łukasz Hardt

lowing one: ceteris normalibus (for ‘‘sufficiently ideal’’ markets) and ceteris paribus (provided the other variables remain unchanged), an increase in demand leads to an increase in price.20 Consequently, interpreting economic laws as only ceteris paribus ones is hardly correct. And now the final point. Viewing economic laws as ceteris normalibus ones clearly makes us conscious that in economics we do not have universal laws and answers. What we only have are laws that are context-specific, and they refer to tendencies and capacities. They are only true without any exceptions in theoretical models. We can do nothing more here than again agree with the following Rodrik’s opinion: “Economists who remain true to their discipline, like Tirole, are necessarily humble […]. Their responses to most questions necessarily take the form of ‘It depends’, ‘I don’t know’” (Rodrik, 2015, p. 209), and we would add that their responses are at best stated in ceteris normalibus terms.

REFERENCES Alchian A. (1955), The Rate of Interest, Fisher’s Rate of Return Over Cost, and Keynes’ Internal Rate of Return, “The American Economic Review”, 45(5), pp. 938–943. Anderson G.M., Goff B.L., Tollison R.D. (1986), The Rise and (Recent) Decline of Mathematical Economics, “Bulletin of the History of Economics Society”, 8(1), pp. 44–48. Aristotle (1961), Physics, University of Nebraska Press, Lincoln. Aristotle (1984), Politics, University of Chicago Press, Chicago. Aristotle (2016), Metaphysics, Hackett Publishing Company, Indianapolis. Aydinonat N.E. (2018), The Diversity of Models as a Means to Better Explanations in Economics, “Journal of Economic Methodology”, 25(3), pp. 237–251. Backhouse R. (1998), The Transformation of US Economics 1920–60, Viewed through a Survey of Journal Articles, in: M. Morgan, M. Rutherford (eds.), From Interwar Pluralism to Postwar Neoclassicism, Duke University Press, Durham, pp. 85–107. Bernanke B. (2010), On the Implications of the Financial Crisis for Economics, Lecture at the Bendheim Center for Finance and the Center for Economic Policy Studies, https://fraser.stlouisfed.org/files/docs/historical/bernanke/bernanke_20100924.pdf (accessed: 1/3/2017). Blaug M. (2003), The Formalist Revolution of the 30’s, “Journal of the History of Economic Thought”, 25(2), pp. 147–156. Boruszewski J. (2014), Problem pomiaru w semantyce neooperacjonalistycznej, “Filozofia Nauki”, 22(3), pp. 67–80. Brunner O. (1939), Land und Herrschaft, Wissenschaftliche Buchgesellschaft, Darmstadt. 20 To

be precise, Schurz writes about ceteris rectis laws (see, footnote no 20).

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

Bunge M. (1999), Dictionary of Philosophy, Prometheus Books, Amherst. Carnap R. (1938), Foundations of Logic and Mathematics, in: O. Neurath, Ch. Morris, R. Carnap (eds.), International Encyclopedia of Unified Science, vol. 1, University of Chicago Press, Chicago, pp. 139–213. Cartwright N. (1983), How the Laws of Physics Lie, Oxford University Press, New York/Oxford. Cartwright N. (1989), Nature’s Capacities and their Measurement, Clarendon Press, Oxford. Cartwright N. (1994), Fundamentalism vs. the Patchwork of Laws, “Proceedings of the Aristotelean Society”, 94, pp. 279–292. Cartwright N. (1999), The Dappled World: A Study of the Boundaries of Science, Cambridge University Press. Cartwright N. (2007a), Causal Powers: What are They? Why do we need Them? What can be done with Them and What Cannot? Contingency and Dissent in Science Project, Technical Report 04/07, LSE, London. Cartwright N. (2007b), Hunting Causes and Using Them: Approaches in Philosophy and Economics, Cambridge University Press, Cambridge. Clower R.W. (1954), An Investigation into the Dynamics of Investment, “The American Economic Review”, 44(1), pp. 64–81. Colander D. (2011), How Economists got it Wrong: A Nuanced Account, “Critical Review: A Journal of Politics and Society”, 23(1-2), pp. 1–27. Colander D., Goldberg M., Haas A., Juselius K., Kirman A., Lux T., Brigitte S. (2009), The Financial Crisis and the Systematic Failure of the Economics Profession, “Critical Review: A Journal of Politics and Society”, 21(2), pp. 249–267. Crespo R. (2009), Nancy Cartwright, Millian and/or Aristotelian, “Sapientia”, 65, pp. 113–139. Debreu G. (1959), Theory of Value: An Axiomatic Analysis of Economic Equilibrium, Yale University Press, New Heaven. FED (2018), How does Monetary Policy Influence Inflation and Employment?, https:// www.federalreserve.gov/faqs/money_12856.htm (accessed 29/10/2018). Frigg R., Hartmann S. (2018), Models in Science, in: E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2018 Edition), https://plato.stanford.edu/ archives/sum2018/entries/models-science/ (accessed 1/9/2018). Guala F. (2005), The Methodology of Experimental Economics, Cambridge University Press, Cambridge. Haavelmo T. (1944), The Probability Approach in Econometrics, “Econometrica”, 12, pp. iii–vi+1–115. Hands D. (2001), Reflection Without Rules: Economic Methodology and Contemporary Science Theory, Cambridge University Press, Cambridge. Hardt Ł. (2016), The Recent Critique of Theoretical Economics: a Methodologically Informed Investigation, “Journal of Economic Issues”, 50(1), pp. 269–287. Hardt Ł. (2017), Economics without Laws. Towards a New Philosophy of Economics, Palgrave Macmillan, Cham. Hardt Ł. (2018), Prawa ceteris rectis w ekonomii, “Gospodarka Narodowa”, No. 1, pp. 9–31.

Łukasz Hardt

Hart O., Moore J. (1991), A Theory of Debt Based on the Inalienability of Human Capital, “Quarterly Journal of Economics”, 109(4), pp. 841–879. Hausman D. (1998), Causal Asymmetries, Cambridge University Press, Cambridge. Hoover K. (2001), Causality in Macroeconomics, Cambridge University Press, New York. Hoover K. (2002), Econometrics and Reality, in: U. Mäki (ed.), Fact and Fiction in Economics, Cambridge University Press, Cambridge, pp. 152–177. Hume D. (1748/1993), An Enquiry Concerning Human Understanding, Hackett Publishing Company, New York. Kesting P., Vilks A. (2004), Formalism, in: J.B. Davis, A. Marciano, J. Runde (eds.), The Elgar Companion to Economics and Philosophy, Edward Elgar, Cheltenham, pp. 283–298. Keynes J.M. (1938/1978), Letter to Roy Harrod, in: E. Johnson, D. Moggridge (eds.), The Collected Writings of John Maynard Keynes, 14, pp. 296–297. Klaes M. (2001), Begriffsgeschichte: between the Scylla of Conceptual and the Charybdis of Institutional History of Economics, “Journal of the History of Economic Thought”, 23(2), pp. 153–179. Klaes M., Sent E.M. (2005), A Conceptual History of the Emergence of Bounded Rationality, “History of Political Economy”, 37(1), pp. 27–59. Koselleck R. (1972a/1979), Begriffsgeschichte und Sozialgeschichte, in: R. Koselleck (ed.), Vergangene Zukunft, Suhrkamp, Frankfurt a. M., pp. 107–129. Koselleck R. (1972b/2004), Begriffsgeschichte and Social History, in: R. Koselleck (ed.), Futures Past (translated by Keith Tribe), Columbia University Press, New York, pp. 75–92. Krueger A.O. (1991), Report on the Commission on Graduate Education in Economics, “Journal of Economic Literature”, 29(3), pp. 1035–1053. Lange M. (2002), Who’s Afraid of Ceteris Paribus Laws? Or: How I Learned to Stop Worrying and Love Them, “Erkenntnis”, 57(3), pp. 407–423. Lange M. (2009), Laws and Lawmakers. Science, Metaphysics and the Laws of Nature, Oxford University Press, Oxford. Lawson T. (1997), Economics and Reality, Routledge, London. Lawson T. (2014), A Conception of Social Ontology, in: S. Pratten (ed.), Social Ontology and Modern Economics, Routledge, Oxon, pp. 19–52. Mäki U. (2009), MISSing the World. Models as Isolations and Credible Surrogate Systems, “Erkenntnis”, 70(1), pp. 29–43. Mäki U. (2018), Rights and Wrongs of Economic Modelling: Reﬁning Rodrik, “Journal of Economic Methodology”, 25(3), pp. 218–236. Mäki U., Piimies J.P. (1998), Ceteris Paribus, in: J. Davis, W. Hands, U. Mäki (eds.), The Handbook of Economic Methodology, Edward Elgar, Cheltenham, pp. 55–59. Marshall A. (1920/2013), Principles of Economics, Palgrave Macmillan, Basingstoke. Mill J.S. (1836/2008), On the Definition and Method of Political Economy, in: D. Hausman (ed.), The Philosophy of Economics, Cambridge University Press, Cambridge, pp. 41–58. Mill J.S. (1843), A System of Logic. Ratiocinative and Inductive, John W. Parker, London. Mill J.S. (1863), Utilitarianism, Parker, Son, and Bourn, London.

ECONOMIC MODELS AND CETERIS NORMALIBUS LAWS

Nietzsche F. (1981), Zur Genealogie der Moral, in: K. Schlechta (ed.), Friedrich Nietzsche: Werke in Drei Bänden, Vol. 2, Hanser, München. Pareto V. (1906/2014), Manual of Political Economy, Oxford University Press, Oxford. Persky J. (1990), Retrospectives: Ceteris Paribus, “Journal of Economic Perspectives”, 4(2), pp. 187–193. Reiss J. (2013), Philosophy of Economics, Routledge, London. Reutlinger A., Schurz G., Hüttemann A. (2017), Ceteris Paribus Laws, in: E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2017 Edition), https:// plato.stanford.edu/archives/spr2017/entries/ceteris-paribus/ (accessed: 1/4/2018). Rodrik D. (2015), Economics Rules: The Rights and Wrongs of the Dismal Science, Oxford University Press, Oxford. Rodrik D. (2018), Second Thoughts on Economics Rules, “Journal of Economic Methodology”, 25(3), pp. 276–281. Rothbard M. (2006), Austrian Perspective on the History of Economic Thought, vol. 2, Ludwig von Mises Institute, Auburn. Ruggles R. (1954), The Value of Value Theory, “The American Economic Review”, 44(2), pp. 140–151. Samuelson P. (1952), Economic Theory and Mathematics – An Appraisal, “The American Economic Review”, 42(2), pp. 56–66. Schurz G. (2004), Normic Laws, Nonmonotonic Reasoning, and the Unity of Science, in: S. Rahman S. (ed.), Logic, Epistemology, and the Unity of Science, Kluwer, Dordrecht, pp. 181–211. Schurz G. (2014), Ceteris Paribus and Ceteris Rectis Laws: Content and Causal Role, “Erkenntnis”, 79(S10), pp. 1801–1817. Smith A. (1994), The Wealth of Nations, Random House, New York. Suppes P. (1970), A Probabilistic Theory of Causality, North-Holland, Amsterdam. Woodward J. (2000), Explanation and Invariance in the Special Sciences, “The British Journal for the Philosophy of Science”, 51(2), pp. 197–254. Woodward J. (2003), Making Things Happen, Oxford University Press, Oxford.

MODELE EKONOMICZNE I PRAWA CETERIS NORMALIBUS STRESZCZENIE W artykule omówiono charakter praw ekonomicznych. Zamiast rozumieć je w kategoriach ceteris paribus, twierdzi się, że prawa ekonomiczne powinny być opatrzone klauzulą ceteris normalibus, która może być rozumiana w dwojaki sposób. Po pierwsze, jako stwierdzenie, że dane prawo jest prawdziwe tylko w warunkach określonego modelu. W takiej sytuacji, im dana domena empiryczna jest bliższa strukturze modelu, tym wyższe prawdopodobieństwo, że konkluzje modelu (tj. prawa ekonomiczne) poprawnie opisują tę domenę. Jednak nigdy nie mamy do czynienia z pełnym izomorfizmem między modelami a ich domenami empirycznymi, a więc prawa ekonomiczne opisują jedynie tendencje w rzeczywistości empirycznej. Dochodzimy więc do innego rozumienia klauzuli ceteris normalibus: prawa nią opatrzone nie opisują regularności, ale odnoszą się do poten-

Łukasz Hardt

cjalności i możności. Określają to, czego powodowanie leży w naturze danego czynnika sprawczego. Takie prawa nazywamy prawami normalnościowymi (normic laws). W artykule, badając naturę praw ekonomicznych, przedstawiono również historię zastosowania klauzuli ceteris paribus w ekonomii. Ponadto niniejszy artykuł nawiązuje do interesującej debaty dotyczącej modeli i praw ekonomicznych, zawartej w książce D. Rodrika Economics Rules z 2015 roku. Słowa kluczowe: filozofia ekonomii, modele i prawa w ekonomii, ceteris paribus, ceteris normalibus, ontologia rzeczywistości społecznej. Klasyfikacja JEL: B41, B10, B50

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

Robert Mróz*

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS – AN APPLICATION TO METHODOLOGICAL VALUE JUDGEMENTS CONTAINED IN THE AUSTRIAN BUSINESS CYCLE THEORY AND THE REAL BUSINESS CYCLE THEORY1

ABSTRACT The paper aims to apply a simple “model of a model”, presented in detail in a companion paper, which draws on Max Weber’s discussion of values in social sciences. The argument in that paper was that to compare economic models in a thorough way, one should include in such comparisons value judgements expressed or assumed in these models. So, our “model of a model” should include these judgements, which is not common in the literature on economic modeling. The value judgements can be roughly divided into methodological and evaluative judgements, the latter concerning desirable policies, ethical issues, etc. In this paper the focus is on the former. Therefore, a case study is presented to show how some differences between the models in the Austrian Business Cycle Theory and the Real Business Cycle theory can be traced to the methodological * 1

Faculty of Economic Sciences, University of Warsaw; e-mail: rmroz@wne.uw.edu.pl

This research was financed by a research grant within the framework of the The National Programme for the Development of Humanities (NPRH) (grant no. 2bH 15 0266 83).

Robert Mróz

value judgements embedded in those models. These include the differing understanding of what the science of economics should look like, and choices between realisticness and simplicity. Some comments are also provided about the evaluative judgements in economic models. Keywords: Max Weber, Austrian Business Cycle Theory, Real Business Cycle Theory, methodological value judgements, evaluative value judgements, economic modeling. JEL codes: B41, B53

1. INTRODUCTION In a companion paper (Mróz forthcoming), summarized here in section 2, I argued that the literature on economic modeling should incorporate topics from the literature on value judgements in economics. As the first step towards the realization of this goal, I proposed a simple framework that could be used to appraise economic models, including their normative components. Its starting point was a classic take on the role of values in social sciences, i.e. Max Weber’s (1904/1949, 1895/1980) understanding of what it means for social sciences to be objective. Weber thought that even though economics is never truly value-free, this does not mean it is not objective as long as aims, interests, and values of a given researcher are clearly articulated. Another building block of the above-mentioned theory is the observation that modern economics is a model-based science (cf. Morgan, 2012). It means that, if there is a need to be able to identify values and political ends in economic research, then this need arises for modeling in particular. One of the well-known accounts of economic models, by Uskali Mäki (2009, 2011, 2013), was then built upon to formulate the amended account of models as used by economists. Mäki’s “model of a model” goes beyond standard realist descriptions of models as tools for representing some parts of the world. It includes a modeler using the model for a particular purpose, and an audience the model is addressed to. But if value judgements are to be included, then even more detail needs to be added to the description of the modeling agent. It was shown that Weber’s account is well suited to become the basis of such endeavor. The conclusion was that the advocated view provided a fuller account of economic models as it allowed to understand some differences between particular models as resulting from differing methodological or ethical value judgements, and not only, e.g., from their purposes or intended audiences. Having formulated this general proposition, what remains to be done is to show how it works in practice. A short outline of such a practical application was presented in the companion paper. Research on business cycles in Austrian tradition was juxtaposed with studies in the Real Business Cycles tradition. However, given that the purpose of that paper was just to introduce the pro-

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

posed approach, it was left for the current text to delve into details on particular examples. The main part of the current paper is, therefore, devoted to a detailed analysis and comparison of the canonical models in the Austrian Business Cycle Theory (ABCT) and in the Real Business Cycle (RBC) theory. These two traditions are chosen on purpose, as they share a lot in terms of normative outlook on free markets and state intervention in the economy. The normative recommendations that are built on the basis of these models are similar, and they point to the conclusion that activist policy in the face of business cycles is undesirable. This common normative ground between the schools allows me to downplay any differences that arise between them when it comes to evaluative value judgements (which concern desirable policies, ethical issues, etc.), and instead bring to light differences in methodological value judgements. Standard discussion about value-free economics is concerned with evaluative judgements, and this is also what Weber had in mind when he said that social scientists should be open about their values. It is not surprising that the debate centers on this issue as it is directly connected to the involvement of economists and their models in constructing economic policies of countries. Some topics, such as income and wealth inequality, are so widely discussed today that it is certainly desirable (a value judgement itself!) to bring to light the normative assumptions held by economic researchers. But it does not mean that methodological value judgements are unproblematic and the discussion about them should be discarded. As will be presented in the course of this paper, many methodological choices are not purely fact-based, and are not only driven by the attempt to build models that are good representations of the world. This then impacts the way these models are built, including the complexity of their explanatory mechanisms and the way their description (the language used to present them) looks like. Nonetheless, a short discussion of evaluative value judgements in economic modeling will also be provided, partly to highlight some difficulties in applying my framework to uncovering such judgements in economic models. One hopes these difficulties will be overcome in the future and a full-fledged case study will be presented that will do the opposite to the case study presented here – downplay the methodological differences, highlight the evaluative differences between the models. Given the above, the paper proceeds as follows. Section 2 briefly recounts the theory (the “model of a model”), and the justification for it, proposed in the companion paper. It also recalls the distinctions between various types of value judgements. Section 3 shortly presents canonical models in the ABCT and the RBC theory. Section 4, then, discusses different aspects of Austrian and RBC models, with the focus on methodological value judgements separating the two. It is intended to show why certain differences between particular models cannot be explained by standard components of philosophical accounts of economic models. Section 5 comments on evaluative value judgements in economic models. Conclusions follow.

Robert Mróz

2. THE MODEL OF A MODEL It is a famous Weberian stance that statements of fact are one thing and statements of value are another, and one should not confuse them in scientific research. This idea draws a strong contrast between the two, which runs counter to the newer developments in philosophy (Putnam, 2004; Putnam & Walsh, eds., 2011). For our purposes it is not necessary to adhere to such a strong version of the fact-value distinction, especially given that it is rather clear that research in economics is full of value judgements (more on this below and in the companion paper (Mróz forthcoming)). What is important is that the Weberian view serves as a useful starting point for formulating a prescription that is the basis of the account of economic modeling (the “model of a model”) proposed in the companion paper and illustrated with an example in the current paper. Before this account will be described, let us shortly recall the types of value judgements that can be involved in economic modeling. On the most general level, there are value judgements involved even in choosing one study area over some other, as it demonstrates that the person choosing attaches greater importance to the former than to the latter. Or there are other general normative commitments in science, such as accepting the primacy of logic and evidence over authority. Methodological value judgements, more narrowly understood, weigh characteristics such as simplicity, choice of formalisms, internal and external consistency, predictive power, etc., against each other (cf. Shrader-Frechette, 1994, ch. 3). Evaluative (or ethical, for simplicity) judgements, on the other hand, correspond to claims that something is good or bad, just or unjust, desirable or undesirable, etc. (cf. Baujard, 2013). Additionally, prescriptive judgements can be characterized as corresponding to statements of recommendation.2 In economics, they will most commonly be associated with policy recommendations, e.g. of the form “if the policy goal is X, then A, B, and C should be done to achieve this goal in the most effective manner.” But of course, there can be methodological prescriptions (the Weberian prescription that facts and values should always be disentangled is just that), which means that this last category cuts through the former two. Max Weber was primarily focusing on values which appear when one is analyzing things from the perspective of a specific policy end,3 but in current paper the methodological value judgements will be of primary interest. As a practical matter, what follows from the Weberian stance is a prescription that a scientist should “put her values on the table”, meaning – be open and 2 This

partition corresponds to what economists know very well, namely – J.N. Keynes's (1891/1999) division of economics into positive economics (dealing with facts), normative economics (dealing with values), and the art of economics (formulating prescriptions to help achieve any given aim). 3 In fact, Weberian account of value judgements in science was much broader and more nuanced, as was shortly summarised in the companion paper (see also Bruun, 2001, 2007). From the perspective of the current paper, however, the brief characterisation presented above is sufficient.

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

clear about them. However, in line with modern developments in analytical philosophy, but also in economics (in the works of Amartya Sen, for instance), it needs to be noted that such a clear-cut separation between facts and values is impossible in each and every instance of economic reasoning. Economics is permeated with value judgements, from choices between simplicity and realisticness of models to assuming possibility (or impossibility) of interpersonal comparisons of utility; from the widespread use of social value functions in welfare economics to specific arguments about inequality, so strongly visible in today’s popular discourse. In light of this, it is perhaps impossible to strictly follow Weber’s prescription at all times. But, as used in this paper, this Weberian perspective serves more as a guiding light than a strict code of conduct, and it does not assume one needs to be able to fully separate facts and values in each and every case. The modern claim that there is no strong ontological difference between the two does not mean that any attempt at separating them will be doomed to fail. On the contrary, as argued in Mróz (forthcoming), “it would rather reinforce the need of such separation inasmuch as we are able to do so, if only to avoid as much confusion as possible. And for the purpose of this paper, such lack of clear distinction would also not nullify, but rather reinforce the need to incorporate value judgements in our view of models and modeling practice in economics.” Given this Weberian prescription, and the fact that economics is a model-based science, the companion paper then proposed a view on economic models that took into account the value judgements expressed by the modeling agent. Modifying a well-known account by Mäki (2013, p. 91), it arrived at the following: “[ModRep2] Agent A, expressing value judgements contained in set V, uses (imagined) object M as a representative of (actual or possible) target R for purpose P, addressing audience E, at least potentially prompting genuine issues of resemblance between M and R to arise, describing M and drawing inferences about M and R in terms of one or more model descriptions D, and applies commentary C to identify and coordinate the other components.” (Mróz, forthcoming). The part in cursive letters is the addition to Mäki's account. In line with what was stated above about the limited possibility of separating values from facts, it might never be possible to specify all elements of V. But this is fine as long as some elements can be specified. Additionally, the word “expressing” is used here instead of some other, like “making”, to allow for the possibility that some value

Robert Mróz

judgements in a given model will not be made consciously or deliberately.4 Therefore, uncovering these judgements will often mean reconstructing them on the basis of analysis of a given model. The original Weberian prescription states that one has to explicitly distinguish between the normative and the positive elements of a given piece of research because normative elements will have an impact on policy conclusions. The argument here is that if this is the case, then it will be good to have an account allowing us to conceptualize how something like this could be done. The above “model of a model” is a simple account of this sort. Given this, we can see how the discussion about economic models can, and should, take into account value judgements expressed in these models. This is done below using a case study.

3. THE AUSTRIAN BUSINESS CYCLE THEORY AND THE REAL BUSINESS CYCLES MODEL The case study contrasts some aspects of the Austrian Business Cycle Theory (ABCT) and the Real Business Cycles (RBC) models.5 What follows are short descriptions of the canonical versions of both, and then the analysis of value judgements involved. The canonical model of the ABCT can be (crudely) summarized as follows (basing mainly on Garrison (2001) and Young (2015)). The central notion in the model is that of the structure of production. It is introduced instead of an aggregate “K” usually used in major mainstream business cycle models to denote capital stock. Using such aggregates is, according to the Austrians, a sure way to obscure real mechanisms responsible for the creation of cycles. The structure of production involves idealized stages of production, as visible in the so-called Hayekian triangle (fig. 1). Given this notion, the model works as follows. When the individuals voluntarily decide to increase their savings and abstain from current consumption, the supply of savings available as loanable funds increases. This is signaled to the producers and investors by the drop in the interest rate. Producers are notified that consumers are willing to decrease their short-term consumption in exchange for increased consumption in the future, which is an incentive to invest in more “roundabout”, and thus more efficient, methods of production. These methods, in turn, allow the producers to satisfy the increased future demand. Graphically, this would correspond to lengthening the horizontal leg of the triangle (injection of capital to the production stages far from consumption, creation of new stages) and to shortening its vertical leg (subtracting capital from the stages close to 4 Even though current paper deals mainly with methodological value judgements, this modification of Mäki’s account is in principle meant to encompass all types of value judgements. 5 As shown in Mróz and Hardt (forthcoming), the fact that the ABCT is a “theory” is not important for present purposes as it, and surely its specific instantiations in the literature, can be understood as a model.

Mining

ReямБning

EARLIER STAGES

Source: Garrison, 2001, p. 47.

VALUE

STAGES OF PRODUCTION/PRODUCTION TIME

Manufacturing

Distributing

Figure 1. The structure of production in the ABCT

LATER STAGES

Retailing

OUTPUT OF CONSUMER GOODS

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS... 77

Robert Mróz

consumption).6 This alignment of entrepreneurial decisions with consumer preferences generates sustainable growth, which requires abstaining from consumption in the short run in exchange for the expanded production and consumption in the future. Things happen differently when the interest rate is lowered not because of a voluntary increase in savings, but because of a central bank engaging in monetary easing, or because the fractional-reserve banking sector used its inbuilt money-creation mechanism. It is not necessary, for present purposes, to dig into the specifics of the model. Suffice it to say that such monetary intervention induces, in the Austrian model, a disconnect between entrepreneurs’ plans and consumers’ preferences. Entrepreneurs observe what they think is an increase in loanable funds, while consumers want to take advantage of reduced interest rates and increase their consumption. This imbalance then results in a bust when it transpires that there is not enough funds to finish the investments in more roundabout production methods.7 The bust is thus viewed as a necessary correction after a period of misalignments (called “malinvestment”) in the time structure of production in the economy.8 What has to be stressed at this point is that the ABCT, even in this simple form, is not restricted to blaming the boom-bust cycles solely on central banks. The setup of the banking system (e.g. the reserve requirements and, consequently, the associated multiplier in money creation leading to additional distorting effects) and other institutional considerations9 are vital components of the explanation in this model. Additionally, while graphical tools such as Hayekian triangles are sometimes used for expository purposes, the main descriptive work in the model is done in narrative style, i.e. using English vocabulary. The canonical version of the RBC theory, on the other hand, stresses other factors and is formulated using other tools, namely mathematical equations with 6 It

is important to note that the Hayekian triangle represents value, and not physical production and consumption. Barnett and Block (2006) thoroughly criticise, from within the Austrian paradigm, the use of Hayekian triangles even for expository purposes. However, as my aim here is not to provide a detailed and perfectly accurate depiction of the ABCT, but rather to highlight some of its methodological features, this criticism is largely irrelevant for the purpose at hand. 7 In principle the central bank could keep the boom going by continually injecting additional money into the economy, but on the Austrian account this will eventually result in significant inflationary pressures. Then, the central bank either switches to contractionary monetary policy (thus making loanable funds harder to attain and, consequently, uncovering the untenability of investment projects for which there are no funds) or induces hyperinflation. 8 This is what is usually presented as the ABCT, however one needs to remember this model was proposed by Mises ([1913] 1934) and Hayek (1933, 1935) in the context of the Great Depression. Young (2015) is an interesting attempt at showing that this is but a variant in a broader family of models. His own model stresses the time structure of consumption and the risk structure of the economy (instead of the traditional time structure) to better account for what happened in the US economy during the Great Recession of 2008. 9 For example, Young (2015) shows how changes in the risk structure of the economy, related mainly to operations of Fannie Mae and Freddie Mac, impacted the boom phase of the cycle (which, on the Austrian account, sows the seeds for the bust phase without any additional external shocks).

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

attached description. In the short presentation of this well-known model, I follow the basic textbook exposition by Romer (1996, chapter 4). The RBC is based on the Ramsey-Cass-Koopmans (RCK) growth model, which is “the natural Walrasian baseline model of the aggregate economy” (Romer, 1996, p. 151). Assuming exogenous technological progress, it presents how long-run economic growth without fluctuations comes about – to introduce fluctuations, one has to input some disturbances, such as exogenous shocks, heterogeneity among agents, or market imperfections, none of which are present in this model. It is a fully microfounded representative-agent model in which – as opposed to the Solow model where the savings rate is assumed constant – households (assumed to be identical and infinitely lived) make optimal consumption/ savings decisions. Therefore, there is an interaction between the utility-maximising households, who supply savings, and profit-maximising firms, who demand investment. In any case, the solution to the model is an optimal growth path that realizes households’ preferences over the time-path of consumption. The solution is (simplifying a great deal) arrived at by specifying a steady-state and dynamics of a system of two differential equations, one in capital and one in consumption. There are some obvious convergences with the Austrian model in the way some parameters work, which is not surprising given that these properties are rather intuitive. For example, the RBC model is constructed in such a way that saving is more desirable at higher interest rates, or that higher time-preference (more impatient agents) translates into more consumption today, but at a price of less capital accumulation, and therefore higher interest rates and less consumption in the long run. Two modifications are introduced to the RCK model in baseline RBC models to account for fluctuations in the economy. First, one has to introduce some external shock, and traditionally these were either shocks to the technology (translating into intertemporal changes in the production function), which change the amount produced from a given quantity of inputs (Kydland and Prescott, 1982; Long and Plosser, 1983), or changes in government purchases, which change the quantity of goods available to the private sector for a given level of production (Christiano and Eichenbaum, 1992; Baxter and King, 1993). These shocks are real as opposed to monetary, or nominal, disturbances – hence the name of the model family. Second, variations in employment have to be allowed to fit the observed facts about the business cycle. RBC models allow for changes in employment by introducing work-time as an argument in the households’ utility function. A classic example of a cycle-generating shock is a positive but temporary productivity shock which momentarily increases the output for any given level of inputs. Then the household decisions come in. One is the trade-off between consumption and savings. Having a greater output at their disposal, and valuing future consumption, a household will respond by consuming some of the increase but also investing more to increase future consumption. This explains one of the observed facts – that investment spending is more volatile than consumption spending. The other decision is the trade-off between labor and leisure. Higher

Robert Mróz

productivity means it is desirable to substitute work today for work in the future and decrease leisure, but it also means workers are earning more today, which can discourage additional work-time. Moreover, labor is pro-cyclical, which means the substitution effect is stronger than the income effect. As such a one-time shock translates into increased investment and, consequently, more capital in the future, it has not only one-time, but also persistent impact. A series of such shocks generates a boom. Conversely, a series of bad shocks generates a recession. Without the shocks, there are no business cycles in the RBC models. What is crucial here is that, in contrast to the ABCT, there is only one causal source of the business cycle. The other crucial thing is that households and firms respond optimally all the time. This means that, given the shocks, cycles of booms and recessions are preferred to different paths for the economy. This results in policy recommendations that are similar to the Austrian ones. As the ABCT implies that it is mainly the government involvement in the economy that generates the crises, the government should stay out of the economy. The RBC implies, in turn, that the crisis is the most efficient response of the economy to the external shocks and any attempts at improving the situation by using fiscal or monetary policy are misguided. This alignment of policy recommendations between the schools is useful for the current purpose, as it allows to push aside any potential differences in evaluative judgements involved and serves to underline the methodological value judgements.

4. METHODOLOGICAL VALUE JUDGEMENTS IN THE ABCT AND RBC MODELS In light of this feature, let us now go back to the model of a model presented in section 2, and compare the ABCT and the RBC models with regard to its components. For now, let us stick to Mäki’s account without the inclusion of value judgements. Both models have arguably the same general target, i.e. the cyclical fluctuations in economic performance. They also have the same purpose as they try to explain these fluctuations by identifying their causes. It is also highly plausible that they are intended to represent something. It is surely true for realistically-minded Austrian economists that they try to represent a part of real world. It is also not far-fetched to suppose something like this for the RBC theorists, especially as the intricate arguments in philosophical debates about whether models represent something or not are mostly the domain of philosophers of science and not practicing economists. It seems that some form of realism about representation is the default position among practitioners of economic modeling. When it comes to the audience, on one level it is the same as it consists of professional economic researchers. On another level, however, it is different. The RBC models are aimed mostly at mathematical economists operating in what we

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

could call the “mainstream”, while the reception of the ABCT is largely limited to the narrow Austrian audience. On the other hand, the Austrian economists are interested in getting their ideas across to the mainstream. So definitely there is at least some degree of convergence between the intended audiences of these two classes of models. The difference is most visible when it comes to the model description. The RBC model is expressed in mathematical terms with some added interpretation in English, while the Austrian model is expressed in the narrative style, without mathematics, but sometimes, as presented here, with the help of simple graphical devices that are useful for exposition purposes. And it is of course obvious that the substance of the explanation of how the boom-bust cycles come about is completely different between the two models. As mentioned in the companion paper, it is not appropriate to stop here, only noting these similarities and differences. One should go further, from Mäki’s model of a model to the updated version ModRep2 presented in section 2, because this allows a better understanding of why the content of these models, their intended audience (to some extent), and their description are different even though the target and purpose are the same. Of course, the difference between ModRep and ModRep2 lays in the latter’s explicit inclusion of value judgements. In the case of the ABCT and the RBC it will be most fruitful, as mentioned above, to focus on methodological value judgements. On the most general level, the ABCT underscores the crucial role of the structure of production, which follows from the typically Austrian insistence on the importance of capital theory in economics (cf. Garrison, 2001). The New Classical school, which produced the RBC models, together with all the other strands of mainstream economics, abstracts from capital theory, thus treating it as irrelevant to the explanation of business cycles. This is of course a consequence, to some extent, of the differing understanding between the schools of how the world actually works, but it does not seem far-fetched to suppose some of such differences trace back to the most general type of methodological value judgements mentioned before, the ones involved in choosing some research area over some other. Going further, it is easily noticeable that both models operate in line with the principle of methodological individualism (in economic parlance, they are microfounded). Both models concern decisions of individual actors and how these decisions impact the outcomes on the macro-level of the economy. But it seems that the understanding of methodological individualism differs between the schools. This methodological doctrine was introduced by Max Weber, most notably in Economy and Society (1922/1968).10 In Weber’s version, it claims that social phe10 Austrian economists, who consider methodological individualism an indispensable part of economics, would say that this principle was at work already in the writings of Menger (e.g. 1883/1985). As shown by Heath (2015), however, Menger’s version was actually different to Weber’s, and later exponents of the school, such as Mises, Hayek, and Lachmann, went with the Weber’s view.

Robert Mróz

nomena have to be explained by reference to individual actions, which in turn follow from the intentional states motivating individual agents. For him, the requirement of individualism followed from the vision of social sciences as verstehende (understanding/interpretive) sciences. Individual action is the only “subjectively understandable” component of sociological or economical explanations – given that actions are motivated by intentional mental states, we have interpretive access to them by virtue of our ability to comprehend the acting agent’s underlying motives. For Weber, this means that social sciences are different than natural sciences in that they allow this subjective understanding of actions (cf. ibid., p. 15). This, in turn (and noting that only individuals can possess intentional states), means that explanations invoking individual actions should take central role in social science as without knowing why people do what they do, we are unable to understand macro-level phenomena which result from the actions. So the goal of the explanation is understanding of the social phenomena. Austrians, following Mises (1949), often present economics as part of a broader science of human action, praxeology. Hülsmann (2001, p. 36) sums it up by saying that the Austrians are focused on analyzing human action while the economic mainstream is focused on the analysis of quantities of things that are subject to human action. This is very much tied to the Weberian notion of verstehende social science. Lachmann (1971, 1991), in particular, explicitly connects the interpretation of individual actions and expectations (as well as the meaning of institutions in economic and social life) to the Weberian tradition.11 When it comes to business cycle theories, Lachmann (1943) argues that all of them ultimately rest on some presuppositions about expectations. To conceptualize these expectations in an adequate manner, it is necessary to understand how agents interpret changing conditions. Because these interpretations vary, it is a mistake for an economist to treat them as fixed or governed by some mechanical updating rule. Instead, a proper economic explanation renders the individuals’ interpretations of the world intelligible (cf. Martin, 2015). This is but one way of explaining the firm Austrian stance against the use of mathematical methods in economics, or at least methods that are standard in the mainstream economics. For example, the mathematical structure of the baseline RBC model fixes a lot of parameters of the household’s choice. In the Austrian view, such fixing is not warranted if one wants to understand the dynamic nature of individual choices and actions. The Austrian approach, similar to Weber’s in this regard, is closely tied to their advocacy of radical subjectivism (cf. Yeager, 1987; Lavoie, 1991; Martin, 2015). For instance, Austrians will typically reject Marshall’s scissors metaphor in which subjective utility and objective cost of production influence prices. For them, as for Buchanan (1969), costs are also subjective. This relates to the conception of human action as multifaceted and dynamic, which does not lend itself to mathematical analysis in the spirit of mainstream economics. So the version of methodological individualism espoused by 11 Another

crucial Austrian text is Hayek (1942–44).

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

the Austrians, resulting from its inherent connection to Weber’s vision of verstehende social science and to the subjectivist stance, is at least in part responsible for the model description being devoid of mathematics in the case of ABCT. The RBC models, sitting within the general-equilibrium paradigm, also rely on some understanding of individual action – the one that is embedded in the microeconomic homo oeconomicus. Therefore, it is rather clear that they fulfil the demands of methodological individualism. It is also clear that any form of marginalist economics is to some extent subjectivist (e.g. when it comes to the concept of value). But the crucial difference here is that the mainstream economists, RBC theorists among them, are not so radically subjectivist as the Austrians, and they do not adhere to the strict connection between individualism, subjectivism and verstehende social science (see Hülsmann’s distinction above). A definition of methodological individualism largely devoid of such connections could be traced to Watkins (1957), who juxtaposed what he called “half-way” explanations with “rock-bottom” explanations. The former do not need to specify any mechanism on the individual level, but the latter do. The rock-bottom explanations should be preferred on this view, but not because they allow us some unique Weberian understanding, but because they are just deeper and more detailed. This does not mean, however, that half-way explanations are completely useless – they are some explanations, after all. These explanations could for example result from purely statistical analysis of correlations between economic variables. The New Classical school was known for their calls for microfoundations, especially on the basis of the Lucas critique (Lucas, 1976), but it never meant statistics could not supply any explanations. For Austrians, however, statistics alone explains nothing. The foregoing discussion allows us to assert the following. The commitment to methodological individualism in Austrian economics is inextricably tied to subjectivism, which is, in turn, tied to a specific vision of what social science should be – namely Weberian verstehende science. Subjectivism opens one way of criticism of mathematical methods prevalent in mainstream economics. This then translates into how specific models are being expressed, and in our case – it explains why the description (understood as one of the elements of ModRep2) of the ABCT looks the way it does. On the other hand, methodological individualism embedded in the RBC models is not tied to such commitments about the nature of social sciences, and economics in particular. If anything, it rather follows Popper (1945), who claimed (in the section tellingly titled “The Unity of Method”) that both social and natural sciences are concerned with “causal explanation, prediction and testing” (ibid., p. 78), and there is no necessary difference of method between them.12 Such understanding is of course only one of many sources of the drive towards mathematisation in the 20th-century mainstream 12 For Polish-language readers, a comprehensive reference for both value judgements in mainstream economics and a defence of methodological monism of all empirical sciences is Czarny (2010).

Robert Mróz

economics. The literature on this topic is extensive13 and there is no need to delve into it in this paper. What is important from the perspective of this paper is that the difference between the descriptions employed to present these two models to the audience stems, ultimately, from the discord in the respective visions of what economics should be and what should be its primary focus. Should economists be in the business of understanding, or just in the business of causal explanation similar to natural sciences? But of course, this is not something that can be argued for or against solely on the basis of scientific arguments pertaining to the nature of the studied subject matter. At some point the fact-based arguments about what method is more adequate to the subject matter end. It can reasonably be asserted that both visions can have place in social sciences, and both can provide their own types of insights (following up on the Hülsmann’s distinction – it seems that both the science of human action and the science of the quantities subject to human action have a role to play). Therefore, ultimately, at least part of the difference lies in the preference towards one or the other mode of doing economics. In other words, it lies in a methodological value judgement. The discussion of methodological individualism and subjectivism shows one issue with employing the framework proposed in ModRep2. In this example it was necessary to go very far beyond the confines of the discussed models to understand the underlying value judgements. The big picture, the general outlook on what economics is or is not, needed to be invoked to understand why descriptions of these two models vary. Such analyses, then, run the risk of quickly becoming overwhelming. To identify all implicit value judgements in some evaluated model one might have to guess the meaning of some statements, consult the whole body of theory, the author’s biography, normative opinions expressed elsewhere, etc. This concern was already mentioned in the companion paper, and the response was that the prescription urging the user of a piece of research to identify the value judgements was intended as a piece of practical guidance. It does not require the reader to uncover all the value judgements in the process of interpreting a model, as this would be impractical, but it elucidates certain awareness and attitude towards economic models. It seems especially useful in light of the prevalent opinion in economics that it is by and large a value-free science. This first response notwithstanding, it can also be shown that not all value judgements are so deeply buried and require so much work to uncover. For this, consider the second example relating to the ABCT and RBC models. As mentioned above, the ABCT should not be interpreted as pointing to only one single source of cyclical fluctuations. With the Austrians emphasising the structure of production, the activities of the central bank and the banking sector 13 See

e.g. Weintraub (2002). For the examples from the post-Great Recession debates on mathematisation of economics, see e.g. Colander et al. (2009); Colander (2011). For a critique of the mathematical economics from the perspective of the philosophy of science, see e.g. Lawson (1997, 2003, 2009).

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

operating under the fractional reserve are one source of disturbances, but the institutional setting will also impact how the cycle develops. The complexity of capital, the workings of the price system as an information-conveying mechanism, the entrepreneurial calculations – all this also enters the explanation (cf. Young, 2015). Therefore, as it does not point to only one single cause of recessions, for our purposes this model could be termed “multicausal.” On the opposing end, the RBC model has exactly one source of cycle-generating disturbance, and in the canonical version it is the exogenous technology shocks. So it is more of a particular, single hypothesis, and so we could term this model “monocausal” in this sense. The weight the Austrian economists attach to the realisticness of their models, and to their ability to explain how the world actually works, is well-known. The view that “understanding the causes of aggregate fluctuations is a central goal of macroeconomics” (Romer, 1996, p. 146), including the RBC models, is also widespread in mainstream macro. Given this, we were able to conclude earlier that both of these models are meant to explain some features of reality. But given that one of these models is multicausal, and the other is monocausal, the complexity of their respective explanations differs. This has implications for the methodological assessment of these models. The baseline RBC model was first proposed to account for a few stylized facts about the business cycles, some of which were mentioned before. But it is of course known that the calibrated model14 fails to fit the data in many respects (e.g., it underestimates the volatility in labor input relative to the volatility of output). This means that, if the underlying idea that exogenous shocks are responsible for business cycles is sound, there must be some other, or additional, source of shocks. Or maybe one has to add the nominal dimension to the model after all, given that there is none in the baseline version (which is the path macroeconomics actually took with the introduction of price and wage stickiness, etc.). This means that the explanation was at best incomplete from the very start and it was so, crucially, judging by the criteria that the theorists set up for themselves (i.e., the fit with the actual data). But the model was still considered one of the most important developments in modern macroeconomics. This suggests that a trade-off was in play between the realisticness and success of the explanatory mechanism, on the one hand, and something else, on the other hand. Normally, if there is a trade-off between realisticness and something else, this “something else” is simplicity or precision. Given the monocausal nature of the explanation in the RBC model, simplicity is definitely there. As for precision, it is generally assumed (sometimes implicitly) as desirable in mainstream economics, as evidenced by the ubiquitous drive for specific numerical results. The Austrians, on the other hand, will claim that considerations of realisticnesstrump all other possible choices, at least to a point where further realistic14 The baseline calibrated model referenced here is as described in Romer (1996, Ch. 4.9), which means it is taken from Prescott (1986) and Hansen (1985), with parameter values assigned by Hansen and Wright (1992).

Robert Mróz

ness becomes overwhelmingly impractical (because obviously the Austrian explanation contains various simplifications and omissions as well – cf. Mróz and Hardt (forthcoming)). The multicausal nature of their model – which makes the model more complicated in its explanatory mechanism – is a direct outcome of such thinking. What these models represent, then, is the outcome of methodological choices on the realisticness-simplicity axis. There is no obvious, strictly fact-based way of determining which of these choices is better. The drive towards realism will be in some sense better if the goal is to explain the whole phenomenon at once. But if the goal is to provide a baseline model, and then to work on it by adding new layers, new parameters, sources of shocks, etc., then simplicity will perhaps be better, as it will allow other researchers to understand the basic features of the model before adding something to it. Given this, even a simple, “unrealistic” explanation still fulfils the demands of the RBC model’s purpose, which is to explain. It is still some explanation, after all. Therefore, the methodological choices, such as the one involved in deciding between realisticness and simplicity, will always involve methodological value judgements related to what a given researcher prefers in her particular model. (Note how, in contrast to the previous example involving individualism and subjectivism, the discussion in this one was confined to the models themselves to a much greater degree.) The two examples discussed in this case study do not exhaust all possibilities of uncovering methodological value judgements in these two models. Much could be said, for example, about the difference between rational expectations assumed in the RBC model and the understanding of expectations among Austrian economists15, or about the use of the concept of the representative agent in mainstream economics.16 What these examples do show, however, is that at least some important differences between models do stem from methodological value judgements.17 These differences cannot be traced to the modelers’ purposes, intended 15 Yeager (1987, p. 17): “Since expectations are formed by people, they are understandably loose, diverse, and changeable.” 16 One of the consequences of using this concept is that in such models there is no possibility to obtain emergent properties on the macro-level, so properties which would not be reducible to the properties on the micro-level. In modern mathematical economics this assumption is lifted, e.g., in agent-based models. 17 Here it is perhaps instructive to draw a connection between methodological value judgements and methodological assumptions. It is of course difficult to compare models without discussing their methodological assumptions, and it is being habitually done in most such exercises. But it does not mean that the latter are synonymous with methodological value judgements. While methodological assumptions will often concern some very specific characteristics of models (like the choice of values of specific parameters in the production function; the choice between rational and adaptive expectations; the choice of estimation methods in empirical models), methodological value judgements are a broader category encompassing, as we have seen above, considerations such as what is the nature of economic sciences, what is the proper balance between simplicity and realisticness, etc. There is, therefore, some convergence between the two concepts (the choice between various types of expectations can be seen as a methodological assumption, but also as a methodological value judgements – as hinted at above), but value judgements should be seen as a broader concept.

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

audiences, issues of representation (whether the model is intended to represent something or not), or to model descriptions. This means that to include value judgements in model comparison exercises is to add an important new element to these comparisons that is not, as of yet, represented in the literature on economic modeling.18

5. A COMMENT ON EVALUATIVE JUDGEMENTS This paper is mostly concerned with analyzing methodological value judgements, but it is more in the spirit of Weber’s discussion of value-free science to focus on the evaluative judgements which enter the normative considerations of economists when it comes to policy recommendations. These types of judgements are also at the heart of the discussion about the value-free science. The value added of the previous section is, therefore, to show that methodological value judgements, even though they do not seem to impact policy advice directly in the way their evaluative counterparts do, have an important influence on how the models look like, including how the explanatory mechanisms in these models look like. To balance the previous section, in which models were compared in a manner that allowed to push evaluative differences aside and focus on methodological differences, it would be best to engage in another case study, one which would push methodological differences aside and focus on evaluative differences. Therefore, it would be best to focus on two models from the same family, or school of thought (which would ensure methodological similarity), that differ in their evaluative components. It is, however, not a straightforward task in economics as schools or traditions of thought, which produce families of models, are most often defined not only on the basis of their methodological convictions, but also on the basis of their evaluative convictions. Consider New Classicals and Austrians on one side, and Keynesians on the other. Normative convictions of the former two (e.g., related to the proper role of governments in the economy, or in the time of crisis specifically) are underpinned by the agreement between two schools that there exists a kind of natural harmony, or a kind of equilibrating process, in the market. Normative convictions of the latter, however, are underpinned by the opposing notion – that the market system does not have a built-in equilibrating mechanism. (And such convictions are then impacting normative opinions on issues such as growth, inequality, regulations, etc.) Therefore, one of the characteristics of all models produced within 18 It

is perhaps worth noting that Austrian economists would normally deny that value judgements enter the positive science of praxeology (cf. Rothbard, 1976). But such statements normally refer to the evaluative judgements, and not to the methodological value judgements, mostly because the evaluative judgements are the main issue of the debate on whether science can be “value-free”. Therefore, this Austrian position has no direct bearing on the case study shown here.

Robert Mróz

the New Classical school, or within the Austrian school, is that the modelers will share relevant convictions, and the analogue is true for Keynesians. This means that no readily apparent differences in evaluative judgements (regardless whether they are openly expressed or just assumed, or maybe just embraced unknowingly) will exist between models belonging to one and the same family. This, in turn, means that most of the differences in evaluative judgements between a given pair of models will stem from the broader paradigms these models were formulated in. In other words, most normative differences (i.e. differences in evaluative judgements) seem not to be embedded at the level of individual models, but at the level of broader thought traditions or schools. This is similar to the difference, discussed above, in the vision of economics as science between the Austrians and the mainstream theorists. Such visions define schools, and the differences between schools then impact how the specific models look like. It is dissimilar, however, to the second example described in the previous section, about the trade-off between realisticness and simplicity (which is a simplification itself, one might add). That kind of methodological value judgement is decided on a model-by-model basis, irrespective of a broader thought tradition a researcher might belong to. (There exist, of course, RBC models much more complicated than the baseline version.) The hypothesis stated above – that most of the evaluative differences between models are in fact differences between schools, so evaluative differences will be accompanied by methodological differences19 – is not a good starting point if a case study showing these differences needs to be analogous to the one presented in the previous section in that it will highlight one type of value judgements while suppressing the other. Such case study would ideally pick models from the same family, but then the intuition is that the evaluative differences would be hard to grasp. For now, this has to be seen as a limitation of the account to modeling proposed in the current paper. In its present form, the account is well-suited to analyze differences in methodological value judgements between models, but less so to analyze differences in evaluative judgements, at least as long as the models belong to the same school of thought. This does not mean, however, that a relevant case study has to pick models from the same school of thought. It is in principle possible to account for evaluative judgements on the basis of models picked from different families, just remembering to control for their methodological differences. However, given the space that would be required to introduce a new set of models and then to analyze them while controlling for these irrelevant differences, this is left for future research to pick up. And, in any case, the discussion in this section is based on broadly sketched remarks and general outlook on how economics looks like, which gives rise only to certain intuitions and hypotheses and not to definitive 19 For

example, while both Austrians and RBC theorists require microfoundations, there is no such thing in traditional Keynesianism. Note, however, that the implication does not necessarily work the other way around and, as shown by the ABCT and RBC case study, methodological differences are not always accompanied by evaluative differences.

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

conclusions. So it is not in fact settled that a case study cannot be undertaken that would be analogous to the one presented in this paper, only that certain difficulties seem to exist in doing so. The companion paper mentioned mainstream monetary economics as a potentially promising avenue for this kind of endeavor. The current paper dealt primarily with methodological value judgements, therefore different targets were chosen for the case study. What remains to be done, then, is to engage in more detail the issue of evaluative judgements in economic models.

6. CONCLUSION This paper built on the idea, novel in the literature on economic modeling, that it is worth incorporating into this literature the ideas from the discussions on value judgements in economics. These value judgements can be, for the purposes of this study, divided into two groups: methodological and evaluative (with the additional third category, prescriptive judgements, cutting through them and making it possible to classify some judgements as methodological prescriptions and some as evaluative prescriptions). A complete argument supporting such incorporation should show that an account of economic models which includes value judgements is able to provide some new dimension to understanding the content, purpose, description, etc., of economic models. The current paper went some way towards providing such an argument by comparing two models (ABCT and RBC) and highlighting some of their respective methodological value judgements. These judgements, which cannot be reduced to purely fact-based decisions of researchers, are responsible for non-trivial characteristics of these respective models, such as the relative complexity of their explanations of business cycles as well as the languages used to provide their descriptions. Therefore, the inclusion of value judgements (or at least of their methodological subset) in appraising economic models proves useful in providing a fuller account of these models.

REFERENCES Barnett W., Block W. (2006), On Hayekian Triangles, “Procesos de Mercado”, Vol. 3, No. 2, pp. 39–141. Baxter M., King R.G. (1993), Fiscal Policy in General Equilibrium, “American Economic Review”, Vol. 83, pp. 315–334. Baujard A. (2013), Value Judgements and Economics Expertise, Working Paper GATE 2013–14. Bruun H.H. (2001), Weber on Rickert: From Value Relation to Ideal Type, “Max Weber Studies”, 1, pp. 138–160. Bruun H.H. (2007), Science, Values and Politics in Max Weber’s Methodology (New Expanded Edition), Ashgate Publishing, Aldershot.

Robert Mróz

Buchanan J.M. (1969), Cost and Choice, Markam, Chicago. Christiano L.J., Eichenbaum M. (1992), Current Real-Business-Cycle Theories and Aggregate Labor-market Fluctuations, “American Economic Review, Vol. 82, pp. 430–450. Colander D. (2011), How Economists Got It Wrong: A Nuanced Account, “Critical Review”, Vol. 23, No. 1–2, pp. 1–27. Colander D. et al. (2009), The Financial Crisis and the Systematic Failure of the Eco nomics Profession, “Critical Review”, Vol. 21, No. 2, pp. 249–267. Czarny B. (2010), Pozytywizm a sądy wartościujące w ekonomii, Szkoła Główna Handlowa, Warszawa. Garrison R. (2001), Time and Money: The Macroeconomics of Capital Structure, Routledge, New York. Hansen G.D. (1985), Indivisible Labor and the Business Cycle, “Journal of Monetary Economics”, Vol. 16, pp. 309–327. Hansen G.D., Wright R. (1992), The Labor Market in Real Business Cycle Theory, Federal Reserve Bank of Minneapolis “Quarterly Review”, Vol. 16, pp. 2–12. Hayek F. v. (1933), Monetary Theory and the Trade Cycle, Augustus M. Kelley, New York. Hayek F. v. (1935), Prices and Production, Augustus M. Kelley, New York. Hayek F. v. (1942), Scientism and the Study of Society (parts I–III), “Economica”, Vol. 9, pp. 267–291; Vol. 10, pp. 34–63; Vol. 11, pp. 27–39. Heath J. (2015), Methodological Individualism, The Stanford Encyclopedia of Philosophy, E. N. Zalta (ed.), online: https://plato.stanford.edu/entries/methodological-individualism/ Hülsmann J.-G. (2001), Garrisonian Macroeconomics, “The Quarterly Journal of Austrian Economics”, Vol. 4, No. 3, pp. 33–41. Keynes J.N. (1891/1999), The Scope and Method of Political Economy, Batoche Books, Kitchener. Kydland F.E., Prescott E.C. (1982), Time to Build and Aggregate Fluctuations, “Econometrica”, Vol. 50, p. 1345–1370. Lachmann L. (1943), The Role of Expectations in Economics as a Social Science, “Economica”, Vol. 10, pp. 12–23. Lachmann L. (1971), The Legacy of Max Weber, Glendessary Press, Berkeley, CA. Lachmann L. (1991), Austrian Economics: A Hermeneutic Approach, in Economics and Hermeneutics, D. Lavoie (ed.), Routledge, London, New York, pp. 132–144. Lavoie D. (1991), The Progress of Subjectivism, in Appraising Modern Economics, M. Blaug, N. de Marchi (eds.), Edward Elgar, Aldershot, pp. 470–486. Lawson T. (1997), Economics and Reality, Routledge, London, New York. Lawson T. (2003), Reorienting Economics, Routledge, London, New York. Lawson T. (2009), The Current Economic Crisis: Its Nature and the Course of Academic Economics, “Cambridge Journal of Economics”, Vol. 33, pp. 759–777. Long J.B., Plosser C.I. (1983), Real Business Cycles, “Journal of Political Economy”, Vol. 91, pp. 39–69. Lucas R.E. (1976), Econometric Policy Evaluation: A Critique, “Carnegie-Rochester Conference Series on Public Policy”, Vol. 1, pp. 19–46.

WEBERIAN PERSPECTIVE ON VALUE JUDGEMENTS IN ECONOMIC MODELS...

Martin A. (2015), Austrian Methodology: A Review and Synthesis, in The Oxford Handbook of Austrian Economics, P.J. Boettke, C.J. Coyne (eds.), Oxford University Press, Oxford, pp. 13–42. Mäki U. (2009), MISSing the World. Models as Isolations and Credible Surrogate Systems, “Erkenntnis”, Vol. 70, No. 1, pp. 29–43. Mäki U. (2011), Models and the Locus of Their Truth, “Synthese”, Vol. 180, pp. 47–63. Mäki U. (2013), Contested Modeling: The Case of Economics, in: Models, Simulations, and the Reduction of Complexity, U. Gähde et al. (eds.), De Gruyter, Berlin/Boston. Menger C. (1883/1985), Investigations into the Method of the Social Sciences with Special Reference to Economics, New York University Press, New York. Mises L. v. (1913/1934), The Theory of Money and Credit, Yale University Press, New Haven, CT. Mises L. v. (1949), Human Action, Yale University Press, New Haven, CT. Morgan M.S. (2012), The World in the Model: How Economists Work and Think, Cambridge University Press, Cambridge. Mróz R. (forthcoming), Value Judgements and Economic Models – a Weberian Perspective, “Studia Metodologiczne”. Mróz R., Hardt Ł. (forthcoming), Economic Modeling – Beyond Isolation and Con struction, “Acta Oeconomica”. Popper K. (1945), The Poverty of Historicism III, “Economica”, Vol. 11, pp. 69–89. Prescott E. C. (1986), Theory Ahead of Business-cycle Measurement, “CarnegieRochester Conference Series on Public Policy”, Vol. 25, pp. 11–44. Putnam H. (2004), The Collapse of the Fact/Value Dichotomy and Other Essays, Har vard University Press, Cambridge, MA. Putnam H., Walsh V. (eds.) (2011), The End of Value-Free Economics, Routledge, London. Romer D. (1996), Advanced Macroeconomics, McGraw-Hill. Rothbard M.N. (1976), Praxeology, Value Judgements, and Public Policy, in The Foun dations of Modern Austrian Economics, E. G. Dolan (ed.), Sheed & Ward, Kansas City, pp. 89–111. Shrader-Frechette K.S. (1994), Ethics of Scientific Research, Rowman & Littlefield, Lanham. Watkins J. W. N. (1957), Historical Explanation in the Social Sciences, “British Journal for the Philosophy of Science”, Vol. 8, pp. 104–117. Weber M. (1922/1968), Economy and Society, G. Roth, C. Wittich (eds.), University of California Press, Berkeley, CA. Weber M. (1895/1980), To Methodological Value Judgements Contained in the Austrian Business Cycle Theory and the Real Business Cycle Theory. The National State and Economic Policy (Freiburg Address), “Economy and Society”, Vol. 9, No. 4, pp. 428–449. Weber M. (1904/1949), ‘Objectivity’ in Social Science and Social Policy, in: The Methodology of Social Sciences, E.A. Shils, H.A. Finch (eds.), The Free Press, Glencoe, Il. Weintraub E.R. (2002), How Economics Became a Mathematical Science, Duke University Press, Durham, NC.

Robert Mróz

Yeager L. (1987), Why Subjectivism? “The Review of Austrian Economics”, Vol. 1, pp. 5–31. Young A.T. (2015), Austrian Business Cycle Theory: A Modern Appraisal, in: The Oxford Handbook of Austrian Economics, P.J. Boettke, C.J. Coyne (eds.), Oxford University Press, Oxford, pp. 186–212.

WEBEROWSKIE PODEJŚCIE DO SĄDÓW WARTOŚCIUJĄCYCH W MODELACH EKONOMICZNYCH – ZASTOSOWANIE DO METODOLOGICZNYCH SĄDÓW WARTOŚCIUJĄCYCH ZAWARTYCH W AUSTRIACKIEJ TEORII CYKLI KONIUNKTURALNYCH ORAZ W TEORII REALNYCH CYKLI KONIUNKTURALNYCH STRESZCZENIE W artykule zastosowano prosty „model modelu” – szczegółowo opisany w powiązanym tekście – w którym wykorzystano poglądy Maksa Webera na rolę wartości w naukach społecznych. W tymże powiązanym tekście argumentowałem, że, aby porównywać modele ekonomiczne w gruntowny sposób, należy w tych porównaniach uwzględnić sądy wartościujące wyrażane lub przyjmowane w tych modelach. Zatem nasz „model modelu” powinien zawierać te sądy, co nie jest powszechne w literaturze na temat modelowania ekonomicznego. Sądy wartościujące można w uproszczeniu podzielić na metodologiczne i ewaluatywne. Te drugie dotyczą kwestii etycznych, pożądanych kierunków polityki itd. W niniejszym artykule skupiam się na tych pierwszych. W związku z tym prezentuję studium przypadku i pokazuję, w jaki sposób niektóre różnice pomiędzy modelami związanymi z austriacką teorią cykli koniunkturalnych i modelami związanymi z teorią realnych cykli koniunkturalnych powstają ze względu na odmienne metodologiczne sądy wartościujące zawarte w tych modelach. Mamy tu m.in. do czynienia z różnym rozumieniem istoty nauk społecznych, a także z niekompatybilnymi wyborami pomiędzy realistycznością i prostotą. W końcowej części artykułu znalazł się również komentarz na temat sądów ewaluatywnych w modelach ekonomicznych. Słowa kluczowe: Max Weber, austriacka teoria cykli koniunkturalnych, teoria realnych cykli koniunkturalnych, metodologiczne sądy wartościujące, ewaluatywne sądy wartościujące, modelowanie ekonomiczne. Klasyfikacja JEL: B41, B53

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

Krzysztof Nowak-Posadzy*

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS1 ABSTRACT Economics became a model-based science. As economic modeling, namely model-building and model-appraisal, involves both informal craft and formal procedures, the aim of the article is twofold. Firstly, to enter into the discussion about the practice of economic modeling by drawing attention to the possibility of two implicit modes of model appraisal in economic science. We will start with a thought-provoking recent proposal by philosophically and methodologically sensitive economist Dani Rodrik, whose insights into the art of model selection has already triggered a number of in-depth philosophical commentaries, and has been the subject of a few successful formal reconstructions by philosophers of economics. Secondly, the aim is to supplement this philosophical view of model selection with an account that aims at underlabouring for the reconstruction of a distinct implicit mode of model criticism, which for various reasons has been pretty much absent in recent philosophical and economic discussions. Keywords: philosophy of economics, economic model, model appraisal, model selection, model criticism. JEL codes: B41, A14, B40 *

Institute of Philosophy, Adam Mickiewicz University in Poznań; e-mail: k_nowak@amu. edu.pl 1 This research was financed by a research grant within the framework of The National Programme for the Development of Humanities (NPRH) (grant no. 2bH 15 0266 83).

Krzysztof Nowak-Posadzy

The epistemic genre of creating and reasoning with models requires a craft skill working with highly formal instruments. (Morgan, 2012, p. 399) Economists need to err to be right, and if they get their errors right then they will be right and able to model with no error. (Louçã, 2007, p. 255)

1. INTRODUCTORY REMARKS: ECONOMIC MODELING – BETWEEN SCIENCE AND ART Since the publication of Mark Blaug’s seminal book titled “Methodology of economics, or, how economists explain” (Blaug 1980[1992]), the philosophy of economics profession has put a lot of effort and invention into recovering the practical dimension of economic science (e.g. de Marchi, 1992; Backhouse, 1994; Boumans, Leonelli, 2013). This philosophical focus on the research practice in economic science was inspired by two conspicuous methodological and epistemic trends: 11

the shift in the general philosophy of science from the analyses of the cognitive status and methodological features of scientific theories to the studies of theoretical and empirical models. Such a transition followed both focusing on how science is actually done, rather than on how science should be done, as well as questioning the received view of scientific theory (e.g. Hacking, 1983; Suppe, 1989; Cartwright, 1999; Zeidler, 1997);

the growing plurality of models in economic science (e.g. Mäki, 1997; Sent, 2006; Dow, 2007; Dow, 2012; Gräbner, Strunk, 2018) due to which economics became a model-based science (cf. Morgan, 2012; Maas, 2014; Boland, 2014; Rodrik, 2015). An economic model has thereby gained the status of the basic unit of production, communication and utilization of economic knowledge in the intradisciplinary, interdisciplinary and extradisciplinary exchanges. This gradual transformation of economics into modeling science should not be surprising when we compare the path economics follows to the path already set out by number of natural, cognitive and computer sciences (cf. Gelfert, 2016; Magnani, Casadio, 2016; Magnani, Bertolotti, 2017). As it will be presented in this paper, the growing plurality of models in economics can be heuristically expressed by means of two metaphors – economic science as a library and economic science as a marketplace of ideas.

As a consequence of these two trends, the majority of philosophers of economics declare today the willingness to investigate in detail how economists actually practice their trade. On the one hand, since contemporary economics became

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

a model-based science, the nature, specificity and features of economic models, as contrasted with economic theories,2 have become the objects of detailed philosophical investigations (see: Erkenntnis, 2009). On the other hand, since contemporary philosophy of science differentiates between scientific theorizing and scientific modeling (e.g. Hacking, 1983), some philosophers of economics have recently offered their accounts of what the economic modeling actually is (e.g. Mäki, 2013; Spiegler, 2015; Gilboa et al., 2018). Still, there are some essential difficulties and substantial differences within the philosophy of economics profession3 about how to philosophically and methodologically investigate the economic research practice (e.g. economic modeling) that to a large extent rests on the tacit (or implicit) type of knowledge, craft skills, knowing-how, connoisseurship or expert judgment. As it is not intended in this article to discuss the nature and features of economic models, the general focus will be placed on the practice of economic modeling. To do so, let us first make three specifications that will motivate, organize and guide our further investigations: (1) the philosophical account of economic modeling that is acknowledged and used here is a revised version of Peter Spiegler’s account, according to which: “an economist E uses model M to establish a certain likeness with the target T for intended purposes P and the success of E in accomplishing P is judged against disciplinary norms N” (Spiegler, 2015, p. 25; cf. Mäki, 2013, p. 91); (2) the cycle of economic modeling is decomposed into two phases: (a) model-building (constructing, formulating) and (b) model appraising (testing, checking, validating, assessing), which fits into Spiegler’s conceptual framework, as well as other traditional accounts from the general philosophy of science, philosophy of economics and economic methodology (e.g. Żytkow, 1999; Mäki, 2008; Boland, 2014). The important point to be made here is that the success of an economist in model-building and model appraising is judged against certain suitable intradisciplinary norms that are sufficiently shared in a certain way by the members of economics profession. It is then argued that in the context of model appraisal two sets of research 2 For the discussion about the meaning of “theory” and “model” in economics see: (Klein, Romero, 2007; Goldfarb, Ratner, 2008). 3 However, despite the shift towards the practice of economics, there are still three possible formulas of methodological investigations in the philosophy of economics: (i) normative investigation – that aims at elaborating recommendations (guidelines) regarding good research practices and thus at revising received set of practices, procedures and scientific products in a given discipline (e.g. Lawson, 1997); (ii) descriptive investigation – that aims at systematizing explicit declarations of researchers and thus at reporting the state of affairs concerning the set of practices, procedures and scientific products in a given discipline (e.g. Weintraub, 1989); (iii) reconstructive investigation – that aims at reconstructing implicit (tacit) presuppositions behind research practice in a given discipline, and thus to complement (supplement) a given set of practices, procedures and scientific products with a certain context, within which its specific construction (grammar) emerges (e.g. Mäki, 1994).

Krzysztof Nowak-Posadzy

norms may come into play, standing behind the two implicit modes of model appraisal, namely norms of models selection and norms of model criticism. (3) the community of economic modelers “in order to make the model work [is] using (a) a tacit, craft-based, knowledge as much as (b) an articulated, scientific, knowledge” (Morgan 2012, p. 25; see also: Morgan, Magnus 1999) and that claim not only conforms to Spiegler’s account on disciplinary norms, but also enables to use some insightful philosophical contributions about two dimensions of scientific practice and dual nature of scientific knowledge by Percy Williams Bridgman (1959), Gilbert Ryle (1949) and Michael Polanyi (1966) or Harry Collins (2010). It is then argued that in the context of model appraisal one can distinguish between formal explicit criteria and informal implicit modes of model selection and model criticism. By combining (2a), (2b), (3a) and (3b) we obtain a simple and tentative, yet helpful for philosophical investigation of economic science, typology (table 1). Table 1. Typology for philosophical investigation of economic modeling model-building phase

mode appraisal phase

explicit dimension

formal methods

explicit criteria

tacit dimension

informal styles (genres, strategies)

implicit modes (ways of thinking, attitudes)

Source: own elaboration.

As the discussion about the explicit dimension of economic modeling (quadrant I and II) can in no way be restrained to the exclusive domain of the economics profession, the deliberations about the tacit dimension (quadrant III and IV) can by no means be delegated solely to the domain of the philosophy of economics. On the one hand, a number of philosophers of economics (more systematically) and some economists (more occasionally) have contributed to the discussion about styles (genres, strategies) of model-building and modes (ways of thinking, attitudes) of model appraisal in economic science. On the other hand, many economists (more systematically by either innovating new methods or importing existing ones from mathematics and statistics) and some philosophers of economics (more occasionally) have contributed to an increasing plurality of formal methods of model-building and explicit criteria of model-appraisal in economic science. An excellent example of the insights into discussion about both dimensions and phases of the practice of economic modeling is the one offered recently by philosophically and methodologically sensitive economist, Dani Rodrik in his extensively commented book titled Economics rules. Why economics works, when it fails, and how to tell the difference (Rodrik, 2015).

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

The aim of this article is twofold. Firstly, to engage in the discussion about the practice of economic modeling by drawing attention to the possibility of two implicit modes of model appraisal in economic science. This implies departing from occasional, yet thought-provoking, recent proposal by Rodrik, whose insight into the art of model selection has already triggered a number of in-depth philosophical commentaries, as well as has received so far a few coherent formal reconstructions by philosophers of economics. The second aim is to supplement this received philosophical view of model selection with an account that aims at underlabouring for further reconstruction of a distinct, implicit mode of model criticism which, for various reasons, has been pretty much absent in the reflection on economics. The structure of this paper is as follows. In the introductory remarks we sketch the basic distinction between formal methods and informal styles of model-building, as well as between explicit criteria and implicit modes of model appraisal. In the next section we explore the metaphor of economics-as-a-library-of-models, as recently offered by Rodrik. It is then argued, that, by making use of the heuristic potential of this metaphor, Rodrik attempts to redirect the economics profession’s attention from formal methods of model-building to the neglected informal mode of model appraisal, namely, the art of model selection (section 2). As this shift has been also addressed by some contemporary philosophers of economics, we will summarize a formal reconstruction of the art of model selection by Till Grüne-Yanoff and Caterina Marchionni (section 3). Afterwards we will argue that the mode of model selection does not exhausts the very art of model appraisal in economic science. There are good philosophical, theoretical and statistical reasons to distinguish then the second implicit mode of model appraisal, namely the art of model criticism. The discussion of the latter will start with delineating the metaphor of economics-as-a-marketplace-of-models (section 4). This metaphor is a useful starting point to turn to both the philosophical tradition that interprets science as a process of detecting, collecting and correcting errors, as well as the statistical and econometric tradition that explicitly differentiates model selection and model criticism (section 5). Such an account makes it possible to address in concluding remarks the question of peculiarities of the art of model criticism in economic science, analogically to Rodrik’s insights into the art of model selection.

2. EXPLORING THE METAPHOR OF ECONOMICS AS A LIBRARY OF MODELS As it has been already pointed out, Dani Rodrik in his broadly discussed and thought-provoking book made a diagnosis of the state of the art in contemporary economic science. According to him, “models are both economics’ strength and its Achilles’ heel; they are also what make economics a science” (Rodrik, 2015,

Krzysztof Nowak-Posadzy

p. 5). Furthermore, as Mary Morgan convincingly put it, since the discipline of economics has always been considered both an art and a science (Morgan, 2012, p. 400), it may then prove useful to apply the latter distinction to the practice of modeling and to recognize that economic modelers “depend upon both tacit and articulated knowledge in making sense of their (…) findings and judging their relevance” (ibid., p. 34). When we look at, for example, judging the relevance of economic models, the success of an economist in model-building and model appraisal is assessed against intradisciplinary methodological and epistemic tacit norms, as designed by insiders to insiders and mutually negotiated by them. These norms are commonly shared to a sufficient degree by the members of the economics profession. However, it is the articulated knowledge about formal methods and explicit criteria that was traditionally of primary interest to economic methodologists and philosophers of economics. Given both Michael Polanyi’s famous dictum that for all knowledge that is intersubjectively communicable and controllable there exists some tacit and unarticulated knowledge (Polanyi, 1966) and taking into account growing economists’ and economic methodologists’ interest in the nature and function of tacit knowledge in economic decision-making (Perraton, Tarrant, 2007), it is puzzling that only a few economists and philosophers of economics focused their attention on the unspecifiable or hard-to-specify art of economic modeling. It is even more striking, when one considers that the tacit knowledge, skills and connoisseurship may play in economics the function of “a flexible methodological glue for doing that science” (Morgan, 2012, p. 399). Having said that, one has to take up the challenge of addressing an initial question about how to investigate in a philosophical and methodological way (in contrast to sociological, psychological and ethnographical ways) the tacit dimension of research practice (e.g. modeling) in economics. The philosophical and methodological literature on economic modeling provides us with at least three possible ways of addressing the question about how to investigate the economic research practice that to a large extent is based on tacit knowledge: 11

according to the first, “this can be done by studying the documents one will find at the sites of practice. These documents can include a variety of printed materials: almanacs, dictionaries, guides, handbooks, instructions, reports, teaching materials, tutorials, yearbooks and anything else one can find on desks and shelves at the research site or in the corners to where they have been thrown away out of frustration or because they became redundant. For a philosophy of science-in-practice, these documents have proven to be very informative sources to gain a deeper understanding of specific research practices, particularly of those where theories do play a minor role (…) and where knowledge is to a large extent intuitive” (Boumans, 2016, p. 30);

according to the second, this can be done by running a controlled field trial experiment that aims at assessing different ways of doing economics and econometrics. “The basic idea of the experiment is very simple, namely to

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

take a specified data set and let several researchers carry out the same set of applied econometrics tasks but with different methods, approaches and beliefs” (Magnus, Morgan, 1997a, p. 460). In order to learn about how tacit knowledge “fills the gap between methodological treatises and successful applications” (p. 462), “we asked all participants to join us in an attempt to throw light on this important aspect of econometric research. As a mechanism for keeping track of the process we suggested the ‘logbook’. It is commonplace in other scientific fields to keep lab notebooks. These record the procedures used, the various steps taken as the research progresses, false avenues, interim results, and other details the scientist wishes to keep track of. These records, directly or indirectly, can reveal much about the research process. We asked all our participants to keep such notebooks, which we call ‘logbooks’ and which form an important element of the experiment” (Magnus, Morgan, 1997b, p. 471); 11

according to the third way, this can be done by exploring the genesis, structure and function of metaphors that the practitioners of scientific modeling constructed and utilized in their attempts to conceptually grasp a certain, but not yet fully articulated in the formal language, component of research practice.4 The traditional philosophical view of metaphors in science admitted them a certain heuristic role but only at the early stage of formulation of unclear intuitions in the research process (Zeidler, 2014, p. 40). This situation changed when the interaction view of metaphor (Black, 1962), as well as the cognitive view of metaphor (Lakoff, Johnson, 1980) were elaborated. As a consequence, the role of metaphor in scientific practice started to be conceived as an important cognitive factor on various stages of the process of production of scientific knowledge.5 An example of an account that makes use of a certain metaphor is the one offered by economist Dani Rodrik (2015). Although he does not explicitly refer to the very term of “metaphor” while discussing the state of affairs in economic science, he does implicitly make use of the metaphor of economics as a library of models (Rodrik, 2015, pp. 5, 46, 84). 4 As

there is no direct access in philosophical/methodological investigations to the tacit dimension of economic research practice the exemplification of the tacit dimension depends on the understanding of the “tacit”. This term can be defined as either (1) “unarticulated” or (2) “unaware”. Concerning (1), the philosopher/methodologist is able to reconstruct the tacit dimension by focusing on either (1a) informal (verbal) procedures or (1b) suggestive (verbal) hints. Concerning (2), the philosopher/methodologist is able to reconstruct either (2a) accepting (declarations) or respecting by the researcher the basic set of normative beliefs. These beliefs constitute the so-called methodological consciousness. The structure of researchers’ methodological consciousness consists of five major components: epistemological (cognitive norms), methodological (directives), meta-scientific (aim of scientific cognition), metaphysical (world-view), and anthropological (vision of Man). The methodological consciousness can be seen as a dimension in which the culture intersects with scientific practice. 5 For the role of metaphors in economics and interactions between models and metaphors in economic research practice see: (Hardt, 2016).

100

Krzysztof Nowak-Posadzy

The Rodrik’s attempt to metaphorically grasp economic science as a library of models will now be shortly discussed. Such a discussion is necessary to underline the role the metaphor of library plays in drawing the economics profession’s attention to the neglected art of model selection. The genesis of the metaphor of library will be only signaled.6 Instead, the main focus will be on the structure and, especially, the function of this metaphor. To present the structure of the metaphor, it is helpful to recall below some relevant quotations from Rodrik’s book (2015): 11

“Rather than a single, specific model, economics encompasses a collection of models. The discipline advances by expanding its library of models and by improving the mapping between these models and the real world. The diversity of models in economics is the necessary counterpart to the flexibility of the social world. Different social settings require different models. Economists are unlikely ever to uncover universal, general purpose.” (Rodrik, 2015, p. 5; emphasis added);

„models enable the accumulation of knowledge, by expanding the set of plausible explanations for, and our understanding of, a variety of social phenomena. In this way, economic science advances as a library would expand: by adding to its collection (ibid., p. 46; emphasis added);

“[e]conomics advances by expanding the collection of potentially applicable models, with newer ones capturing aspects of social reality that were overlooked or neglected by earlier ones.” (ibid., p. 183);

“[e]conomics advances also by better method of model selection [and] this is more craft then a science” (ibid., p.183–184);

„[economic] methods are as much craft as they are science. Good judgment and experience are indispensable, and training can get you only so far.” (ibid., p. 83; emphasis added);

“economics as a collection of models, along with a system of navigation among models” (ibid., p. 208; emphasis added).

The structure of Rodrik’s reasoning by means of the library metaphor is as follows: he starts with acknowledging the analogy between a science and a library. Just as a library is a collection of books, economic science is viewed as a collection of models, i.e. the basic units of scientific knowledge production, communication and utilization. The library of economic models is thus growing 6 Before the structure and function of this metaphor, as constructed and applied by Rodrik to analyze in an original manner the state of affairs in economic science, will be explored, it may prove useful to indicate the genesis of the library metaphor. The metaphor in questions has its long history, especially in philosophical essays and novels. For example, it appears in Jorge Luis Borges’ The Library of Babel, Umberto Eco’s The Name of The Rose or Stanisław Lem’s Solaris.

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

101

through a continuous model-building process, that is, via the production activities of economic modelers. It is then the economic model-building practice that contributes to the expansion of the library of models. But what is necessary to the accumulation of knowledge in economic science is the method of model selection. The latter are necessary because economists in their daily practice, as Rodrik argue, deal with a variety of models at the same time. So the informal procedure of model selection comes into play in order to decide which model to apply. The informal nature of the way economists often proceed speaks for the view according to which model selection is more a craft then science. Moreover, Rodrik is fully aware of that the accumulation of knowledge in economics, if really occurred, concerns not only articulated and explicit knowledge, but also tacit and craft-based. The latter is of major importance especially in making models useful. The idea Rodrik advocates for draws support from both the history of economic thought and methodology of statistics. In the first case, it was John M. Keynes who pointed out that “economics is a science of thinking in terms of models joined to the art of choosing models which are relevant to the contemporary world” (Keynes, 1938). In the latter case, “[t]he current statistical methodology is mostly model-based, without any specific rules for model selection or validating a specified model” (Rao, 2004, p. 2). It is quite clear that by making use of the heuristic potential of the metaphor of economics as a library of models, Rodrik attempts to redirect the economics profession’s attention from formal methods of model-building to the neglected informal mode of model appraisal, namely, the art of model selection and craft of navigating among multiple models.

3. A RECONSTRUCTION OF THE ART OF MODEL SELECTION IN ECONOMICS Rodrik’s initial diagnosis and description of the very art of model selection in the practice of economic modeling attracted attention of philosophy of economics’ profession. His remark that economists in their crafting the navigation among models act often “informally and suggestively” (Rodrik, 2015, p. 184) has motivated Till Grüne-Yanoff and Caterina Marchionni (2018) to take an attempt to reconstruct the art of model selection more “formally and conclusively” (Rodrik, 2015, p. 184), or, to put it in authors terms, to reconstruct such an art as a “mechanical procedure” (Grüne-Yanoff, Marchionni, 2018, p. 4) containing built-in, at least, potential “decidability” criterion (pp. 1, 2, 5). In what follows we will shortly discuss this methodological contribution. However, due to both the detailed character of this contribution and size limitations of this article, it is hardly possible to deal with all of the interesting and thought-provoking analyses offered there. As the aim of this paper is to investigate the modes of model

102

Krzysztof Nowak-Posadzy

appraisal in economics, we will focus only on first step in Rodrik’s remarks on the art of model selection, as made by modeling practitioner’: 11 the informal procedure of selecting the set of candidate models from those epistemically available (Rodrik, 2015, p. 93); 11 the informal procedure of selecting the set of critical candidate models from the set of candidate models (ibid., p. 94–98); 11 the informal empirical tests via four verification strategies (ibid., pp. 93–94). As we focus on the first stage, we will take into considerations only the problem of so-called inclusion criteria for the choice of the set of candidate models, as Grüne-Yanoff, Marchionni (2018, p. 6) call it. Let us now proceed to a short discussion of Grüne-Yanoff’s and Marchionni’s account. In their conceptual framework they integrate three items: concept of model pluralism7, ModRep8 and set-theoretic toolbox9; this framework enables them the reconstruction of informal procedure of model selection, informal because of both its form – Rodrik’s verbal and underspecified description and its genesis – Rodrik’s unsystematic and unmethodical reflection on modeling experience (Rodrik, 2018, p. 3). From the ModRep Grüne-Yanoff and Marchionni pay attention to two components: model’s target (T) and model’s purpose (P). This way they can specify the very concept of model pluralism as the following: “The core idea is in the form of model pluralism, according to which multiple models of the same target T are acceptable as long as one model of T 7 By making use of the concept of model pluralism they directly refer to the Rodrik’s metaphor of economics as a library of models. However, indirectly they refer to the distinction made by Uskali Mäki (1997, pp. 37–39) between “plurality of Xs and pluralism about X”. For example, a statement that there is a plurality of economic models is a descriptive one, whereas any statement about economic model pluralism is of normative character. The normative character of the statement about pluralism takes the form of either justification or prescription. So it is in the case of the idea of model pluralism – it either justifies an existing plurality of economic models or in case of its absence (or insufficient degree) prescribes plurality of economic models by appealing to some ontological or epistemological reasons, e.g. progress in economics by adding to collection of models and improving methods of model selection. 8 ModRep is here the shortcut for the philosophical accounts on (economic) modelling. Actually, it is worth to mentioned two proposal: first was elaborated by Ronald Giere, according to which “[s]cientists use models to represent aspects of the world for specific purposes” (Giere, 2004, p. 742). Later on, this account was extended by Uskali Mäki, according to which “[ModRep] Agent A uses (imagined) object M as a representative of (actual or possible) target R for purpose P, addressing audience E, at least potentially prompting genuine issues of resemblance between M and R to arise, describing M and drawing inferences about M and R in terms of one or more model descriptions D, and applies commentary C to identify and coordinate the other components” (Mäki, 2013, p. 91). 9 What is, however, interesting in their general approach is that they, on the one hand, conduct their formal reconstruction in set-theoretic terms and, on the other hand, admit they have no clear attitude towards the semantic view of model. There seems to be an analogy between their “mechanical procedure” with built-in “decidability” criterion and the semantic view, according to which the intended scope of application is built-in in a model. Whether this analogy, if right at all, exposes some consistency difficulties, is left open.

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

103

is more useful for one purpose P, and another model of T is more useful for another purpose P' ” (Grüne-Yanoff, Marchionni, 2018, p. 1). Such a formulation enables them to draw attention to the neglected function of confronting a model with empirical evidence, namely “selection of the appropriate model for a specific target T for a particular purpose P” (Grüne-Yanoff, Marchionni, 2018, p. 1; cf. Rodrik, 2015, p. 90). As we have already mentioned, they re-described the informal procedure of selecting the set of candidate models from those epistemically available in terms of the inclusion criteria. Such an decision is supported by references to Rodrik’s contextual usage of terms such as “intuitive”, “intuitively”, “intuition”, “reasonable”, “reasonably”, “simplicity”, “simplified”, “simplistic”, “plausible”, “story lines”, “narratives”. Rodrik’s, formally underspecified, usage of these terms is identified by Grüne-Yanoff and Marchionni as the first gap in his description. To fill this gap they enumerate five basic terms as an inclusion criteria (Grüne-Yanoff, Marchionni, 2018, p. 4) and next they explicate, what is quite puzzling, only four (p. 6). We will come back to this puzzling omission later on. Below the four specified and separate criteria for selecting the set of candidate economic models will be presented (p. 6): 11

“intuitiveness: T is a member of a theoretical basis T, which is intuitive;

reasonableness: D is a member of the set of reasonable derivation rules D;

plausibility: A1, …, Ak, …, An are members of the set of plausible assumption A;

narrative relevance: A1, …, Ak, …, An ∉ R (R is a representation of model user’s purpose), i.e. inferences are not trivial.”

The criterion omitted in their reconstruction, yet repeatedly appearing in Rodrik’s informal description, is the criterion of simplicity. Below we will make an attempt to correct this omission by both recalling the standard formulation of the logical simplicity criterion as referred to scientific theory, as well as by offering a tentative non-standard formulation of such an criterion that utilizes Grüne-Yanoff’s and Marchionni’s solutions. The standard formulation of the criterion of simplicity in science was offered, for example, by the methodologist of empirical sciences Jan Such (1982). Such distinguished two types of simplicity – mathematical and logical. Between these two types there is a negative correlation. The logical simplicity of a theory can be presented in the following form (Such, 1982, p. 119):

logical simplicity =

informational content number of initial assumptions

(1)

where an informational content of a theory is “determined by the set of its implied theorems (being logical consequences of its postulates)” and initial assumptions “are mutually independent”.

104

Krzysztof Nowak-Posadzy

Let us briefly comment on this formula: the more logical consequences a theory contains, the logically simpler it is; the less initial assumption a theory has, the logically simpler it is. Now, the question arises whether we can extend such an account on logical simplicity to the case of the inclusion criteria for candidate model selection. A careful exploration of the four criteria specified by Grüne-Yanoff and Marchionni brings a tentative solution, according to which the omitted criterion of simplicity may be expressed as follows (own elaboration):

simplicity =

narrative relevance . plausibility

(2)

Such a formulation may explain the omission of the criterion of simplicity in the discussed reconstruction due to derivative character of this criterion in GrüneYanoff and Marchionni proposal. However, this problem surely deserves a deepened and separate examination. Let us close this section with recalling Rodrik’s remark that “[e]conomics advances also by better methods of model selection” (Rodrik, 2015, p. 183). Exposing that economic model selection is “more a craft then a science” (ibid., p. 184) means that economists’ navigation among models proceed sometimes “informally and suggestively”. As we have pointed out, Grüne-Yanoff and Marchionni in their formal reconstruction focused on the first element of this expression, namely on “informally” (Rodrik, 2015, pp. 89, 90). For further investigation it is helpful to systematize the interpretation of the term “informally” and understand it as: 11 either “verbally”: –– “economists employ a wide a range of strategies (…) from the informal and anecdotal to the sophisticated and quantitative” (ibid., p. 109, emphasis added); –– “practitioners’ views about the real world develop much more heuristically, as a by-product of informal conversations and socialization among themselves” (ibid., p. 171, emphasis added). 11 or “enthymematically”: –– “the idea that economists have to carry multiple models in their heads simultaneously” (ibid., p. 84, emphasis added); –– “the core of model-based science is the representational capacity of the model, which is deployed to serve the modeler’s purpose with the norms of the discipline in mind.” (Spiegler, 2015, p. 25, emphasis added). As Grüne-Yanoff and Marchionni left aside the second element of the expression, namely that economists in their research proceed often “suggestively” and did not make any systematic reference to it, we will take an attempt to clarify this element. Our view is that “suggestively” may relate to at least two, formally underspecified and philosophically underdeveloped, peripheral issues in economic modeling:

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

105

presupposing theoretical background for a given economic model that enables and secures the tractability of model and contributes to its soundness. Such a presupposed economic-theoretical background has something in common with intuitiveness criterion, as specified by Grüne-Yanoff and Marchionni; 11 implicating scope of application of a given economic model that secures against misapplication of such a model and indicates its intended application. Such an implicated applicability has something in common with establishing a model’s purpose that aims at delimiting the intended scope of model’s application. We will come back to this issue later on. 11

Foregoing considerations were intended only to clarify the idea that model appraisal in economics involves not only formal methods and informative tests but also informal modes and suggestive hints.

4. EXPLORING THE METAPHOR OF ECONOMICS AS A MARKETPLACE OF MODELS The expression of economics-as-a-library-of-models, as coined by Rodrik to draw economics profession’s attention on the art of model selection and craft of navigating among economic models that has long been underspecified at the very heart of economic science, is a metaphor, not the metaphor. Therefore, the art of model selection, as informally expressed by Rodrik and formally reconstructed by Grüne-Yanoff and Marchioni, does not exhausts the very art of model appraisal in economic science. There are good philosophical, statistical and economic-theoretical reasons to distinguish the second implicit mode of model appraisal, namely the art of model criticism. The discussion of the latter will depart from constructing and exploring the metaphor of economics-as-a-marketplace-of-models. Let us, however, start our exploration with discussing a more general metaphor, namely that of science as a marketplace of ideas. As the philosopher of economics Jesus Zamora Bonilla correctly pointed out, “the thesis that ‘science is like a market’ has often been taken as an assumption about the working of some ‘epistemic invisible hand’ mechanism behind the process of scientific research. This vision of science as a ‘marketplace for ideas’ was not originally a technical notion in the analysis of scientific research, but rather a common metaphor ‘floating in the air’” (Zamora Bonilla, 2012, p. 15). This primary metaphor, once carefully extended, may prove useful to heuristically grasp economics science as a marketplace of models. Analogically to section 2, the genesis of the metaphor of marketplace of ideas will be only signaled. Instead, the main focus will be on the structure and, especially, function of this metaphor as related to economic science. Let us start the discussion with the genesis of the metaphor in question. The genesis can be traced to the works of philosopher, logician and political economist John Stuart Mill, especially, to his classic paper titled On liberty (1859). According to Mill, an open society simply benefits from freedom of expression

106

Krzysztof Nowak-Posadzy

and speech contributing to the creation, presence and pertinence of a wide range of alternative ideas, let it be scientific ones. It was philosopher of science Karl Popper who explicitly applied this way of thinking to science when he stated that “the advance of science depends upon the free competition of thought, and thus upon freedom, and that it must come to an end if freedom is destroyed” (Popper, 1959[2005], p. 279). Such a view was further developed by another philosopher of science, Paul Feyerabend who claimed to be following Mill’s tradition, at least, in elaborating his pluralistic view of science. According to Feyerabend, there are two main rules that guides scientific research: the principle of tenacity that claims that scientists stick to their theoretical strategies as long as possible, despite initial problems they encounter, believing in potential of their theories; and the principle of proliferation that claims that scientists produce and supply theoretical alternatives, the more the better for scientific progress, and the number of which should be maximized (Feyerabend, 1970; Bschir, 2015). In a situation when the proliferation effect dominates over the tenacity effect, proliferation of scientific products contributes to the marketplace of ideas. What is, however, missing in Feyerabend’s account of science as marketplace of ideas is the lack of some possible principle for “rejection or elimination of ideas” (Godfrey-Smith, 2003, p. 116). It was William W. Bartley who developed the metaphor of marketplace of ideas one step further. He did not argu that the primary function of the market is to reject or eliminate some ideas. Instead, he focused on the function of recognizing errors. As he convincingly pointed out, “the university is a marketplace of ideas where new ideas are welcome, and falsehoods can be challenged without recrimination (…). Markets are particularly useful in directing attention to error. The detection of error is the dismal function of the marketplace” (Bartley, 1990, p. xvi). A metaphor that underlines the detection, and not rejection, of error in science accommodates some recent accounts of research practice in economics. On the one hand, according to Rodrik, “[m]odels rarely get rejected in economics” (Rodrik, 2018, p. 2). On the other hand, as Lawrence Boland stressed, “[t]esting by attempting to falsify someone’s theory or explanation is just one of many types of criticism. And it is criticism or more specifically, a critical attitude that is the hallmark of science” (Boland, 2003, p. 235). As we have already said, Rodrik opted for the way of thinking in economics profession about model appraisal that he labeled as informal art of “model selection”. Doing this, however, he explicitly dismissed another, prima facie traditional and declaratively supported by most economists, way of thinking about model appraisal, according to which economics profession’s “party line holds that economics advances by improving existing models and testing hypotheses. Models are continually refined until the true universal model comes into view. Hypotheses that fail the test are discarded; those that pass are retained” (Rodrik, 2015, p. 84; emphasis added). What Rodrik meant by “the party line” in economic science is the falsificationist ideal widely attributed do the works of the philosopher of science Karl Popper (Rodrik, 2018, p. 1–2).

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

107

According to Rodrik, this program of methodology, when applied to economic research, does not accurately describe the way economic modelers practice their trade i.e. it does not accurately describe the way economic knowledge is accumulated (e.g. learning process), as well as the way scientific progress in economics occurs. Behind Rodrik’s expression there is the old question of whether Popper’s falsificationism is a descriptive or a normative program, question recently raised by various philosophers of economics. Following a broad discussion, there is a consensus among philosophers that falsificationism in Popper’s version is not a descriptive stance, so neither the history of science, sociology of science nor economics’ insiders’ reports can invalidate it. Still, if it is a normative position, there remains the problem of whether Popper’s falsificationism passes the test of self-reflexivity (whether falsificationism is falsifiable?) (see: e.g. Nowak, 1992). As the detailed discussion about rights and wrongs (or pros and cons) of falsificationist program of economic methodology is not the purpose of this paper, it is sufficient to close this section with a remark that Popper’s approach has still some untapped potential for the philosophy of economics. As Wade Hands put it, a “context dependent criticism is a prime desideratum” (Hands, 2004, p. 152).

5. UNDERLABOURING FOR THE RECONSTRUCTION OF THE ART OF MODEL CRITICISM IN ECONOMICS The metaphor of economics-as-a-marketplace-of-models, as discussed in the previous section, constitutes a heuristically useful basis to draw inspiration in our preliminary exploration of the art of model criticism in economic science from three theoretical traditions: 11

the philosophical tradition that interprets science as a process of detecting, collecting and correcting errors by way of systematic criticism (e.g. Popper, 1959[2005]; 1972[1994]; Bartley, 1982; Mayo, 1996; Boumans, Hon, Petersen, 2014);

the statistical tradition that explicitly differentiates two phases of statistical modeling, namely model selection and model criticism, as well as elaborates respectively to these two phases different statistical tests and formal procedures (e.g. Box, 1976, 1980; Mayo, 1996; O’Hagan, 200110; Rao, 2004; Staley, 2012);

the economic-methodological tradition that investigates the way economists make the potential errors during economic modeling more transparent (e.g. Morgenstern, 1963; Mayo, Spanos, 2004, 2010; Reiss, 2008; Louçã, 2007). 10 “Model

criticism may be defined briefly as the process of checking the model assumption (…) without reference to alternative models or assumptions. It is intended as an open-minded phase of investigation to identify any problems with the model. Formulation of explicit alternatives comes after the model criticism phase has identified some problems.” (O’Hagan, 2001, p. 1)

108

Krzysztof Nowak-Posadzy

Let us start our analysis of the art of model criticism in economic science by discussing the approach within the philosophy of science that interprets science as a process of detecting, collecting and eliminating errors, mistakes and flaws by means of the systematic criticism (or to put it differently, by exercising critical attitude and making use of critical tests). Popper’s idea that science is entirely based on the method of “trial-and-error” (or “conjecture and refutation”), occupies a special place within this tradition: “The way in which knowledge progresses, and especially our scientific knowledge, is by unjustified (and unjustifiable) anticipations, by guesses, by tentative solutions to our problems, by conjectures. These conjectures are controlled by criticism; that is, by attempted refutations, which include severely critical tests. They may survive these tests; but they can never be positively justified: they can neither be established as certainly true nor even as ‘probable’ (in the sense of the probability calculus). Criticism of our conjectures is of decisive importance: by bringing out our mistakes it makes us understand the difficulties of the problem which we are trying to solve. This is how we become better acquainted with our problem, and able to propose more mature solutions: the very refutation of a theory – that is, of any serious tentative solution to our problem (…) is how we can learn from our mistakes.” (Popper, 1963, p. vii). Such a view has already been and, actually, still is widely discussed. As summarizing this discussion goes beyond the aim and scope of this paper, it is sufficient to state that the problem with interpreting, tailoring for special sciences and applying Popper’s “trial-and-error” (or “conjecture and refutation”) method is that it can be either easily understandable or a complex issue. In either case, however, there are still some controversies. In what follows, it will be argued that Popper’s “trial-and-error” idea can be easily understandable (but it does not mean that widely shared) when it is referred to the question of growth of scientific knowledge11. Popper formulated a schema of the growth of scientific knowledge through error-elimination by way of systematic rational criticism. According to him, this general tetriatic schema is found “more and more useful as a description of the growth of theories”, and, what is important, this “tetriatic schema can be elaborated in various way” (Popper, 1972, p. 287). Below, a number of variants of this schema will be gradually presented: Diagram 1. PS

Source: Popper, 1972, p. 243. 11 The question of growth of scientific knowledge, as stated and answered by Popper, has its equivalent in Rodrik proposal, namely that there occurs an accumulation of knowledge in economic science due to the process of model-building and model selection.

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

109

where: PS – problem-situation the scientist is facing, TS – tentative solution to the problem-situation, EE – error-elimination that controls tentative solutions, PS EE PS TS – diachronic relation. Still, this sequence is not intended by Popper to be a cycle, therefore: PS (on the left) ! PS (on the right): Diagram 2. PS1

PS2

Source: Popper, 1972, p. 243.

where: PS1 – old problem-situation, PS2 – new problem-situation. Still, there can be a multiplicity of trials for error-elimination focused on only one tentative solution: Diagram 3. EEa PS1

EEb

PS2

EEn Source: own elaboration.

where: <EEa, EEb, …, EEn> – a finite and non-empty set of trials for error-eliminations that jointly contribute to the emergence of one new problem-situation. Still, there can be a multiplicity of trials for error-elimination that independently contribute to emergence of many new problem-situations: Diagram 4. PS1

EEa

PS2a

EEb

PS2b

EEn

PS2n

Source: own elaboration.

where: <EEa, EEb, …, EEn> – a finite and non-empty set of trials for error-eliminations which independently contribute to the emergence of a number of separate new problem-situations; <PS2a, PS2b, …, PS2n> – a finite and non-empty set of new problem-situations.

110

Krzysztof Nowak-Posadzy

Still, there can be a multiplicity of tentative solutions offered and considered by scientist, therefore TS is not a unit set: Diagram 5. TSa PS1

TSb

PS2

TSn Source: Popper, 1972, p. 243.

where: <TSa, TSb, …, TSn> – a finite and non-empty set of tentative solutions. Still, there can be a multiplicity of trials for error-eliminations taken by a given scientist or other scientists: Diagram 6. PS1

TSa

EEa

TSb

EEb

TSn

EEn

PS2

Source: own elaboration.

where: <EEa, EEb, …, EEn> – a finite and non-empty set of trials for error-eliminations that jointly contribute to the emergence of one new problem-situation. Still, there can be a multiplicity of new problem-situations that emerge as a consequence of different and separate trials for error-elimination: Diagram 7. PS1

TSa

EEa

PS2a

TSb

EEb

PS2b

TSn

EEn

PS2n

Source: Popper, 1972, p. 287.

As it has been already pointed out, Popper’s “trial-and-error” (or “conjecture and refutation”) method can be easily understandable or a complex issue. The simple enumeration of various schemas of growth (in a diachronic sense) of scientific knowledge can be seen as a relatively easily understandable task. But even this Popper’s proposal and achievement has been widely discussed and often put into question (cf. Worrall, 1995). As it is not the aim of this paper, we will not attempt to answer the question whether these various schemas accurately describe the growth of knowledge in economics. However, Rodrik’s usage of the

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

111

library metaphor indicates he seemingly opted for the approach, according to which the accumulation of knowledge in economics occurs by adding to the collections of models. In any case, the problem of growth of knowledge certainly deserves a separate treatment. However, what is more important here for us is to consider the usefulness of the general method of trial-and-error by way of systematic criticism for the reconstruction of the art of model criticism in economic science. Interpreting and applying (in a synchronic sense) Popper’s ideas to the question of scientific (here: economic) modeling is surely a complex issue. By labeling such a task a complex one, one is expected to skillfully tailor Popper’s general method of trial-and-error to the practice of economic modeling. To meet this expectation the reconstruction of the art of economic modeling in general and model criticism in economics in particular is needed. Such a reconstruction will proceed in one preliminary and three main consecutive steps (respectively: supplementation, specification and refinement). The preliminary step is to establish the base-scheme for further reconstructive operations. The idea is to choose one schema from the available set (see: diagrams 1–7) previously presented. The schema that fits best to our investigations is diagram 6. Let us now move to the first main step. As we want to tailor and apply this schema to a special science such as economics, we have to supplement this framework with an account of economic modeling. To supplement means to complete a given component of the scheme with a certain context, within which its specific disciplinary grammar emerges. To be specific, it is the initial problem-situation (PS1) that is going to be specified. What is advocated here is that the problem-situation (PS1) in economics, which is a model-based science, can be understood as follows: “an economists E uses model M to establish a certain likeness with the target T for intended purpose P and the success of E in accomplishing P is judged against disciplinary norms N” (cf. Spiegler, 2015, p. 25; cf. Mäki, 2013). As we want to focus on the model appraisal phase of economic modeling, we have to specify (clarify) how actually theoretical economic models are appraised. On the ground of recent developments in the philosophy of economics and economic methodology, the dominating view is that theoretical economic models “are not appraised by ex post empirical testing. Such models are assessed by whether they satisfy their purpose” (Boumans, 2005, p. 15).12 This is in line with the supplementary Spiegler’s account on economic modeling, but to demonstrate how such a view corresponds to Popper’s traditional account on method of criticism in scientific practice, according to which tentative theories (solutions) are 12 “There are several reasons why economists and others appraise theories and models. They may want to judge whether theories and models are worthy of study, whether one can rely on them in research and practice, or whether one can judge them to be true or false or predictively adequate. Different purposes may call for different appraisals.” (Eels, Hausman, 2008, p. 248).

112

Krzysztof Nowak-Posadzy

tested empirically, is quite challenging. To disentangle this puzzle let us recall the argument made by another philosopher of economics, Francesco Guala (2005). As Guala, during his investigation into the peculiarities of theoretical modeling and experimenting in economics, pointed out (Guala, 2005, pp. 219–220), tentative theoretical models13 cannot be tested empirically in the same way as Popperian tentative solutions (theories, hypotheses) can be. In the trial of validation of theoretical models an element that was absent in the traditional discussions about theory testing, namely fitting the modeler’s purpose, comes into play. Establishing a purpose by the economic modeler aims at delimiting the intended scope of model’s application. It is so, because theoretical models, in contrast to what the semantic view of theories offers, do not have any built-in indicators or instructions where and how to apply it. The latter question is delegated to the intended purpose established by economic modeler. While discussing the art of model criticism it is also important to note that inferring the domain of application from theoretical models by trained economists is to large extent rather a matter of informal economist’ craft than formal procedures. Let us come back to the question of appraisal and empirical testing of economic models. It is a given application of the model, as indicated by modeler in the purpose, that can be empirically tested and not the theoretical model per se: “Scientists are pragmatic people, and although some paradigmatic applications are considered more important than others, a model is always useful to a degree, as long as it is applicable to some situation (or, more precisely, as long as it is more helpful in understanding a certain situation than are other rival models). The fact that a model turns out not to work under certain circumstances does not count as a refutation of the model but only as a (failed) test of its applicability in a given domain.” (Guala, 2005, p. 220). What Boumans and Guala have advocated for does not automatically mean that in our underlabouring for the reconstruction of the art of model criticism in economics we should abandon the Popperian concept of error-detection and error-elimination. What is needed here is the refinement of this concept so that it becomes more tailor-made for the contemporary research practice in economics. This is how we arrived at the third step, which actually, consists of two stages. During the first one, in order to deal with the current practice of economic modeling, we have to take a synchronic approach. To do so, a new element has to be introduced into the base-scheme (6), namely the feedback between TS and EE. By adding the feedback relation, we can, at least tentatively, consider the synchronic view. The refinement we postulate here will proceed by utilizing an 13 For

the interesting account on the tentative epistemic strategies as adopted by researchers in applied sciences in the process of constructing models (see: Carrier, 2004).

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

113

account on scientific modeling in general and the model verification in particular as elaborated by formal methodologist of empirical sciences, Jan Żytkow. According to Żytkow, a scientific modeling is a multilevel feedback process (Żytkow, 1995, 1999). As discussing his entire philosophical account on scientific modelling in details is not necessary here14, the main focus will be on the model appraisal which Żytkow calls “model verification”. It is important to know that, for Żytkow, “[m]odeling can be effective only if verification accompanies each cycle in model construction, providing feedback long before the final solution is reached” (Żytkow 1995, pp. 177–179). This idea is in line with Boumans’ claims that in a number of contemporary practices of theoretical modeling in economics the model appraisal (assessment, validation) is intertwined with model-building (Boumans, 2005, p. 3). By interpreting scientific modeling as multi-layer task, Żytkow is able to conceptually integrate the idea of negative or positive feedback into his model of scientific modeling. As he summarized his own account on the model appraisal: “The most efficient evaluation occurs at the levels prior to the final verification, according to an AI principle: ‘evaluate partial solutions as early as possible’.” (Żytkow, 1999, p. 323). As it has been already pointed out, by integrating the feedback relation into the base-scheme (diagram 6) we can, at least tentatively, receive the synchronic view concerning the method of trial-and-error: Diagram 8. PS1

TSa

EEa

TSb

EEb

TSn

EEn

PS2

Source: own elaboration.

An attempt to incorporate the feedback relation to the base-scheme of Popper’s method of trial-and-error is not something that is illegitimate as having no support in Popper’s writings. To support this claim let us refer to Popper’s statements15: “Criticism (…) is always an attempt to find (and to eliminate) a mistake, a flaw, or an error within the theory. It is (…) the negative feedback by which we control the construction of our theories” (Popper, 1994, p. 162; emphasis added). 14 “The

process of model construction meanders through many feedback loops, as solutions to problems specific at a given step may require changes made to the constructions at the earlier steps” (Żytkow, 1999, p. 322). 15 By focusing on the base-scheme of Popper’s method of trial-and-error it is the tentative solution that is a certain variable which represents a variety of scientific outcomes (theories, hypothesis, models).

114

Krzysztof Nowak-Posadzy

This is how we arrive at the second stage of our attempt to refine Popper’s concept of error-correction (elimination), so that it better fits the contemporary practice of economic modeling. On the one hand, what was essential to scientific investigations, according to him, is the criticism practiced by scientists. Exercising scientific criticism aims at two error-detection (identification) and error-elimination (correction). But still, Popper taking the normative approach to methodology, did not examined in details how actually scientists practice their criticism or to put it in differently, how they craft “the art of recognizing and avoiding errors”, as Paul Feyerabend (1970, p. 18) called it. According to the latter, “scientist who works in a particular historical situation must learn how to recognize error and how to live with it, always keeping in mind that he himself is liable to add fresh error at any stage of the investigation. He needs of a theory of error in addition to the ‘certain and infallible’ rules (…) A theory of errors will therefore contain rules of thumb, useful hints and these suggestions to historical episodes so that one sees in detail how some of them have led some people to success in some situation (…) Good books on the art of recognizing and avoiding errors will have much in common with good books on the art of singing or boxing” (ibid., p. 18–19). On the other hand, Popper often differentiated between “critical or severe tests” and “critical attitude or susceptibility to criticism”, both being features of the scientific approach. However, Popper in his canonical view is not entirely clear about how to understand both the relation between “critical tests” and “critical attitude” and the specific function that “critical attitude” may play in scientific research. This state of affairs gives us some room for an interpretation. Our starting point is that these two concepts, “severe tests” and “critical attitude”, were, and still are, subjected to further philosophical processing quite independently. It is because these concepts refer to two distinct dimensions of scientific practice: “severe test” relates to the explicit dimension consisting in formal methods and explicit or articulated criteria of model-building and model appraisal, whereas “critical attitude” relates to the tacit dimension that consists in informal strategies and implicit modes of model-building and model appraisal. Various conceptual difficulties arise from the fact that in daily practice of scientific (here: economic) modeling these two dimensions inevitably overlap, as Mary Morgan convincingly stated (Morgan, 2012, p. 25). Regarding the concept of “critical or severe tests”, besides of numerous methodological specifications being already offered in the broad literature, it was creatively developed by philosopher of science and statistician, Deborah Mayo. She is a supporter of the so-called error-statistical philosophy of science, a tradition that seemingly prioritizes analyses of “critical or severe tests” over “critical attitude”. However, as it will be presented later on, it leaves some room to assume that critical attitude may play some function in scientific research. What is more important at that moment is that Mayo treats her “account of testing as implementing and improving upon the Popperian idea

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

115

of critical rationalism by cashing out the notion of a severe test” (Mayo, Spanos, 2010, p. 114). Certainly, a severity is the formal requirement of a test of a given theory, model or hypothesis. According to Mayo, in the search for methods of criticism that would be more accurate to contemporary practice of scientific modeling: “we need to distinguish what critical rationalism has been since Popper – Popperian critical rationalism – and a forward-looking theory of criticism, which we may call progressive critical rationalism (…) Progressive critical rationalism would proceed by developing tools for severe tests. Such tools seek reliable probes of errors.” (Mayo, Spanos, 2010, p. 117). The concept of error occurring in science occupies a prominent position in Mayo’s project of progressive critical rationalism aiming at refinement of efficient methods of criticism. Mayo supports Popper’s initial idea that “all our knowledge grows only through the correcting of our mistakes”, but, nevertheless, she argues that his approach is lacking an explicit theory of error.16 The latter is expected to conceptualize the genesis, types and functions of errors, or, to put it in a different way, “conditions of error, the various kinds of error and variety of effects errors can have” (Boumans, Hon, Petersen, 2014, p. 19). Mayo’s main focus can be thus expressed by five basic questions: why is it useful to study the art of dealing with errors in research practice? What are the major stages of the process of dealing with error? What are the types of errors? What does it mean that scientists can learn from errors? And, finally, how does this type of learning proceed in science?17:

16 As John Worrall strongly put it: “What exactly constitutes scientifically valuable criticism, for example? Does producing the most valuable criticism involve holding all theories equally open to correction? How exactly is ‘error’ established in science? What exactly do we learn from our mistakes (‘truer’ theories or only ones that have higher empirical adequacy)? Can some theories, although always strictly speaking tentative, nonetheless become probable to a reasonably high degree? Are successive ‘trials’ informed by the successes and failures of previous ones? And, if so, exactly how? (…) The two criticism of Popper’s own attempt to fill out the details of the general scheme that I shall discuss here. The first is that he basically mischaracterized the process of ‘error-elimination’ in science. And the second is that he basically mischaracterized the process by which ‘tentative theories’ are proposed. Put baldly: Popper’s view was that science is entirely based on the method of ‘trial-and-error’, ‘conjecture and refutation’, and yet – so these criticisms allege – he seriously misidentified the nature both of the process of identifying error in science and of the process of theory-production or ‘conjecture’.” (Worrall, 1995, pp. 76–77). 17 “The view that we learn from error, while commonplace, has been little explored in philosophy of science. When philosophers of science do speak of learning from error-most notably in the work of Popper-they generally mean simply that when a hypothesis is put to the test of experiment and fails, we reject it and attempt to replace it with another. Little is said about what the different types of errors are, what specifically is learned when an error is recognized, how we locate precisely what is at fault, how our ability to detect and correct errors grows, and how this growth is related to the growth of scientific knowledge” (Mayo, 1996, p. xii).

116

Krzysztof Nowak-Posadzy

“The history of mistakes made in a type of inquiry gives rise to a list of mistakes that researchers either work to avoid (before trial planning) or check if committed (after-trial checking). For example, when inferring the cause of an observed correlation, such a repertoire of errors might include a set of questions: Is the correlation spurious? Is it due to an extraneous factor? Are we confusing cause and effect? Corresponding to such a repertoire of errors is a “reservoir of models.” I call them models of error.” (Mayo, 2004, p. 321). As we defined science as a process of trial-and-error, not only the traditional stages of error-detection and error-elimination have to be specified, but also a new intermediary stage has to be introduced. According to philosopher of science and epistemologist Douglas Allchin, one of the most significant, yet insufficiently addressed, issues concerning the scientific research, is to catalog or collect actual and potential errors (Allchin, 1999). In practice, researchers usually assemble a discipline-specific informal catalog of past mistakes, flaws or errors. Thereby, they arrange and keep track of an error repertoire (Mayo, 1996, pp. 5, 18). It is therefore a challenging task for philosophers of science to analyze different types of errors in different disciplines and recognize some patterns across disciplines, as well as for methodologists of the special sciences (here: economics) to draw more on a latent discipline-specific reservoir of errors and reconstruct the way scientists actually cope with errors by building and using formal models of error and informal potential error scenarios (cf. Staley, 2014, p. 40). Distinguishing three stages of dealing with errors in scientific research (detection, collection, elimination) does not, however, give us a complete enough picture of the process of trial-and-error. What is needed is a fourth stage, namely learning from error. It will be argued that the learning from error is to some extent a matter of “critical attitude”, that is, rather of an informal and implicit kind. It may prove useful here to recall and briefly address three questions: what are the types of errors?18 what does it mean that scientists can learn from errors? and how does the learning from error occurs in scientific research? As types of error are to a large extent discipline-specific, the focus will be on the special science of economics. As it has been already mentioned, there are some economic-methodological works that investigate the way economists make errors occurring in their daily modeling practice more transparent. An example of such an account is given by Francisco Louçã (2007). Louçã in his contribution analyzed the history of the concept of error in economic theory, models and equations, as well as changing economics profession’s epistemic strategies to 18 Mayo

introduced four standard or canonical types of error in statistical modeling and experimental sciences: (i) mistaking chance effects or spurious correlations for genuine correlations or regularities; (ii) mistakes about the quantity or value of a parameter; (iii) mistakes about a causal factor; (iv) mistakes about experimental assumptions (Mayo, 1996, p. 18).

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

117

handle them. He distinguished seven types of error as characteristic to economic science and research (ibid., p. 214), namely: 11 measurement errors; 11 influence of omitted variables; 11 intrinsic randomness in human agency; 11 theoretical misspecification of the model; 11 functional misspecification; 11 general inadequacy of the model; 11 irregularities (‘aberrations’). As long as there are different types of error, what is required are different, i.e., discipline-specific strategies for detecting, collecting and eliminating them (cf. Mayo, 2014, p. 59). However, in order to successfully avoid errors or, at least, to formulate potential error scenarios a scientist has to be open to learn from errors. This is how we arrive at the questions of what it means that scientists can learn from errors and how the learning from error occurs in scientific research? A possible answer, advocated in this paper yet surely not exclusive, is that, the learning from error, being a purpose of criticism, is “often done relatively informally” (Cox, Mayo, 2010, p. 285) or, to put it differently, it follows “an informal pattern of reasoning” (Mayo, 2000, p. 322). Once we consider criticism not only as targeted at detection, collection and elimination of errors, but also as aimed primarily at informal and implicit learning from error, we can cash out Popper’s notion of critical attitude. This argument has support in Popper’s writings: “[W]hat characterizes the scientific approach is a highly critical attitude towards our theories rather than a formal criterion of refutability: only in the light of such a critical attitude and the corresponding critical methodological approach do ‘refutable’ theories retain their refutability.” (Popper, 1968, p. 94). Still, there remains the question of so-called learning effect of criticism as it actually takes place in daily practice of scientific research. Leaving this question open, there are good reasons to regard criticism in scientific research as having rather an oscillatory character. As William Bartley insightfully pointed out, “[c]lumsily applied eradication of error may also eradicate fertility. Criticism must be optimum rather than maximum, and must be deftly applied.” (Bartley, 1982, p. 133). So, the main purpose of model criticism in economics that is targeted, at least, at making transparent (exposing) the latent errors, is to secure and not to reduce the possibility of learning from modeling failures, blind alleys and obstacles by the widest possible audience on the marketplace of economic models. As long as “[s]cientists are rarely fully explicit about or even aware of why their methods and strategies work when they do” (Mayo, 2014, p. 77), the key is to reconstruct both how economic models are insured against errors and how economic modelers practice their art of recognizing and avoiding errors.

118

Krzysztof Nowak-Posadzy

6. CONCLUDING REMARKS: MODEL SELECTION AND MODEL CRITICISM IN ECONOMICS – COMPLEMENTS OR SUBSTITUTES? The aim of this article was to juxtapose and consider two theses about economic modeling. According to the first one, practices of theoretical economic modeling, namely model-building and model-appraising, involve both formal procedures and informal craft. Concerning formal procedures, philosophy of economics takes into account formal methods and techniques of economic model-building and explicit criteria of economic model-appraisal. When skilled craft is allowed into consideration, and that was of primary interest in this paper, informal styles (genres, strategies) of economic model-building and implicit modes (ways of thinking, attitudes) of economic model appraisal come to the fore in philosophical investigations. According to the second thesis, the practice of economic model appraisal involves, apart from formal and explicit criteria, implicit and informal modes of model appraisal. These are the joyful, yet long forgotten in economics, art of skillful navigation among models (as recently recalled and discussed e.g. in: Rodrik, 2015, 2018) and the dismal, yet potentially useful for economics, art of agile or optimal learning from errors occurring in modeling (as discussed e.g. in: Mayo, Spanos, 2010). The art of smart navigating among economic models has already triggered a number of in-depth philosophical commentaries, as well as has received so far a few formal reconstructions by philosophers of economics. As a results of the intellectual exchange between economists and philosophers of economics, the formal procedure of model selection in economics was reconstructed and exposed for further discussion (see: section 3). Certainly, answering the question of how economic modelers actually choose among models is an urgent and practical issue. As Uskali Mäki stated, “[b]oth Keynes and Rodrik think that economists have difficulties with the art of model selection, which gives rise to major wrongs in economic modeling” (Mäki, 2018, p. 17). Still, an equally burning question is the question of how economic modelers actually revise (refine or re-specify) existing models. This problem was not discussed by Rodrik in his book to a satisfactory degree though. What is more, it looks like Rodrik’s rather negative outlook on the practices of improving existing models and testing hypotheses was rooted in his belief that the methodological “party line” does not adequately describe what is really going on in economics (see: section 4). Nevertheless, once we think (in a synchronic not diachronic way) about the question of how economic modelers actually refine or re-specify their model, then the art of skillful learning from modeling and from errors, as well as the mode of model criticism comes to our mind and may become a useful conceptual scheme for further philosophical investigations into the science and art of economic modeling.

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

119

As the underlabouring for the reconstruction of such an art of model criticism in economic science has already been carried out (see: section 5), what is needed next is a more detailed and based on a case study formal reconstruction of the way economists practice the model criticism by, at least, making the potential errors during their modeling more transparent and neutralized. Although, it was not intended in this paper to offer such a reconstruction, it is possible to briefly discuss the main characteristics and stages of such an account: 11

“an economist E uses model M to establish a certain likeness with the target T for intended purposes P and the success of E in accomplishing P is judged against disciplinary norms N” (see: Spiegler, 2015);

an economist E by using model M learns from modeling and the object and effect of learning depends on the disciplinary-specific norms (cf. Grüne-Yanoff, 2009);

the disciplinary-specific norms are conceptualized as the norms of model criticism or as the norms of model selection that, if only strictly followed, organize the process of navigating among models (see: Rodrik, 2015) or the process of learning from mistakes, flaws and errors occurring in modeling (cf. Mayo, Spanos, 2010);

giving priority to the norms of model criticism, an economist detects, collects, eliminates errors and learns from them by running the model specification (S), mis-specification (M-S) and re-specification (R-S) tests and analyses (see: Spanos, 2010; Staley, 2014);

detection (at least, making transparent), collection (at least, informal), elimination (at least, neutralization) of errors and learning from errors involve both articulated knowledge and craft-based skills.

Following these enumerated characteristics and stages may prove useful to guide the reconstruction of the art of model criticism in economic science. Recent literature on modeling in the field of methodology of statistics and econometrics (cf. Box, 1976, 1980; Mayo, 1996; O’Hagan, 2001; Rao, 2004; Mayo, Spanos, 2004; Staley, 2011) seems to provide motivation to distinguish two phases of economic modeling, namely model selection and model criticism. Moreover, in focusing on the very art of model criticism we can utilize some arguments delivered by recent works of economic methodologists who analyze the way economists deal with errors in the practice of modeling (e.g. Mayo, Spanos, 2004; 2010; Reiss, 2008; Louçã, 2007). There seem to be enough reasons to philosophically investigate the dismal art of model criticism by means of systematic detection, collection, elimination of errors and learning from them, as exercised by economists. However, a separate treatment and case study would be needed to address in details the questions of how economists actually refine and re-specify their

120

Krzysztof Nowak-Posadzy

models and how they actually learn from errors. That is why the respective reconstruction should be left for another occasion. What may be, however, addressed here is the question of whether the art of model selection, as discussed by Rodrik (2015, 2018) and Grüne-Yanoff, Marchioni (2018), and the art of model criticism are substitutes or complements. The answer depends on whether we juxtapose and analyze these two modes in a diachronic or a synchronic way: 11

from a diachronic point of view, when we deal with the problem of the growth of knowledge in economics as a model-based science, model selection (in Rodrik’s formulation) and model criticism (in a traditional Popperian formulation as expressed in base-schema – diagram 6 from the section 5) seem to be more like substitutes. This interpretation is supported by Rodrik, according to whom “economic science advances by expanding its library of models” and “models enable the accumulation of knowledge.” (Rodrik, 2015, p. 46) On the other hand, in the Popperian view, according to which the growth of knowledge is not a cumulative process, science (here: economics) advances through the process or error-elimination;

from a synchronic point of view, when we deal with the problem of stages (phases) in the cycle of modelling in economic science, model selection (in Rodrik’s formulation) and model criticism (in a refined version as expressed in the scheme (8) from the section 5) seem to be more like complements. This interpretation is supported, on the one hand, by Żytkow (1999), who conceptually integrated the idea of positive and negative feedback into his framework of model evaluation. On the other hand, by Mayo and Spanos, according to whom the first stage of model criticism, namely “the problem of (…) model specification is a distinct from model selection, insofar as (…) the presence of misspecifications jeopardizes the ground for model selection” (Mayo, Spanos, 2004, p. 1008).

This article was an attempt to distinguish two informal modes of model appraisal in economic modeling. Disciplinary-specific norms of model appraisal may be conceptualized as norms of model selection or norms of model criticism, against which a tentative theoretical solution that economic modeler arrived at is judged. If an economic modeler (or a community of modelers) is more interested in adding to the library of models, then what is prioritized is the practice and upgrading of formal procedures or craft of model selection. If an economic modeler (or a community of modelers) is more interested in contributing to the marketplace of models, then improving formal severe tests or art of model criticism is prioritized. To sum up, as Uskali Mäki has inconclusively, yet thought-provokingly, stated recently, “[t]ime will tell whether Dani Rodrik’s Economics Rules (perhaps in its next editions) will win a place in the series of honoured treatises on the nature

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

121

of economics written by practicing economists, next to works such as J.E. Cairnes's The Character and Logical Method of Political Economy (1875/88), Lionel Robbins's The Nature and Significance of Economic Science (1932/35) and Milton Friedman's The methodology of positive economics (1953)” (Mäki, 2018, p. 1). A growing number of interesting commentaries, as well as systematizing formal reconstructions triggered by Rodrik’s art of navigation among economic models by means of following the model selection procedure, suggest the answer to Maki’s statement could be positive. Still, there is a part of philosophy of economics that in order not to be exposed to the charge of decrying how economic modeling should be carried out, looks forward to a treaty on the dismal, yet constructive, art of model criticism by means of learning from errors, written by some practicing economist.

REFERENCES Allchin D. (1999), Negative Results as Positive Knowledge, “Marine Ecology Progress Series”, 191, pp. 303–305. Backhouse R. (ed.) (1994), New Directions in Economic Methodology, Routledge, London and New York. Bartley W.W. (1982), The Philosophy of Karl Popper: Part III: Rationality, Criticism, and Logic, “Philosophia”, Vol. 11, No. 1–2, pp. 121–221. Bartley W.W. (1990), Unfathomed Knowledge, Unmeasured Wealth: On Universities and the World of Nations, Open Court Publishing Co, U.S. Black M. (1962), Models and Metaphors, Cornell University Press, Ithaca, New York. Blaug M. (1992), The Methodology of Economics or How Economists Explain, 2nd Edition, Cambridge University Press. Boland L. (2003), Foundations of Economic Method: A Popperian Perspective, Rout ledge, London. Boland L. (2014), Model Building in Economics: Its Purposes and Limitations, Cambridge University Press. Boumans M. (2005), How Economists Model the World into Numbers, Routledge, London and New York. Boumans M., Leonelli S. (2013), Introduction: On the Philosophy of Science in Practice, “Journal of General Philosophy of Science”, Vol. 44, Issue 2, pp. 259–261. Boumans M., Hon G., Petersen A.C. (eds.) (2014), Error and Uncertainty in Scientific Practice, Pickering and Chatto Limited, London. Boumans M. (2016), Models of Clutter, “Studia Metodologiczne”, No 36, pp. 27–41. Box G.E.P. (1976), Science and Statistics, “Journal of the American Statistical Asso ciation”, 71(356), pp. 791–799. Box G.E.P. (1980), Sampling and Bayes’ Inference in Scientific Modeling and Robustness, “Journal of the Royal Statistical Society”, Series A, 143(Pt. 4), pp. 383–430. Bridgman P.W. (1959), The Way Things Are, Harvard University Press. Bschir K. (2015), Feyerabend and Popper on Theory Proliferation and Anomaly Import: On the Compatibility of Theoretical Pluralism and Critical Rationalism, “Hopos:

122

Krzysztof Nowak-Posadzy

The Journal of the International Society for the History of Philosophy of Science”, 5(1), pp. 24–55. Carrier M. (2004), Knowledge Gain and Practical Use: Models in Pure and Applied Sciences, in: D. Gillies (ed.), Laws and Models in Science, King’s College Publications, London. Cartwright N. (1999), The Dappled World: A Study of the Boundaries of Science, Cambridge University Press. Collins H. (2010), Tacit and Explicit Knowledge, The University of Chicago Press. Cox D., Mayo D.G. (2010), Objectivity and Conditionality in Frequentist Inference, in: Mayo D.G., Spanos A. (eds.), Error and Inference. Recent Exchanges on Experimental Reasoning, Reliability, and the Objectivity and Rationality of Science, Cambridge University Press. Dow S., Pluralism in Economics, in: Teaching Pluralism in Economics, J Groenewegen (ed.), Edward Elgar Publishing, Cheltenham, UK and Northampton, MA. Dow S. (2012), Foundations for New Economic Thinking. A Collection of Essays, Palgrave Macmillan. de Marchi N. (ed.) (1992), Post-Popperian Methodology of Economics: Recovering Practice, Kluwer, Dordrecht. Eels E., Hausman D.M. (2008), Theory Appraisal, in: S. Durlauf, L. Blume (eds.), The New Palgrave Disctionary of Economics, Second Edition, Palgrave Macmillan, pp. 248–253. Erkenntnis (2009), Issue on: Economic Models as Credible Worlds or as Isolating Tools? Vol. 70, No. 1. Feyerabend P. (1970), Against Method. Outline of an Anarchistic Theory of Knowledge, in: M. Radner, S. Winokur (eds.), Analyses of Theories and Methods of Physics and Psychology, University of Minnesota Press. Gelfert A. (2016), How to Do Science with Models: A Philosophical Primer, Springer International Publishing. Giere R. (2004), How Models Are Used to Represent Reality, “Philosophy of Science”, Vol. 71, No. 5, pp. 742–752. Gilboa I. et al. (2018), Economics: Between Prediction and Criticism, “International Economic Review”, Vol. 59, No 2, pp. 367–390. Godfrey-Smith P. (2003), Theory and Reality: An Introduction to the Philosophy of Science, University of Chicago Press. Gräbner C., Strunk B. (2018), Pluralism in Economics: its Critiques and their Lessons, ICAE Working Paper Series, No. 82, https://www.jku.at/fileadmin/gruppen/108/ ICAE_Working_Papers/wp82.pdf Goldfarb R., Ratner J. (2008), “Theory” and “Models”: Terminology Through the Looking Glass, “Econ Journal Watch”, Vol. 5, No 1, pp. 91–108. Grüne-Yanoff T. (2009), Learning from Minimal Economic Models, “Erkenntnis”, Vol. 70, No. 1, pp. 81–99. Grüne-Yanoff T., Marchionni C. (2018), Modeling Model Selection in Model Pluralism, pp. 1–11, DOI: 10.1080/1350178X.2018.1488572. Guala F. (2005), The Methodology of Experimental Economics, Cambridge University Press.

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

123

Hacking I. (1983), Representing and Intervening: Introductory Topics in the Philosophy of Natural Science, Cambridge University Press. Hands W. (2004), Two Proposals for the Recovery of Economic Practice, in: R. Backhouse (ed.), New Directions in Economic Methodology, Routledge, London and New York. Hardt Ł. (2016), On Similarities in Modelling and Metaphorizing Economic Phenomena, “Studia Metodologiczne”, No 36, pp. 147–174. Keynes J.M. (1938), Keynes’ letter to Roy Harrod on 4 July 1938, http://economia.unipv. it/harrod/edition/editionstuff/rfh.346.htm. Klein D., Romero P. (2007), Model Building versus Theorizing: The Paucity of Theory in the Journal of Economic Theory, “Econ Journal Watch”, Vol. 4, No 2, pp. 241–271. Lakoff G., Johnson M. (1980), Metaphors We Live By, Chicago, Chicago University Press. Lawson T. (1997), On Criticizing the Practices of Economists: a Case for an Interven tionist Methodology, in: A. Salanti, E. Screpanti (eds.), Pluralism in Economics: New Perspectives in History and Methodology, Edward Elgar, Cheltenham, pp. 13–36. Louçã F. (2007), The Years of High Econometrics. A Short History of the Generation that Reinvented Economics, Routledge, London and New York. Maas H. (2014), Economic Methodology: A Historical Introduction, Routledge, London. Magnani L., Bertolotti T. (eds.) (2017), Springer Handbook of Model-based Science, Springer International Publishing. Magnani L., Casadio C. (eds.) (2016), Model-Based Reasoning in Science and Technology Logical, Epistemological, and Cognitive Issues, Springer International Publishing. Magnus J.R., Morgan M. (1999), Methodology and Tacit Knowledge: Two Experiments in Econometrics, Wiley. Mayo D.G. (1996), Error and Growth of Experimental Knowledge, The University of Chicago Press. Mayo D.G. (2000), Models of Error and the Limits of Experimental Testing, in: M. Carrier, G. Massey, L. Ruetsche (eds.), Science at Century’s End: Philosophical Questions on the Progress and Limits of Science, University of Pittsburgh Press. Mayo D.G., Spanos A. (2004), Methodology in Practice: Statistical Misspecification Testing, “Philosophy of Science”, 71, pp. 1007–1025. Mayo D.G., Spanos A. (eds.) (2010), Error and Inference. Recent Exchanges on Experimental Reasoning, Reliability, and the Objectivity and Rationality of Science, Cambridge University Press. Mayo D.G. (2014), Learning from Error: How Experiment Gets a Life (of its Own), in: Boumans M., Hon G., Petersen A.C. (eds.) (2014), Error and Uncertainty in Scientific Practice, Pickering and Chatto Limited, London. Mäki U. (1994), Methodology Might Matter, but Weintraub’s Meta-Methodology Shouldn’t, “Journal of Economic Methodology”, Vol. 1, No 2, pp. 215–231.

124

Krzysztof Nowak-Posadzy

Mäki U. (1997), The One World and the Many Theories, in: Pluralism in Economics, in: A. Salanti, E. Screpanti (eds.), Pluralism in Economics: New Perspectives in History and Methodology, Edward Elgar, Cheltenham, pp. 37–47. Mäki U. (2008), Method and appraisal in economics, 1976-2006, “Journal of Economic Methodology”, Vol. 15, Issue 4, pp. 409–423. Mäki U. (2013), Contested Modeling: the Case of Economics, in: U. Gähde et al. (eds.), Models, Simulations, and the Reduction of Complexity, De Gruyter, Berlin/Boston. Mäki U. (2018), Rights and Wrongs of Economic Modelling: Refining Rodrik, “Journal of Economic Methodology”, DOI: 10.1080/1350178X.2018.1488475, pp. 1–19. Morgan M. (2012), The World in the Model: How Economists Work and Think, Cam bridge University Press. Morgenstern O. (1963), On the Accuracy of Economic Observations, Princeton University Press, New Jersey. Nowak L., (1992), The Idealizational Approach to Science: A Survey, in: J. Brzeziński, L. Nowak (eds.) Idealization III: Approximation and Truth, Vol 25, Rodopi, Amsterdam-Atlanta, pp. 9–63. O’Hagan A. (2001), HSSS Model Criticism, ftp://ftp.uni-bayreuth.de/pub/math/statlib/ DOS/general/first-bayes/pdf/modcrit.PDF Perraton J., Tarrant I. (2007), What Does Tacit Knowledge Actually Explain?, “Journal of Economic Methodology”, Vol. 14, Issue 3, pp. 353–370. Polanyi M. (1966), The Tacit Dimension, The University of Chicago Press. Popper K.R. (1959[2005]), Logic of Scientific Discovery, Routledge, London and New York. Popper K.R. (1963), Conjectures and Refutations. The Growth of Scientific Knowledge, Basic Books, New York and London. Popper K.R. (1968), Remarks on the Problems of Demarcation and of Rationality, in: I. Lakatos & A. Musgrave (eds), Problems in the Philosophy of Science, North-Holland, Amsterdam. Popper K.R. (1972[1994]), Objective Knowledge: An Evolutionary Approach, Oxford University Press. Popper K.R. (1994), The Myth of Framework. In Defence of Science and rationality, Routledge, London and New York. Rao C.R. (2004), Statistics: Reflections on the Past and Visions for the Future, “Communications in Statistics – Theory and Methods”, Vol. 30, Issue 11, pp. 2235–2257. Reiss J. (2008), Error in Economics. Towards o More Evidence-Based Methodology, Routledge, London and New York. Rodrik D. (2015), Economics Rules. Why Economics Works, When it Fails, and How to Tell the Difference, Oxford University Press. Rodrik D. (2018), Second Thoughts on Economics Rules, “Journal of Economic Methodology”, pp. 1–6, DOI: 10.1080/1350178X.2018.1490441. Ryle G. (1949[2002]), The Concept of Mind, The University of Chicago Press. Sent E.-M. (2006), Pluralisms in Economics, in: Scientific Pluralism, S.H. Kellert, H.E. Longino, C.K. Waters (eds.), University of Minnesota Press, Minneapolis, pp. 80–101.

ON TWO MODES OF APPRAISAL OF ECONOMIC MODELS

125

Spiegler P. (2015), Behind the Model: A Constructive Critique of Economic Modeling, Cambridge University Press. Staley K.W. (2012), Strategies for Securing Evidence Through Model Criticism, “Euro pean Journal for Philosophy of Science”, 2, pp. 21–43. Staley K.W. (2014), Experimental Knowledge in the Face of Theoretical Error, in: Boumans M., Hon G., Petersen A.C. (eds) (2014), Error and Uncertainty in Scientific Practice, Pickering and Chatto Limited, London. Such J. (1982), Are there Definitively Falsifying Procedures in Science? In: W. Krajewski (ed.), Polish Essays in the Philosophy of the Natural Sciences, D. Reidel Publishing Company, Dordrecht. Suppe F. (1989), The Semantic Conception of Theories and Scientific Realism, University of Illinois Press, Urbana. Weintraub R. (1989), Methodology Doesn’t Matter, but the History of Thought Might, “Scandinavian Journal of Economics”, Vol. 91, No 2, pp. 477–493. Worrall J. (1995), Revolution in Permanence’: Popper on Theory-Change in Science, “Royal Institute of Philosophy Supplement”, 39, pp. 75–102. Zamora Bonilla, J.P. (2012), The Economics of Scientific Knowledge, In: Handbook of the Philosophy of Science, Philosophy of Economics, U. Mäki (ed.), North Holland: San Diego, pp. 823–862. Zeidler P. (1997), Od modelu semantycznego do modelu teoretycznego w metodologii nauk empirycznych, in: Od logiki do estetyki, R. Kubicki, P. Zeidler (eds.), Wydawnictwo Fundacji Humaniora, pp. 53–68. Zeidler P. (2014), O roli ugruntowanych empirycznie metafor w naukowych programach badawczych, „Filozofia i Nauka”, t. 2, pp. 239–255. Żytkow J.M. (1995), Scientific Modeling: Round Trips to Many Destinations, in: AAAI-95 Spring Symposium on Systematic Methods of Scientific Discovery, pp. 175–180. Żytkow J.M. (1999), Scientific Modeling: Multilevel Feedback Process, In: Model-Based Reasoning in Scientific Discovery, L. Magnani, N. Nersessian, P. Thagard (eds.), Kluwer, pp. 311–325.

O DWÓCH TRYBACH OCENY MODELI EKONOMICZNYCH STRESZCZENIE Ekonomia stała się nauką opartą na metodzie modelowej. Modelowanie ekonomiczne, obejmujące zarówno konstruowanie modeli, jak i ich ocenę, wymaga znajomości formalnych procedur oraz nieformalnego rzemiosła. W związku z tym cel niniejszego artykułu jest podwójny. Po pierwsze, wpisanie się w dyskusję dotyczącą praktyki modelowania ekonomicznego przez zasygnalizowanie funkcjonowania dwóch nieformalnych trybów oceny modeli. Będzie to oznaczać

126

Krzysztof Nowak-Posadzy

omówienie propozycji Dani'ego Rodrika dotyczącej nieformalnej sztuki selekcji modeli, która wywołała pogłębione komentarze filozoficzne i stała się obiektem formalnych rekonstrukcji dokonanych przez metodologów ekonomii. Po drugie, uzupełnienie tego ujęcia poprzez przygotowanie gruntu pod rekonstrukcję odrębnego trybu oceny modeli wyrażającego się w nieformalnej sztuce krytyki modeli. Słowa kluczowe: filozofia ekonomii, model ekonomiczny, ocena modeli, selekcja modeli, krytyka modeli. Klasyfikacja JEL: B41, A14, B40

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

POLEMIKA

Marcin Gorazda*

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT?

ABSTRACT The text is supposed to be a critical response to Łukasz Hardt’s paper on ceteris normalibus laws. Author especially criticises three main Hardt’s theses: 1. Economic laws do not describe regularities, but refer to capacities and powers; 2. Economic laws are only true in theoretical models; 3. Economic laws are valid ceteris normalibus, rather than ceteris paribus. Based on several examples of theoretical models in economics Author argues that: 1. We cannot abandon the requirement of regularities being the necessary component of any scientific law, economics including. The concept of capacities, even if helpful in reasoning on causes and outcomes, is methodologically redundant; 2. Economic laws cannot be true only in theoretical models. They must be (at least within the range assumed by the researcher) true in the domain represented by the particular model. Otherwise, the notion of “laws true only in a model” refers to the inherent tautologies, which truth value are checked exclusively by assumptions and adopted inference rules; 3. The term ceteris normalibus in Hardt’s account is redundant because it simply represents a more general set of assumptions, including ceteris paribus, ceteris rectis, ceteris absentibus, ceteris constantibus. As long as the “normal” circumstances are defined in a model, the clause does not improve our understanding of models. Keywords: economic models, economic laws, regularities, capacities, ceteris paribus, ceteris normalibus. *

Jagiellonian University, Copernicus Center for Interdisciplinary Studies; e-mail: marcin. gorazda@gsw.com.pl

128

Marcin Gorazda

This paper was inspired by Łukasz Hardt’s text, published in the same special edition of “Studia Ekonomiczne”, and is meant to be a critical response to at least some of the crucial Hardt’s theses, namely: 1. Economic laws do not describe regularities, but refer to capacities and powers; 2. Economic laws are only true in theoretical models; 3. Economic laws are valid ceteris normalibus, rather than ceteris paribus. I must admit at the beginning that I strongly appreciate the depth of Hardt’s insight into the problem of the ontology of economic laws, and I respect his stance as a scientific realist (though mine is different). Nevertheless, after reading his text, I have an impression that either there is a deep misunderstanding about very fundamental terms and concepts, or his theses, as indicated above and explained in the paper, are indefensible, even in the light of scientific realism. It would probably be easier to critically review his text from instrumentalist’s or constructivist’s point of view, but it would not contribute much to the current philosophical debate, as the arguments on both sides are pretty well known. Besides, it seems that the contemporary realism, especially presented by Uskali Mäki, who takes into account the purpose of the theory, its language and its possible perception by the audience, does not fall much away from the moderate instrumentalist’s view. My point here is that those three theses, as they are put forward and argued by Hardt are not defensible regardless of the ontological stance we occupy, and at least in reference to the first thesis, it seems to contradict other statements found in the paper.

ECONOMIC LAWS, REGULARITIES AND CAPACITIES Intuitively, the fact that economic laws (assuming they exist, as Hardt also questions their existence (Hardt, 2017)) do not describe regularities, or even that they have nothing in common with regularities sounds odd. Each scientist who investigates the realm of her / his studies with empirical methods knows very well that to formulate a law or a law-like statement requires the demonstration of the certain repeatable connection between the investigated factors or variables. Econometricians know the most about it. But even the classical and neoclassical economists, who constructed their theories on the deemed rational homo economicus, based on regularities, which were repeatable behavioral patterns of a rational agent on the market. They were falsely assumed, but they were. So could it be that Hardt detaches the regularities from the concept of economic law? Reading the paper, one cannot conclude anything different. We read among others: Here comes second understanding of economic laws, precisely they do not describe regularities, but they refer to capacities and powers.

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 129

(…) we should not be surprised that our attempts at understanding cp-laws in economics in terms of some regularities usually fail. What we should focus on is the real domain of economic reality where such entities as the following ones are present, namely powers, mechanisms, tendencies, and structures. He even goes further and claims that we may have a (valid?) economic law, based on tendencies, where the tendency is dormant, and we do not observe even one occurrence of the expected outcome. As a given tendency may be dormant and thus it is not to produce a particular (or anticipated) result, in a similar vein we may have type-causation between A and B without any manifestations of tokens a and b. The above quotations do not leave any room for doubts. According to Hardt, in economic laws regularities are of no importance, are redundant. This strong claim is contradictory to the accounts of philosophers, including realists referred to in the text. It also contradicts this part of the text where Hardt tries to explain how we know about capacities. Let us start from the argumentum ab auctoritate. This argument is so commonly mocked as commonly used, and commented paper has a lot of quotations from undisputable authorities in the domain. The concept of capacities comes from British philosopher of science, Nancy Cartwright. Though she claims that it originally goes back to Mill and even Aristotle, her account is the most comprehensive. Cartright’s idea is much like Hardt’s, but not exactly the same. She would probably never defended the view that in a scientific law, the regularities do not count, even in reference to economic laws. Her concept of a scientific law is based on the so-called “nomological machine” which is (…) fixed (enough) arrangements of components, or factors, with stable (enough) capacities that in the right sort of stable (enough) environment, give rise to the kind of regular behaviour that we represent in our scientific laws (Cartwright, 1999, p. 50). Regular behavior is a necessary feature of the nomological machine and a necessary feature of a scientific law. Nomological machines can be artificially constructed or can be observed in a particular environment in nature. Their main purpose is to let us change the environment, namely to intervene, to compose causes to produce the target effect. Without regularities, it would be impossible. What distinguishes Cartwright from other philosophers, especially those inclined on the one side towards scientific constructivism and on the other side towards scientific fundamentalism, is that she claims that “necessary regular association between properties” is not enough. We need to understand the arrangements of capacities that give rise to them, and those capacities are real. Moreover, laws are not universal. They work only in the special settings which are the nomological machines. Laws of nature hold only ceteris paribus – they hold only relative to the successful repeated operation of a nomological machine.

130

Marcin Gorazda

The example of economic laws, where the capacities are dormant and the expected outcome is not produced although the setting is just right, is an example of the nomological machine not running properly, what may imply that the laws are not observed, and the underlying capacities were misidentified. This statement seems to be pretty obvious for most economists. If the monetary mechanism does not work as it is supposed to work according to the commonly accepted laws, e.g. if lowering interest rates does not lead to higher investments, and it does not lead to them even in one singular occurrence, it does not mean that the tendency is “dormant” (whatever it means). In terms of Cartwright, it means that we are unable to construct the running nomological machine according to our previous blueprint, and it further implies that the assumptions for the machine blueprint might have been false, and the deemed capacities were misread or they changed. Hardt knows it too. But he seems to detach the ontological claims from the epistemological ones. In a different section of his paper he tries to answer the question, how we know about capacities. Out of our empirical experience, of course. It reads: First, we need special arrangements where capacities can show up. Second, measurement of their effects is necessary. Third, capacities can be deduced from probabilities, or, to say more precisely, probabilities can offer us hypothesis concerning capacities’ existence. In this passage the contradiction between the ontological and epistemological claim is clear. The capacities must show up, and even more, they must show themselves so that we can measure them. Dormant capacities are unobservable and even less measurable. If we deduce capacities from probabilities, then providing that we are talking about classical probability calculus and rules of inferences embedded therein, we clearly need regularities to deduce anything from within. Taking the above into account, it seems that eliminating the expected outcome occurrence and the regularities from the concept of economic law is impossible (unless we assume that no such thing exists like economic law, which is another Hardt’s disputable claim expressed elsewhere (Hardt, 2017)). However, instead of pondering over the ontological components of the concept of economic law-like regularities, we may focus exclusively on its correct expression or description in language. From this point of view, we may benevolently read Hardt’s thesis as referring to the mode of expression and not to the necessary components of economic law. In this understanding, very close to Cartwright’s account, laws are about capacities, not because they do not refer to any regularities, but because those regularities occur exclusively in nomological machines and beyond them, they are rarely met. Thus laws are not universal, and whenever we need to take our observed associations beyond the setting of a particular nomological machine, we need a concept of capacities. Instrumentalist would probably say that it is purely the problem of our language whether we express laws in terms of capacities or regularities, which occur with the different

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 131

ratio in different environments, and in certain environments, they do not occur at all. Both expressions are in any case replaceable. Other philosophers, however, would not agree, rightly indicating that to intervene in nature effectively, regularities are not enough. We need the concept of causation or/and mechanism. I would even assume that they might be right. But do we need capacities and are they ontologically valid? Our philosophical generalizations depend very often on the examples we deploy to illustrate them. Cartwright’s account, as well as Hardt’s, is very rich in examples. However while Cartwright uses examples of scientific theories or nomological machines which are well recognized and work good, like Newton’s laws of motion and the solar system, Hardt uses examples of economic laws which seem to fail (especially after the recent financial crisis), like monetary mechanism and deemed relation between the interests rate and investments. Therefore, Cartwright requires the repeatable functioning of the nomological machine to deduce on capacities, while Hardt does not. But both agree that capacities are true and valid. Let us, however, focus for a moment on different examples of models in economics, also connected with monetary flows and recent financial crisis. Economists (at least some of them) rightly concluded that if contemporary models were unable to spot the significant signs of the coming crisis, and even less to instruct the policymakers on how to struggle with it, we should work out better models. Some of them are quite peculiar and seems to be constructed on the entirely different ontological assumptions. In the paper from 2010, Billio et al. (2012, p. 535) proposed a model of measuring the so-called systemic risk. The concept of systemic risk in such models is itself an interesting issue. Instead of investigating the linear, causal relations between the quantifiable variables, systemic risk is embedded in the complex, numerous interconnections between market agents. Those interconnections evolve over time according to the evolutionary logic (i.e. mutation, imitation and adaptation to the dynamically changing environment being the most important factors), and they may, or may not enter the dangerous, irreversible path leading directly to the global financial crisis. The task is to spot the moment (or pattern) signifying the coming disaster. Billio at al. focused their research on the financial agents, namely hedge funds, banks, brokers and insurance companies and they reconstructed the map of their connections, taking into account their monthly returns. They claim to reveal certain causal relations among them, and that their model “can also identify and quantify financial crisis periods, and seem to contain predictive power for the current financial crisis.” Another interesting example, not based on the evolutionary logic, is the model constructed by Nyman et al. (Nyman, et al., 2018). Contrary to the mainstream economists, they decided to study the possible relation between the broadcasted news and certain macroeconomic indices (like consumer sentiment index, economic policy uncertainty index, PMI and others). They built their model on the social-psychological theory of “conviction narratives”, which broadly assumes that narrative drives human action. They investigated three

132

Marcin Gorazda

sources of news, i.e. Bank of England’s daily reports, Reuters, and brokers’ reports to measure the so-called “relative sentiment shift” (RSS). They claim to discover a strong correlation between the sentiment measures and financial crisis, where RSS falls in advance of the crisis and even in advance of the other macroeconomic variables, which seems to lag it rather than lead. It may imply the causal relation so that broadcasted information can be one of the essential causes of the financial crisis. What lessons may philosophers learn from those models? They are undeniably economic models. They are far from being the blueprints for well-functioning nomological machines. But at least they are built on certain regularities, and they reveal a certain level of predictive power. Are we, however, able to identify in those models the searched capacities or tendencies or the nature of things? In the case of agent-based models constructed according to the evolutionary principles, I would say that it is impossible. One cannot say that in the nature of interconnectedness between the financial institution is financial crisis creation, because it is clearly both not true and not informative (too vague). The only nature (actually assumed in the model) is its evolutionary dynamic. In the case of RSS, it could be stated, that it is in the nature of the financial market that narrative drives the agent’s behaviour. But instantly two questions emerge: First, “capacity” so identified does not seem to be compliant with a set of standard economic models, where the agent’s behaviour is driven mainly by economic incentives (like interest rates). If capacities may have truth values, which one is true? Second, do we really need the ontological concept of capacities in such models? Is it not enough to conclude that we (most probably) have discovered the additional, hitherto disregarded causal factor, which seems to act stronger than others? It is not my intention, at this stage to go deeper into this hard, metaphysical dispute. But I claim that the concept of capacities, so apparent in models based on linear equations, recedes in more complex models, where the pattern emerges out of interconnectedness of unquantifiable variables. The attempt at answering the question why there are patterns at all, or why they produce any regularities is one of the most troublesome in the scientific ontology. Whether those are capacities which are responsible for them, or the mathematical structure of the world1, or propensities2 is another story, and all these hypotheses are ontologically very strong and undoubtedly interesting. Methodologically they are however redundant. We do not need this strong ontology to make good science. But we cannot make any good theory or construct any good model without the observed regularities or repeatable patterns, economics including. Models presented above are supposed to illustrate the vagueness of the concept of capacities and its speculative nature. 1 The concept of mathemacity of the world is among others promoted by Michal Heller and is often presented as the justification of the effectiveness of inductive reasoning in science (See: Heller, 2006). 2 The concept of propensities is one of the interpretations of the probability theory given by C.H. Pierce and developed by K. Popper. More can be found in: Załuski, 2008.

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 133

TRUTH IN THEORETICAL MODELS Whenever Hardt writes about theoretical models in economics, I have an impression that under the term “model” we understand something different. According to some passages in his text, his view on models seems to be based on the typical realistic approach. Defining what model is he quotes Mäki: Agent A uses object M as a representative of some target system R for purpose P, addressing audience E, prompting genuine issues of resemblance to arise; and applies commentary C to identify and align these components (Mäki, 2009). In some sections, he seems to stress the importance of the above-underlined features of model’s representativeness and resemblance to the target system. We read: (…) the closer a given empirical domain to the model’s structure is, the higher probability that the model’s insights (i.e., economic laws) are to correctly explain the workings of such a domain. Nevertheless, isomorphism between models and empirical domains is never perfect and thus economic laws only describe tendencies in economic realm. (…) I am to show however that discussing laws without referring to models and empirical phenomena is simply impossible. So, one might have an impression that the genuine tester of the hypothetical truth value of the model is a relation, so far unspecified, between the model and the said empirical domain. However, reading his further text one may deduce that his models lack any reference to the outer world. He writes: Now, let me recapitulate the main findings of our study into the very meaning of ceteris paribus clause. It seems that the only uncontroversial way to successfully defend such laws is just to claim that a given cp statement is only true in a model used for its “production” or, in a second case, if one have a cp law, no matter of its origin, then one can always construct a model in which such a statement is to be true. (…) Having in mind what has been said above, one should agree that economic laws (usually stated with ceteris paribus clauses) can be understood as the laws always true in economic models, and hence ceteris normalibus is just to be conceptualized as being synonymous to “in a model” phrase. So, for instance, saying that ceteris paribus lower interest rates are to stimulate investments can be rephrased that lower interest rates always stimulate invest-

134

Marcin Gorazda

ments only in theoretical models where such a relation holds or, in other words, ceteris normalibus, lower interest rates stimulates investments. If one can always construct (a good?) model in which a ceteris paribus statement (any given cp statement?) is true, then one can “produce” any law. The example with interests rates always stimulating the investments only in theoretical models is especially worrying because if the stimulating effect works always only in a theoretical model, it does not work always in the target system. Consequently, the important causal connection is detached from the empirical domain. Either we have here another inconsistency in the text, or we have to assume that the postulated closeness or resemblance to the given empirical domain does not refer to economic laws. To cope with this claim, I need to make a few remarks on models. Regardless of the vast and unfinished discussion in philosophy what they exactly are, two features seem to be common in most accounts, Mäki’s including. First, models model something. There should always be a reference to a kind of universe which is supposed to be modelled. This universe is often called (as above) the target system (Mäki, 2009). In economics, this target system is the economy or its selected part. Models without the target system are either not models in this meaning or can be at most the interesting, entertaining game for mathematicians, and until they do not find its target system in the human economy, they are useless for economists. Special emphasis should be placed on the phrase ‘target system in the human economy’, because we may figure out the model for which the target system is the imaginary economy in Hobbiton, based on the special set of Hobbits’ preferences, who first and foremost wished to make other Hobbits happy by donating them gifts. The equilibrium is reached when for each Hobbit-agent the value of gifts given, equals the value of gifts received. Is it an economic model? We may say it is. Is there truth in this model? This question leads us to the second important feature of models. Models cannot be true or false. If we interpret the model mechanically (like Phillip’s hydraulic model of economy) we may say that the model accurately or inaccurately represents its target (Reiss, 2013). If we interpret a model as a set of sentences (expressed in the natural language or in mathematical equations, what is the case in most of the economic models), we may ascribe, at most, the truth value to the particular sentences. The components of such models may have different functions. In purely mathematical models we have constants and variables, sentences representing assumptions (axioms), inference rules (law-like sentences or specific causal connections including) and last but not least, the sentences representing the outcomes of applied inference (predictions). Mäki supplements it by the model’s purpose audience and commentaries. Each of those components may be the bearer of the truth value. But what is the correct method of ascribing the truth value to the sentences constructing the model? Usually, we test the correspondence level with the rel-

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 135

evant sentences describing the given target domain. But some sentences may also be tautologies deduced upon the given assumptions and accepted inference rules. In Hobbiton sentences expressing the assumptions and causal connections (law-like sentences) are fictitious, so their truth value is 0. They do not correspond to any known human economy. Therefore we may only check if our model is internally complete and consistent, i.e. to test if given assumptions and causal connections can provide an outcome for each value of a variable and if outcomes are consistent (non-contradictory). If the model is internally complete and consistent, we may say that the fictitious economic law is “true” in a model that means that it provides in each iteration non-contradictory outcomes. The sentences in such model which use the fictitious economic law as the inference rule are tautologies. This is the only model of which I can say that “economic law is true only in it”. Therefore, it looks like Hardt’s claim that “economic laws are true only in theoretical models” sends us to Hobbiton. Another, widely accepted method of testing models’ truth value, is setting the sentences composing the model against sentences describing the corresponding observed phenomena3. We are not able to test directly the law-like sentences, as they are usually hypotheses constructed upon the set of assumptions and disclosed regularities. But we can do that in reference to the statements representing model’s assumptions and to the statements representing outcomes (predictions). Which one are more important in the evaluation of the economic models, is a subject to a dispute referred to, among others by Milton Freedman in his famous essay (Friedman, 2008). Empirical constructivists might say, that both assumptions and law-like statements are of no relevance in respect to their truthfulness, only the accuracy of the predictions counts. If the level of this accuracy is satisfactory for the model’s constructor, taking into account her/his purposes, the model works, period. If it does not, we are free to test any other combination of assumptions and rules. In this account, the truth of assumptions and law-like statements is equally irrelevant, both in a model and in a target system. Others might say that the assumptions are the most important, and if the model does not work, we need to make them more real, to “factualize” them. And realists would rather place the emphasis on the truth (meaning its correspondence to the target system) of the model’s inherent mechanism transforming the input into output, regardless how we call it (causal connection, causal mechanism, economic laws etc.). According to them the economic laws in a given model are true, that means that they sufficiently well represent a certain aspect of a given economic realm. The phrase “truth in the model” is indeed often used by the philosophers of economics. Hardt quotes Rodrik stating that “Models are never true: but there is truth in models”. And Cartwright writes: 3 For the sake of the current discussion let us for a moment forget about the strong constructivists’ view, that those observables are also constructs, burdened with an observer’s previously accepted theories, values and his social context.

136

Marcin Gorazda

(…) the laws of physics are true only of what we made. (…) I have argued that laws are true in the models, perhaps literally and precisely true, just as morals are true in their corresponding fable. (…) Even when the scientific model fit, they do not fit very exactly. This question bears on how true the theory is of the world. (…) This is the question that bears, not on the truth of the laws, but rather on their universality (Cartwright, 1999, p. 48). The corresponding fable, which entails the truthfulness of morals, represents the set of assumptions and restrictions explicitly or tacitly imposed on a model. The truth of economic laws in a model may be evaluated only in combination with this fable, but not universally. But they are true both in a model and in a nomological machine built upon it. If Hardt’s intention was to present the same or similar view, his statements as quoted above are highly misleading as they suggest that the crucial component of a model, the most important bearer of the truth value (at least according to some realists) has no, or insignificant correspondence to the target system and remains in the impound of the theoretical model. They also suggest that if lowering interests rate does not stimulate investments, we have no reason to worry. The economic law is still valid in the theoretical model, and the implied capacity truly exists. On the contrary, we should worry. Using the words of Cartwright, in this case, we do not have any well-functioning nomological machine, so we do not have good reasons to claim that the model represents any economic domain and even fewer reasons to deduce any capacities.

CETERIS NORMALIBUS, RATHER THAN CETERIS PARIBUS What is ceteris paribus clause and what problems it generates is extensively discussed in the philosophical literature (Mucha, 2016; Reiss, 2013; Reutlinger, et al., 2015; Schurz, 2014). Hardt’s stance on that does not fall much away from other philosophers. He rightly noticed that especially in economics, the “other things being equal” may also mean “other things being absent” (ceteris absentibus) or “other things being fixed” (ceteris constantibus), or “other things being right” (ceteris rectis). So economists, more than other scientists, construct models in which they assume that certain factors (or variables) are absent or disregarded (have no impact on other variables), or they are fixed in advance at certain values and these values remain constant regardless of the changing values of other variables (egzo- and endogenous). Those absent, disregarded or fixed variables are quite often confused and put under the collective term ceteris paribus. Philosophers of science try to be more precise, and they often distinguish between them and between the practical testability of the given model and (in terms of Popperian and post-Popperian philosophy of science) its corroboration or falsification, depending on the exact content and characteristics of those “other

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 137

things being equal” (Schurz, 2014). Hardt’s task, however, as he specified in the paper, is not to analyze the exact content of cp-clause in certain or exemplary economic models, but rather to argue that it should be in principle replaced by the ceteris normalibus clause. As the latter is neither commonly used in economics or philosophy, it requires definition. We find it in several fragments of the text. The proposed definition is combined with the theses, already criticized above. Hardt claims that because economic laws can be understood as the laws always true in economic models so ceteris normalibus clause should be synonymous to “in a model” phrase. He further refers to the commonsensical notion of “normality” and admits that though law-statement should be conceptualized as only true in normal circumstances, those circumstances can be found only in a theoretical model. Yet, he does not define the normal conditions in any other way. To sum up, what is normal in economics is specified in the economic model. There are at least two problems with such understanding of ceteris normalibus clause. First, as models quite often represent the selected part of economy in circumstances which are rather abnormal (like financial crisis, e.g. above mentioned model by Billio et al.), the concept of “normal circumstances” introduced by Hardt, is apparently different from concepts of other authors, who previously introduced the notion of ceteris normalibus clause to the philosophy of science (like Schurz, who is also referred to in Hardt’s paper). Second, as the “normal circumstances” are to be found in a model, they are supposed to be defined therein, and by definition, they include all the factors or variables composing those circumstances, or their assumed absence in a model. Thus, the notion of ceteris normalibus designates the set of subsets. The subsets are (supposedly) the assumptions, which has been previously mentioned, quite well recognized in the philosophy of science, being ceteris paribus, ceteris absentibus, ceteris constantibus and ceteris rectis. Is it a matter of terms, or does it have any important ontological or methodological implications? No implications can be found in the paper, so Occam’s razor suggests the additional notion is redundant. But does it mean that the concept of “normal circumstances” plays no part in economic models? They may play, and this is exactly the account of Schurz, but the understanding of the concept is different. Briefly, Schurz would probably never agree that all the conditions composing normal circumstances can be found in a model. The crucial problem is that they cannot because precisely we do not know what they are. To understand this, we need to follow Schurz in his distinction between cp (ceteris paribus) laws and cr (ceteris rectis) “laws”. The general form of the former is: An increase in the value of a variable X leads to an increase (or decrease) in the value of another variable Y, provided that the value of all other variables Z, which are not effects of X (whatever these values are) remain unchanged (Schurz, 2014, p. 1802).

138

Marcin Gorazda

In this formulation it does not matter what the values of “all other variables” are, it matters only that they remain unchanged. Unlike Cr laws, which requires “all other variables” to take the “right” values, without explicitly stating what the right values are. The range of values which is “right” is beyond our current cognition. We only suspect that there is a certain range which is required to sustain the stable relation between variable X and Y, and we can indicate the exemplary values, within which the relation holds. According to Schurz this controversy cp laws and cr “laws” is especially noticeable in theoretical economics. The truth of the law of demand requires not only a cp clause, but also a cr clause: the cp-demand-price relation holds only under the condition of an “ideal market”, which requires that the sellers and buyers are fully informed and free utility maximizers; irrational behaviour and government price regulation (etc.) have to be excluded (Schurz, 2014, p. 1803). Due to this unknown values of “other variables”, the cr “laws” are not eligible to direct falsification. There is however a possibility of eliminating the cr clause (at least in reference to certain law-like formulations) via its replacement by a normality clause and thus construct the co-called normic cr law. In such formulation, the “right” values of the remainder variables are claimed to be the “normal” ones. But this normality is not defined in a model, but outside it, with reference to statistical normality, which further requires minimum regularity observed in the form of “Most As are Bs”. This statistical normality further emerges out of the evolutionary dynamic, which out of the complexity is able to create local orders with the help of self-regulatory, adaptation mechanism. Let us again illustrate the difference with examples. In the above quotation, Schurz rightly notices that one of the cr requirement for law of demand to hold are free, rational, utility maximisers. As we know from various researches of behavioral economists, rational behavior is far from being normal in Schurz’s terms. So this formulation of the law of demand cannot be expressed with the use of cn clause. But in Hardt’s account, as the circumstances are defined in a model of supply and demand, they constitute the “normal” circumstances, and both cp and cr clauses belong to the cn set. We may, however, take into account another, empirically tested regularity which can also be represented by the mathematical equation, namely tips earned by exotic dancers in relation to their ovulation cycle (Miller et al., 2007)4. This model does not require rational agents, but on the contrary, the agents seem at first sight to be quite irrational, spending more money on lap dancers who at the moment of dancing are at the peak fertility. But outside the model, there is a set of assumptions which refer to the situation which is statistically (evolutionary) 4 It

is worth noting that the research described in the paper, was so peculiar that authors was awarded with Ig Nobel prize in 2008 in economics.

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 139

“normal”. Agents offering tips are heterosexual and are not asexual so that they respond relatively to the attractiveness of the exotic dancers, and their taste for the ladies’ beauty is relatively unified. If we placed the experiment in the “abnormal” situation, e.g. in the gays’ club, or if we engaged to the experiment men with a very exceptional (statistically rare) erotic taste, the regularity might disappear. However, all of those “normality” assumptions are not defined in a model. They are not even defined precisely at all. They are implied out of the normal circumstances, in which the regularity holds. As Hardt rejects the concept of regularities as a component of laws, Schurz’s account differs from his also in this point. His ceteris normalibus laws are different from Schurz’s normic cr laws because there are no underlying regularities which they may refer to, and all the variables which compose “normal circumstances” are defined in the model, and not outside it. The ceteris normalibus clause in his account seems to be a redundant term which simply stands for all the model’s assumptions.

CONCLUSIONS 1. If we accept that there is something like economic law and that it should have something in common with the empirically tested, observed realm, regardless of our ontological stance we cannot abandon the requirement of regularities being the necessary component of any scientific law, economics including. Especially we cannot replace the regularity requirement with the ontological concept of capacities. Capacities may help us explain why there are regularities in nature at all, and they help us in reasoning about the possible outcomes, whenever we go beyond the precisely constructed or described nomological machine in terms of Cartwright. Methodologically they are however redundant. We use the regular behavior of the nomological machine or any other setting to reason through an analogy about the possible outcome of other settings, which differs from the former in some points. But if there is no properly functioning nomological machine, any reasoning about capacities has no grounds. 2. If a model predicts certain outcomes, which do not occur in a particular setting, instead of consoling ourselves with the concept of dormant capacities we should rather assume that the model significantly misrepresents its target system. Significant misrepresentation, means, that it misrepresents its target system in reference to possible causal connections or in Hardt’s terms, in reference to alleged capacities. 3. Economic laws cannot be true only in theoretical models. They must be (at least within the range assumed by the researcher) true in the domain represented by the particular model. Otherwise, the notion of “laws true only

140

Marcin Gorazda

in a model” refers to the inherent tautologies, which truth value are checked exclusively by assumptions and adopted inference rules. 4. The term ceteris normalibus in Hardt’s account is redundant because it simply represents a more general set of assumptions, including ceteris paribus, ceteris rectis, ceteris absentibus, ceteris constantibus. As long as the “normal” circumstances are defined in a model, the clause does not improve our understanding of models and makes it even less falsifiable.

REFERENCES Billio M., Getmansky M., Lo, A.W., Pelizzon L. (2012), Econometric Measures of Connectedness and Systemic Risk in the Finance and Insurance Sectors. “Journal of Financial Economics”, June, 104(3), pp. 535−559. Cartwright N. (1999), The Dappled World. A Study of the Boundaries of Science, Cambridge University Press, Cambridge. Friedman M. (2008), The Methodology of Positive Economics, in: D.M. Hausman (eds.), The Philosophy of Economics. An Anthology, Cambridge University Press, Cambridge, pp. 145−178. Hardt Ł. (2017), Economics Without Laws. Towards a New Philosophy of Economics, Palgrave Macmillan, Cham. Heller M. (2006), Racjonalność i matematyczność świata, in: Filozofia i wszechświat, Universitas, Kraków, pp. 37−104. Mäki U. (2009), Realistic Realism about Unrealistic Models, in: The Oxford Handbook of Philosphy of Economics, H. Kinkaid, D. Ross (eds.), Oxford University Press, Oxford, pp. 68−98. Miller G., Tybur, J.M., Jordan B.D. (2007), Ovulatory Cycle Effects on Tip Earnings by Lap Dancers: Economic Evidence for Human Estrus? “Evolution and Human Behavior”, November, 27(6), pp. 375−381. Mucha K. (2016), Kontrowersje związane z klauzulą ceteris paribus, in: ., Metaekonomia. Zagadnienie z filozofii ekonomii, M. Gorazda, Ł. Hardt i T. Kwarciński (eds.), Copernicus Center Press, Kraków, pp. 337−360. Nyman R. et all. (2018), News and Narratives in Financial Systems: Exploiting Big Mata for Systemic Risk Assessment. [Online] Available at: https://www.bankofengland. co.uk/working-paper/2018/news-and-narratives-in-financial-systems[Data uzyskania dostępu: 15 December 2018]. Reiss J. (2013), Philosophy of Economics. A Contemporary Introduction, Routledge, New York. Reutlinger A., Schurz G, Hüttemann A. (2015), Ceteris Paribus Laws. [Online] Available at: https://plato.stanford.edu/archives/spr2017/entries/ceteris-paribus/ [Data uzyskania dostępu: 15 12 2018]. Schurz G. (2014), Ceteris Paribus and Ceteris Rectis Laws: Content and Causal Role, “Erkenntnis”, December, 79(S10), pp. 1801−1801. Załuski W. (2008), Skłonnościowa interpretacja prawdopodobieństwa, Biblos, Tarnów.

ARE THE CONCEPT OF CAPACITIES AND CETERIS NORMALIBUS CLAUSE REDUNDANT? 141

CZY POJĘCIE ZDOLNOŚCI (CAPACITIES) I KLAUZULA CETERIS NORMALIBUS SĄ ZBĘDNE? STRESZCZENIE Tekst stanowi krytyczną odpowiedź na artykuł Łukasza Hardta dotyczący tzw. praw ceteris normalibus. Szczególnie poddaje krytyce trzy główne tezy: 1. Prawa ekonomiczne nie opisują regularności, ale odnoszą się do zdolności (capacities) i mocy (powers); 2. Prawa ekonomiczne są prawdziwe tylko w modelach ekonomicznych; 3. Prawa ekonomiczne są ważne raczej ceteris normalibus niż ceteris paribus. Opierając się na kilku przykładach teoretycznych modeli w ekonomii, autor twierdzi, że: 1. Nie można porzucić wymogu regularności, które stanowią konieczny element każdego prawa naukowego, włączając w to ekonomię. Pojęcie zdolności, nawet jeśli pomocne w rozumowaniu dotyczącym przyczyn i skutków, jest metodologicznie zbędne; 2. Prawa ekonomiczne nie mogą być prawdziwe wyłącznie w modelach teoretycznych. Co najmniej w zakresie założonym przez badacza, muszą one być prawdziwe w domenie reprezentowanej przez dany model; 3. Termin ceteris paribus w ujęciu Hardta jest zbędny, jako że reprezentuje on li tylko bardziej ogólny zbiór założeń obejmujący: ceteris paribus, ceteris rectis, ceteris absentibus, ceteris constantibus. Dopóki warunki „normalne” są zdefiniowane w modelu, dopóty klauzula ta nie poprawia jego rozumienia. Słowa kluczowe: modele ekonomiczne, prawa ekonomiczne, regularności, zdolności, ceteris paribus, ceteris normalibus.

STUDIA EKONOMICZNE 1 ECONOMIC STUDIES NR 1–2 (XCVI–XCVII) 2018

ESEJE

Marian Gorynia*

O PIĘKNIE NAUK EKONOMICZNYCH1

STRESZCZENIE Celem artykułu jest identyfikacja przejawów (aspektów) piękna nauk ekonomicznych. Przez nauki ekonomiczne rozumie się w tym tekście cztery dyscypliny obowiązującej w Polsce do września 2018 r. klasyfikacji dyscyplin, zaliczone do dziedziny nauk ekonomicznych (w kolejności alfabetycznej): ekonomię, finanse, nauki o zarządzaniu i towaroznawstwo. Podstaw koncepcyjnych poszukiwania piękna nauk ekonomicznych należy doszukiwać się w filozofii, a szczegółowiej rzecz ujmując w jej fragmencie, jaki stanowi estetyka. W tekście wykorzystano dwa sposoby pojmowania piękna: piękno apollińskie i piękno dionizyjskie, przy czym ten pierwszy sposób potraktowano jako tradycyjny, klasyczny, a drugi jako alternatywny. Oba te sposoby wykorzystano do zidentyfikowania przejawów piękna nauk ekonomicznych ze względu na przedmiot badania oraz ze względu na zastosowaną metodę badawczą. Podstawową metodą badawczą wykorzystaną przy powstawaniu tego opracowania była analiza literatury przedmiotu. Słowa kluczowe: nauki ekonomiczne, piękno jako kategoria filozoficzna, piękno przedmiotu nauk ekonomicznych, piękno metody nauk ekonomicznych. Klasyfikacja JEL: A10, A11, A12, D00, E00, G00, M00 * Katedra Konkurencyjności Międzynarodowej Wydział Gospodarki Międzynarodowej Uniwersytet Ekonomiczny w Poznaniu; e-mail: Marian.Gorynia@ue.poznan.pl 1 Artykuł powstał na podstawie wykładu wygłoszonego na inauguracji roku akademickiego 2018/2019 na Uniwersytecie Ekonomicznym w Poznaniu w dniu 28 września 2018 roku.

O PIĘKNIE NAUK EKONOMICZNYCH

143

WSTĘP Tytuł opracowania przesądza w pewnym sensie o tym, że nauki ekonomiczne są piękne. Treść artykułu nie jest jednak ślepą apologią pracy wykonywanej przez naukowców uprawiających nauki ekonomiczne. Celem artykułu jest refleksja nad uprawianiem nauk ekonomicznych z położeniem akcentu na aspektach filozoficznych i estetycznych. Nie jest to więc bezkrytyczna afirmacja uprawiania nauk ekonomicznych, ze wskazaniem wyłącznie na pożytki, jakie niosą one ze sobą, ale także zwrócenie uwagi na niebezpieczeństwa, słabości i niedostatki, jakie są przecież ich udziałem. Jak wiadomo – chociażby z teorii rewolucji naukowych Thomasa Kuhna (1968) – każda wiedza jest budowana na zasadzie kumulacji, dokładania cegiełek do dokonań poprzedników. W przypadku tego opracowania inspiracją do zajęcia się zagadnieniem piękna nauk ekonomicznych były prace profesorów Jerzego Wilkina (2009, 2016) i Bogusława Fiedora (2016). Są one niedoścignionym wzorcem rozmachu intelektualnego i głębi refleksji nad pięknem ekonomii. Niemniej jednak w tym miejscu chciałbym zaznaczyć dwie różnice w przyjętym tutaj podejściu. Po pierwsze rozważania podejmowane w tym artykule odnoszą się do całości nauk ekonomicznych jako integralnej dziedziny wiedzy składającej się jeszcze do niedawna z czterech dyscyplin: ekonomii, finansów, nauk o zarządzaniu i towaroznawstwa, a obecnie obejmują dwie dyscypliny: ekonomia i finanse oraz nauki o zarządzaniu i jakości2. Po drugie, o ile prace wspomnianych uczonych były napisane w trybie pytającym, o tyle w tym tekście jest to tryb twierdzący. Nie oznacza to jednak bezkrytyczności wobec nauk ekonomicznych i niewrażliwości na ich słabości. Refleksja nad pięknem nauk ekonomicznych nawiązuje do związków między naukami ekonomicznymi a filozofią i estetyką. Filozofia jest bowiem nauką zajmującą się między innymi pięknem. Piękno jest kategorią filozoficzną i estetyczną. Podjęte tutaj zagadnienie wpisuje się więc w zyskujący na znaczeniu obszar badawczy dotyczący związków między filozofią jako królową nauk a naukami ekonomicznymi jako dziedziną czy subdziedziną, której fundamenty filozoficzne decydują między innymi o jej przydatności dla człowieka. W ramach uwag wstępnych warto jeszcze zacytować słowa J. Wilkina (2016, s. 63), który napisał: „Piękno to wartość, która obok prawdy i dobra przyświeca dociekaniom filozoficznym, a dążenie do niej jest, albo powinno być celem nauki. Zestawienie kategorii piękna i ekonomii jako nauki może wydawać się zaskaku2 Z punktu

widzenia prowadzonych tutaj rozważań zmiany w klasyfikacji dyscyplin dokonane we wrześniu 2018 roku nie mają większego znaczenia. Można przyjąć, że obecna dyscyplina ekonomia i finanse to suma dwóch wcześniej występujących dyscyplin, czyli ekonomii i finansów. Z kolei obecna dyscyplina nauki o zarządzaniu i jakości to wynik połączenia dwóch wcześniej występujących dziedzin, a mianowicie nauk o zarządzaniu i towaroznawstwa (w części). Zob. Rozporządzenie Ministra Nauki i Szkolnictwa Wyższego z dnia 20 września 2018 r. w sprawie dziedzin nauki i dyscyplin naukowych oraz dyscyplin artystycznych (Dz. U. 2018 r. poz. 1818).

144

Marian Gorynia

jące i niezrozumiałe. Ekonomia jest dziedziną wiedzy, z którą pojęcie piękna na ogół się nie kojarzy. Czy w ogóle możemy odnieść je do nauki i jej poszczególnych dyscyplin? Wydaje się, że jeśli już, to raczej do nauki o literaturze, do filozofii, architektury, historii sztuki, teatrologii czy tym podobnych dyscyplin”. Niniejszy tekst stanowi twierdzącą odpowiedź na pytanie Wilkina, z tym że została ona odniesiona do całej dziedziny nauk ekonomicznych, a nie do samej dyscypliny ekonomia.

1. CZYM JEST PIĘKNO? Pojęcie czy idea piękna pełni ważną funkcję w filozofii, historii idei i estetyce. Poświęcono mu wiele miejsca w licznych traktatach. Krótki przegląd, jaki tutaj zostanie przedstawiony, siłą rzeczy musi być pobieżny i powierzchowny, ale jednak podporządkowany celowi tego tekstu, a więc przydatny do operacjonalizacji pojęcia piękna, które ma być odniesione do nauk ekonomicznych. Swoje stanowisko w sprawie piękna formułowali liczni filozofowie, malarze, pisarze, muzycy, a szerzej ludzie sztuki, przedstawiciele nauk przyrodniczych i humanistycznych, a także naukowcy innych specjalności. Wydaje się też, że każdy człowiek ma bardziej lub mniej uświadomione własne poczucie piękna. W powszechnym odbiorze piękno kojarzy się jednak przede wszystkim ze sztuką, a precyzyjniej rzecz ujmując – z wytworami sztuki. Wstępnie można też wyrazić pogląd, że piękno może się odnosić zarówno do nauki, jak i do sztuki – te odmiany piękna silnie z sobą korespondują, zachowując jednak swoje odrębności i subtelności znaczeniowe. Można by więc rozpocząć od postawienia pytań: Czym jest piękno? Czy istnieje jedna, prosta uniwersalna definicja piękna? Jak się okaże niżej, odpowiedzi na te pytania nie są proste. Ze względu na ograniczenia objętości tego tekstu warto przytoczyć wyniki kwerendy na temat rozumienia piękna uzyskane przez B. Fiedora (2016, s. 109–112), który zainspirowany lekturą Historii piękna Umberta Eco zaproponował cztery następujące sposoby pojmowania piękna: 11

piękno apollińskie versus piękno dionizyjskie,

piękno jako ład matematyczny,

piękno jako zgodność z celem,

piękno i wzniosłość.

Rozróżnienie piękna apollińskiego i dionizyjskiego zawdzięczamy Friedrichowi Nietzschemu (2011), który stworzył podział kultury na kulturę apollińską i kulturę dionizyjską. Nietzsche wywodzi ideę piękna apollińskiego od Platona (2007), który wiązał je z harmonią i proporcją, traktując jednocześnie jako wartość autonomiczną (byt idealny), która istnieje niezależnie od swej podstawy fizycznej czy

O PIĘKNIE NAUK EKONOMICZNYCH

145

zmysłowej. Piękno dionizyjskie jest inne – jak pisze Fiedor (2016, s. 110), nawiązując do podziału Nietzschego: „Piękno apollińskie jest według tego wybitnego filozofa zasłoną piękna dionizyjskiego. To ostatnie jest też radosne, ale i często bolesne, a ponadto może też być antytetyczne wobec rozumu, nierzadko pełne bólu, szaleństwa i opętania”. Rozumienie piękna jako ładu matematycznego także nawiązuje do starożytności, w szczególności do prac Platona i Pitagorasa, w erze nowożytnej zaś jego przykładem mogą być poglądy i twórczość Leonarda da Vinci. Renesansowe podejście do piękna jako ładu matematycznego korespondowało jednak z wcześniejszymi koncepcjami: koncepcją matematycznej harmonii wszechświata Bonawentury z Bagnoregio (XIII wiek) oraz z koncepcją przedstawicieli szkoły z Chartres (XII wiek) definiującą „kosmos jako boski ład wszechrzeczy, który przeciwstawia się pierwotnemu, czyli istniejącemu przed aktem boskiego stworzenia świata, chaosowi” (szerzej: Eco, 2010, s. 82–84, za: Fiedor, 2016, s. 111). Pojmowanie piękna jako zgodności z celem wywodzi się od św. Tomasza z Akwinu. Jego zdaniem piękno nie ogranicza się do harmonii polegającej na właściwych proporcjach, ale obejmuje także integralność części, a przede wszystkim zgodność z celem, do którego dana rzecz jest przeznaczona, czyli „milczące współdziałanie treści” (materii rzeczy) (szerzej: Eco, 2010, s. 89, za: Fiedor, 2016, s. 111). W tym miejscu warto wskazać jeszcze na zauważony przez Davida Hume’a „subiektywizm piękna”, ujęty w następujący sposób: „Piękno nie jest własnością przedmiotów samych przez się. Istnieje w umyśle, który je ogląda, a każdy umysł dostrzega inne piękno. Niektórzy widzą brzydotę nawet tam, gdzie inni widzą piękno” (Eco, 2010, s. 245). Z punktu widzenia celu artykułu należy zgodzić się z podkreślonym przez Hume’a subiektywizmem piękna, wątpliwe natomiast wydaje się ograniczenie odnoszenia piękna wyłącznie do przedmiotów, a pominięcie na przykład wytworów rozumu (idei, teorii, modeli itp.). Czwarty, ostatni sposób rozumienia piękna odnosi się do ustalenia relacji między pięknem a wzniosłością. Edmund Burke przeciwstawia klasycznemu apollińskiemu pięknu ideę wzniosłości (Fiedor, 2016, s. 112). Z kolei Fryderyk Schiller określa piękno i wzniosłość mianem dwóch geniuszy, którymi obdarowała nas natura (Fiedor, 2016, s. 112). W nawiązaniu do rozważań Władysława Tatarkiewicza (1970) nad pięknem J. Wilkin (2016, s. 66) zauważa, iż znakiem współczesności jest przekonanie, „że ogólnej teorii piękna nie da się stworzyć”. Według Tatarkiewicza „dłużej niż przez dwa tysiące lat przeważało w kulturze europejskiej przekonanie, że piękno, czyli – harmonia, miara, ład, doskonała proporcja – jest własnością świata i jedną z wielkich idei ludzkości. […] Jest możliwe, a nawet najbardziej prawdopodobne, że idea piękna wróci. – Ale dzisiaj jest w upadku” (Tatarkiewicz, 1970, s. 15). Wydaje się, że sąd Tatarkiewicza jest przesadzony i nazbyt pesymistyczny (w warstwie diagnostycznej). Idea piękna żyje, ale należy raczej mówić o ewolucji jej rozumienia aniżeli o upadku.

146

Marian Gorynia

2. CZYM JEST PIĘKNO W NAUCE? Do udzielenia odpowiedzi na zadane wyżej pytanie może być pomocne – jak się wydaje – przytoczone wcześniej rozróżnienie kategorii piękna apollińskiego i piękna dionizyjskiego wprowadzone przez Nietzschego. Zanim jednak spróbuję na postawione pytanie odpowiedzieć, warto przytoczyć pogląd Józefa Tischnera (2000, s. 72) dotyczący zależności między pięknem a prawdą. Filozof ten pisze: „Pytanie o prawdę to utrapienie filozofa. Utrapieniem poezji jest pytanie o piękno. Nie wolno mieszać ze sobą tych dwóch pytań. U podstaw pomieszania znajduje się bowiem wątpliwe założenie, że prawda albo sama musi być piękna, albo musi prowadzić do piękna”. Ta wypowiedź zasługuje na dwa komentarze. Po pierwsze można przyjąć, że poezja jest w tym wypadku uosobieniem sztuki w szerokim tego słowa znaczenia. Po drugie to, że prawda i piękno nie implikują się wzajemnie, nie oznacza jednak, że prawda nie może być piękna. W wypowiedziach wybitnych uczonych na temat piękna w nauce zdaje się dominować apollińskie rozumienie piękna3. Najczęściej wiąże się ono z prostotą jakiejś teorii lub z możliwością jej ujęcia w postaci zapisu matematycznego, niekoniecznie prostego. Jako przykład można podać za Wilkinem (2016, s. 67) teorię względności Einsteina, „której przypisuje się – mimo całego jej skomplikowania – walor szczególnego piękna matematycznego”. Przekonanie to bardzo trafnie sformułował Michał Heller (2010, s. 26), pisząc: „Matematyczne morze struktur jest nieskończone. Obcowanie z nimi dostarcza silnych przeżyć estetycznych, jak obcowanie z poezją”. Podobne stanowisko zajął słynny fizyk Werner Heisenberg (laureat Nagrody Nobla), który, wypowiadając się na temat zależności ujętych w teorii atomowej, zauważył, że mimo całej swej matematycznej abstrakcji charakteryzują się one niewiarygodną prostotą, dodając, iż nawet Platon nie uwierzyłby, że są tak piękne (za: Wilkin, 2016, s. 67). Co więcej, prowadzi to Heisenberga do wniosku, że: „Tych zależności nie można wymyślić; musiały one istnieć od stworzenia Świata” (za: Wilkin, 2016, s. 67). Można podawać kolejne przykłady obrazujące odniesienie apollińskiego pojmowania piękna do nauki. Kwintesencją takiego podejścia jest pogląd Johannesa Keplera, że „matematyka jest archetypem piękna” (Wilkin, 2016, s. 67). Warto też przytoczyć stwierdzenie Subrahmanyana Chandrasekhara (laureata Nagrody Nobla w dziedzinie fizyki), który napisał: „W istocie, wszystko, co usiłowałem powiedzieć, można zwięźle wyrazić w dwóch łacińskich aforyzmach: Simplex sigillum veri – Prostota to znak prawdy. Pulchritudo splendor veritatis – Piękno splendorem prawdy” (za: Wilkin, 2016, s. 68). 3 Podejście

to koresponduje z tzw. zasadą ekonomii myślenia (brzytwa Ockhama, Lex parsimoniae), zgodnie z którą w wyjaśnianiu, a także w definiowaniu należy dążyć do prostoty („bytów nie należy mnożyć bez konieczności”). Encyklopedia PWN: https://encyklopedia.pwn.pl/haslo/ brzytwa-Ockhama;3881454.html

O PIĘKNIE NAUK EKONOMICZNYCH

147

Tak rozumiana idea piękna apollińskiego odniesiona do nauki przyświecała także niektórym ekonomistom tworzącym zwłaszcza prace z zakresu teorii równowagi ogólnej czy neoklasycznej teorii wzrostu gospodarczego. Opracowania te co do swych założeń upodabniały się metodologicznie do prac z zakresu nauk ścisłych oraz przyrodniczych, gdzie prym wiodła formuła opisu matematycznego. Jak się jednak okazało, piękno tych teorii nie zawsze szło w parze z realizmem. Wobec powyższego spróbujmy zatem rozpatrzyć możliwość odniesienia koncepcji piękna dionizyjskiego do prac naukowych. O ile w naukach przyrodniczych dążenie do apollińskiego rozumienia piękna jest dość powszechne, o tyle przedstawiciele nauk społecznych i humanistycznych zdają się poszukiwać piękna raczej w złożoności i różnorodności, a nie prostocie budowanych teorii. Zwraca na to uwagę wybitny antropolog kultury Clifford Geertz, pisząc: „Jak przypuszczam, elegancja wciąż jeszcze króluje jako cecha uznawana ogólnie za ideał; a jednak na polu nauk społecznych bardzo często to właśnie przez odejście od owego ideału pojawiają się prace naprawdę twórcze” (za: Wilkin, 2016, s. 69). Tenże autor napisał także: „Whitehead zaproponował niegdyś naukom przyrodniczym maksymę «Szukajcie prostoty i nie ufajcie jej»; w odniesieniu do nauk społecznych mógłby równie dobrze zaproponować hasło: «Szukajcie złożoności i porządkujcie ją»” (za: Wilkin, 2016, s. 69). Wydaje się, że zalecenie Geertza można odnieść do całości nauk ekonomicznych w poszukiwaniu swego rodzaju ukierunkowania idei piękna dionizyjskiego w ich stronę. Konkretyzując swoje rozważania na temat piękna ekonomii, Wilkin (2016) uczynił to, wykorzystując głównie dwa aspekty (wymiary) tej dyscypliny: przedmiot ekonomii oraz jej metody badawcze. J. Wilkin pośrednio wskazuje na obecność w ekonomii zarówno elementów piękna apollińskiego, jak i dionizyjskiego. Podejście, jakie zamierzam zastosować w dalszej części artykułu w odniesieniu do całej dziedziny nauk ekonomicznych, wykorzystuje to ujęcie, jednocześnie uwzględniając okoliczności wynikające z koncepcji piękna dionizyjskiego – a więc swego rodzaju dwoistość, dualizm piękna, prowadzące do nadania mu bardziej realistycznego obrazu i zbliżenia do rzeczywistości gospodarczej. Idea piękna apollińskiego odniesiona do nauki może bowiem być ryzykowna z punktu widzenia realizmu poznawczego – niekiedy obraz rzeczywistości tworzony w procesie modelowania bywa jednak nadmiernie oddalony od świata realnego. Z perspektywy prowadzonych tutaj rozważań warto jeszcze nawiązać do stwierdzenia Fiedora (2016, s. 121), który, pisząc o pogłębianiu prawdy naukowej w naukach społecznych, zauważa: „Ta prawda jakże często, a być może że zawsze, jest przy tym konglomeratem (w różnych proporcjach) zarówno klasycznego piękna apollińskiego – harmonii, proporcji i miary – jak i piękna dionizyjskiego: często bolesnego i antytetycznego wobec «czystego rozumu»”. W dziedzinie nauk ekonomicznych proporcje typów piękna pojawiające się w ramach poszczególnych dyscyplin różnią się, niemniej każda z odmian piękna jest w nich obecna, co postaram się wykazać w dalszej części artykułu.

148

Marian Gorynia

3. PIĘKNO NAUK EKONOMICZNYCH W UJĘCIU KLASYCZNYM – ASPEKT PRZEDMIOTU BADANIA4 W momencie przygotowania pierwotnej wersji tego artykułu nauki ekonomiczne stanowiły odrębną dziedzinę wiedzy5. Dziedzina nauk ekonomicznych należała do obszaru nauk społecznych (obok dziedziny nauk społecznych i dziedziny nauk prawnych). W dziedzinie nauk ekonomicznych zostały wyróżnione cztery dyscypliny: ekonomia, finanse, nauki o zarządzaniu i towaroznawstwo. Warto zatem się zastanowić nad tym, czy istniało jakieś wspólne pole badawcze, podstawowy problem badawczy łączący te dyscypliny. Można tutaj wyjść od spostrzeżenia, że zazwyczaj przedstawiciele poszczególnych dyscyplin, subdyscyplin, specjalności, koncepcji, modeli, nurtów, szkół, teorii i paradygmatów badawczych skupiają uwagę na tym, czym ich pola zainteresowań się wyróżniają, pod jakimi względami są specyficzne, na czym polegają ich osobliwości. Dużo mniej uwagi poświęca się kształtowaniu świadomości tego, co jest wspólne dla nauk ekonomicznych, co je wyróżnia spośród innych nauk. Prawdopodobnie większość naukowców uprawiających nauki ekonomiczne zgodzi się z opinią, że podstawowym problemem badawczym w tych naukach jest efektywność wykorzystania rzadkich zasobów w powiązaniu z indywidualnymi i społecznymi skutkami działalności gospodarczej. Zasadne jest jednak pytanie, w jaki sposób wątek ten jest obecny w badaniach prowadzonych w ramach czterech wymienionych dyscyplin. Zgodnie z definicją Lionela Robbinsa (1933), przedmiotem zainteresowania ekonomii jako dyscypliny naukowej jest „ludzkie zachowanie jako związek między danymi celami a ograniczonymi środkami o alternatywnych zastosowaniach”. Środki są równoznaczne z szeroko rozumianymi zasobami. Ekonomia zajmuje się efektywnością rynków w alokacji i koordynacji zastosowań zasobów na różnych poziomach systemu gospodarczego. Najczęściej przyjmuje się, że system ten ma budowę hierarchiczną, tzn. możliwe i uzasadnione jest wydzielanie w nim podsystemów niższego stopnia – w systemie gospodarki globalnej można wyróżniać podsystemy – gospodarki narodowe, w podsystemie gospodarki narodowej można wydzielać podsystemy – branże i sektory, w podsystemie branży można wyróżniać podsystemy – przedsiębiorstwa itd. W związku z tym analizy prowadzone w ekonomii mogą dotyczyć poziomu globalnego, poziomu makro, poziomu mezo, poziomu mikro i poziomu mikro-mikro (por. Gorynia, Kowalski, 2013). Przechodząc do nauk o zarządzaniu, należy przytoczyć opinię Ricky’ego W. Griffina (1999, s. 6), który uważa, że termin „zarządzanie” można zdefiniować z punktu widzenia tzw. perspektywy zasobowej. Zgodnie z nią, każda organiza4 W przygotowaniu

tej części wykorzystano fragmenty opracowania M. Goryni (2018). klasyfikacji dyscyplin, dokonana w 2018 r., nie wnosi z rozważanego tutaj punktu widzenia istotnych zmian – dwie „nowe” dyscypliny powstały bowiem w zasadzie z połączenia czterech „starych” dyscyplin. W związku z tym prowadzone rozumowanie odnosi się do czterech „starych” dyscyplin. 5 Zmiana

O PIĘKNIE NAUK EKONOMICZNYCH

149

cja wykorzystuje zasoby pozyskiwane z otoczenia. Są to ograniczone zasoby: ludzkie, finansowe, materialne i informacyjne. Zarządzanie polega na takim dobieraniu i koordynowaniu wykorzystania rzadkich zasobów, aby możliwe było osiągnięcie celów organizacji. Innymi słowy, zarządzanie zajmuje się alokacją i koordynacją wykorzystania zasobów wewnątrz firmy (organizacji). Warto zaznaczyć, że nauki o zarządzaniu odnoszą się co do zasady do obiektów poziomu mikro. W dyscyplinie „finanse” wątek rzadkich zasobów jest również szeroko obecny. Zasobami są w tym przypadku środki finansowe, które są przedmiotem inwestowania z myślą o osiągnięciu zysków. Można też ująć to w taki sposób, że będące dobrem rzadkim (ograniczonym) zasoby finansowe alokowane są pomiędzy różne możliwości inwestowania, które mają przynosić pożytki właścicielowi środków. Termin „efektywność” sprowadza się tutaj do umiejętności pomnażania posiadanych aktywów. Należy mieć na uwadze to, że dyscyplina „finanse” obejmuje cztery grupy jednostek: finanse publiczne, finanse przedsiębiorstw, finanse rynków i instytucji finansowych oraz finanse prywatne (Flejterski, 2007, s. 72). Czwarta dyscyplina nauk ekonomicznych, czyli towaroznawstwo, mimo swej specyfiki posiada jednak także silne akcenty efektywnościowe, stanowiące jednoznaczną przesłankę zakwalifikowania jej do nauk o efektywności, czyli nauk ekonomicznych. W uchwale Prezydium Centralnej Komisji do Spraw Stopni i Tytułów z dnia 29 maja 2007 roku przyjmującej dokument zatytułowany „Zakres merytoryczny dyscypliny naukowej «towaroznawstwo» w ramach dziedziny «nauki ekonomiczne»” zapisano, że towaroznawstwo jest nauką interdyscyplinarną, łączącą elementy nauk ekonomicznych, przyrodniczo-technicznych i społecznych (CK, 2007). W towaroznawstwie zwraca się szczególną uwagę na kształtowanie i analizę jakości wyrobów, opierając się na badaniach oczekiwań klientów zewnętrznych i wewnętrznych oraz na wyrażaniu tych oczekiwań poprzez zdefiniowanie parametrów jakościowych i technologicznych. Podniesione wyżej cechy towaroznawstwa można, jak się wydaje, sprowadzić do stwierdzenia, że – zajmując się jakością wszelkich komercyjnych wytworów ludzkiej działalności – dyscyplina ta jest zorientowana na efektywność. Badanie jakości oraz formułowanie dyrektyw zapewniających jej przestrzeganie można uznać za odmianę działań proefektywnościowych. W tym sensie towaroznawstwo jest istotnym uzupełnieniem zainteresowań badawczych pozostałych dyscyplin nauk ekonomicznych, zwraca bowiem szczególną uwagę na przyrodniczo-techniczne podstawy efektywności. Jeszcze inaczej rzecz ujmując, towaroznawstwo jest swego rodzaju pomostem przerzuconym pomiędzy dyscyplinami zajmującymi się twardymi, przyrodniczo-technicznymi aspektami jakości i efektywności a ekonomicznymi, finansowymi i menedżerskimi wymiarami prowadzenia działalności gospodarczej. Dzięki towaroznawstwu ocena i kształtowanie efektywności stają się wielowymiarowe, bardziej kompleksowe i pełniejsze. Wypada także zaznaczyć, że obserwując rozwój nauk ekonomicznych jako dziedziny, dość łatwo zauważa się częściowe nachodzenie na siebie zakresów

150

Marian Gorynia

poszczególnych dyscyplin (Gorynia, Jankowska, Owczarzak, 2005; Rudolf, 2016; Klincewicz, 2016). Z jednej strony może to świadczyć o tendencji do integracji dyscyplin w ramach dziedziny, a z drugiej może być intepretowane jako osłabienie uzasadnienia dla istnienia aż czterech dyscyplin w dziedzinie nauk ekonomicznych. Wydaje się jednak, że częściowe nachodzenie na siebie zakresów poszczególnych dyscyplin jest czymś naturalnym, spotykanym często także w innych obszarach i dziedzinach nauki. Jest to także zazwyczaj jedna z przesłanek do podejmowania badań interdyscyplinarnych. Wskazany wyżej wspólny rdzeń nauk ekonomicznych nie oznacza wcale, że wykluczona jest w ich ramach różnorodność i możliwość występowania wielu paradygmatów (wieloparadygmatowość). Różnorodność zainteresowań badawczych w dziedzinie nauk ekonomicznych wiąże się przede wszystkim z rozległością problematyki badawczej w ramach czterech dyscyplin. W tym miejscu można nieco uwagi poświęcić wyróżnikom poszczególnych dyscyplin, a nawet subdyscyplin. Ekonomia jako dyscyplina wyróżnia się tym, że z punktu widzenia ontologii w zasadzie zajmuje się wszystkimi możliwymi poziomami bytu, co znajduje odbicie w takich jej tradycyjnych składnikach jak mikroekonomia i makroekonomia. Do tego dochodzą nowsze komponenty, takie jak: ekonomia globalna i regionalna (na poziomie ponadpaństwowym), mezoekonomia oraz ekonomia mikro-mikro (Gorynia, Kowalski, 2013). W tym artykule ze względu na jego ograniczoną objętość nie da się w pełni udokumentować różnorodności wszystkich poziomów analizy w ekonomii. Tytułem ilustracji można zauważyć, że jeśli chodzi o fragment mikroekonomii skupiający uwagę na firmie, daje się wyróżnić następujące ekonomiczne teorie: neoklasyczną teorię przedsiębiorstwa, menedżerskie teorie firmy, behawioralną teorię firmy, teorię agencji, teorię praw własności, teorię kosztów transakcyjnych, ewolucyjną teorię firmy, teorię produkcji zespołowej oraz teorię zarządzania strategicznego (por. Gorynia, 1998; Koenig, 1993; Noga, 2009). Z kolei w odniesieniu do współczesnej makroekonomii liczba propozycji klasyfikacyjnych dotyczących jej części składowych jest olbrzymia, ale w uproszczeniu można wymienić dwie główne: ekonomię głównego nurtu (mainstream) oraz nurt ekonomii heterodoksyjnej. Do ekonomii głównego nurtu zalicza się zazwyczaj następujące składniki: szkołę neoklasyczną (czasami nową ekonomię klasyczną, choć nie są to identyczne składniki), monetaryzm, teorię racjonalnych oczekiwań, teorię realnego cyklu koniunkturalnego oraz keynesizm (por. Kundera, 2004; Niekipielow, 2016, s. 39; Bałtowski, 2016; Giza, 2016; Bochenek, 2017). Aleksander Niekipielow (2016) do ekonomii głównego nurtu zalicza także ekonomię branży (industrial economics), ekonomię pracy (labour economics), ekonomię międzynarodową (international economics) oraz ekonomię informacyjną (informational economics). Ten sam autor w nurcie heterodoksyjnym wymienia nową ekonomię instytucjonalną (new institutional economics), ekonomię rozwoju (development economics), ekonomię ewolucyjną (evolutionary economics), ekonomię ekologiczną (ecological economics) oraz ekonomię fizyczną

O PIĘKNIE NAUK EKONOMICZNYCH

151

(physical economics). Do najważniejszych szkół nurtu heterodoksyjnego należy także zaliczyć ekonomię behawioralną (Daniel Kahneman, Amos Tversky, Richard Thaler) oraz ekonomię eksperymentalną (Vernon Smith, Kevin McCabe). Inni autorzy do ekonomii heterodoksyjnej zaliczają także szkołę historyczną, szkołę neoaustriacką oraz teorię wyboru publicznego (Kundera, 2004). Bardzo interesującą analizę zależności pomiędzy ekonomią ortodoksyjną, nieortodoksyjną i heterodoksyjną prowadzi Lásló Csaba (2016; 2017). Wśród mniej znanych, bardziej niszowych kierunków badań we współczesnej ekonomii Andrzej Wojtyna (2017) wymienia: ekonomię narracyjną (Robert Shiller), ekonomię tożsamości (George Akerlof i Rachel Kranton) oraz ekonomię manipulacji i oszustwa (George Akerlof i Robert Shiller). Odnosząc kryterium poziomu analizy do nauk o zarządzaniu, należy podkreślić, że dyscyplina ta skupia się na poziomie mikro (firmy, organizacje, ludzie, grupy itp.). W obrębie nauk o zarządzaniu występują następujące teorie: nurt klasyczny (klasyczna teoria organizacji), prakseologiczna teoria organizacji, kierunek empiryczny, ujęcie systemowe, organizacja jako maszyna cybernetyczna, psychologia organizacji, socjologiczna teoria organizacji, koncepcja gry organizacyjnej, ujęcie sytuacyjne – contingency approach, organizacja jako teatr – perspektywa dramaturgiczna, organizacja ucząca się, zarządzanie strategiczne, zwinne zarządzanie, koncepcja zdrowia organizacji, koncepcja pozytywnego potencjału organizacji (por. Koźmiński, 1983, 1987; Gorynia, 1999a, 1999b, 2000; Krzakiewicz, Cyfert, 2013; Klincewicz, 2016). Cechą dyscypliny „finanse” jest, podobnie jak w przypadku ekonomii, pokrywanie polem prowadzonych analiz wszystkich poziomów ontologicznych działalności gospodarczej, ze zwróceniem szczególnej uwagi na aspekty finansowe. Podstawowe, wyjściowe znaczenie mają cztery paradygmaty charakterystyczne dla filozofii nauk społecznych: funkcjonalistyczny, interpretatywistyczny (interpretacyjny), radykalny humanistyczny i radykalny strukturalistyczny (Ardalan, 2008). Na każdym poziomie analiz prowadzonych w dyscyplinie finanse mamy de facto do czynienia ze współistnieniem wielu paradygmatów bardziej szczegółowych niż wymienione przez przywołanego autora. Jeśli zostanie obniżony poziom abstrakcji, to pojawia się przeciwstawienie paradygmatu neoklasycznego finansom behawioralnym (Szyszka, 2009, s. 15–43). Z kolei gdyby zawęzić obszar refleksji do finansów przedsiębiorstw, wówczas podkreślenia wymagałoby dość silne powiązanie tej subdyscypliny z teorią firmy. Inaczej rzecz ujmując, paradygmaty „teorii finansów przedsiębiorstw” w dużym stopniu nawiązują do paradygmatów teorii firmy (Bolton, Scharfstein, 1998; Zingales, 2000; Baker, Gibbons, Murphy, 2000). W tym kontekście można, jak się wydaje, także mówić o paradygmacie nawiązującym do podejścia instytucjonalnego w ekonomii. Z rozpatrywanego punktu widzenia dyscyplina towaroznawstwo (nauki o jakości) w pewnym zakresie jest podobna do nauk o zarządzaniu. O ile nauki o zarządzaniu zajmują się poziomem mikro odnoszonym do organizacji, o tyle przedmiot zainteresowania towaroznawstwa stanowią konkretne wyroby i usługi ze skupieniem uwagi na aspektach jakościowych, rozpatrywanych z punktu

152

Marian Gorynia

widzenia efektywności. Można zatem zasadnie mówić o mikroekonomicznej orientacji towaroznawstwa jako dyscypliny. Przedmiot zainteresowań towaroznawstwa dobrze opisuje model tzw. simpleksu, zgodnie z którym podstawowe cechy produktów (towarów) są określane przez pojęcie jakości produktu (wyrobów lub usług), który to termin jest swoistym towaroznawczym indykatorem. Indykator taki powstaje przez wzajemne sprzężenie zespołu trzech podstawowych parametrów: parametrów ekonomicznych (analiza kosztów we wszystkich fazach cyklu życia produktu); parametrów przyrodniczych (projektowo-technologicznych) oraz parametrów heurystycznych (aspektów społecznych oraz wybranych zagadnień zarządzania produktem w aspekcie zrównoważonego rozwoju). Łącznie parametry te mają decydujący wpływ na poziom zadowolenia klienta. Prostą konsekwencją wynikającą z przytoczonej definicji jest uznanie, że w dyscyplinie „towaroznawstwo” są uzasadnione badania we wszystkich obszarach składających się na „towaroznawczy simpleks”, tzn. zarówno w obszarze nauk przyrodniczo-technologicznych, jak i w obszarze nauk ekonomicznych oraz innych, składających się na pojęcie heurystyki (np. zarządzania normatywnego i marketingu produktu). Przedstawione uwagi i ich rozwinięcia w przywołanej literaturze pozwalają na sformułowanie spostrzeżenia, że poszczególne dyscypliny nauk ekonomicznych mają cechę, którą można by określić jako wieloparadygmatowość. Jeśli przyjąć, że zgodnie z definicją paradygmat może być rozumiany jako zestaw najważniejszych problemów teoretycznych wiążących się z jakimś badanym zagadnieniem, to łatwo jest skonstatować, że w zasadzie w odniesieniu do wszystkich zagadnień badawczych nauk ekonomicznych mamy do czynienia z sytuacją równoległego funkcjonowania wielu paradygmatów, modeli, teorii, nurtów, koncepcji itp. Można zatem mówić o występowaniu swoistego rynku podejść do badanych zagadnień. Niekiedy używa się nawet określenia „dżungla teorii”, tak jak w odniesieniu do teorii zarządzania uczynił Harold Koontz (1961). W przypadku dyscypliny „ekonomia” uwagę na tę okoliczność zwrócił Dani Rodrik (2015), który – parafrazując Koonzta – pisze o czymś, co można by nazwać dżunglą modeli używanych w ekonomii. Rodrik (2015) zwraca uwagą na ich dualizm. Z jednej strony modeli jest wiele, gdyż odnoszą się do różnych komponentów gospodarki, która jest złożona. Z drugiej strony ekonomiści nie ustają w poszukiwaniu jednego, całościowego modelu służącego do opisu całej gospodarki. Mamy zatem dylemat, czy ekonomia powinna się opierać na wielu kontekstowo, sytuacyjnie dopasowanych cząstkowych modelach, czy też zmierzać do budowy jednej ogólnej teorii. Ta druga droga, zdaniem Rodrika, nie jest słuszna. Podobne konstatacje podkreślające wieloparadygmatowość dyscyplin można odnieść także do finansów i towaroznawstwa. Zasygnalizowany rynek paradygmatów charakterystycznych dla poszczególnych dyscyplin podlega swoistej ewolucji. Jedne podejścia dają społecznie użyteczne wyniki i przyjmują się na dłużej, rozwijają się, ewoluują, inne zaś stosowane są coraz rzadziej, popadają w zapomnienie i przepadają. Zostają te, które dają przydatne rezultaty. Funkcjonujące paradygmaty potwierdzają zatem (lub

O PIĘKNIE NAUK EKONOMICZNYCH

153

nie) swoją przydatność i płodność przy budowaniu modeli czy systemów twierdzeń teoretycznych (Sztompka, 1985). Taka ocena przydatności możliwa jest jednak ex post, czasami z dużym czasowym przesunięciem. W podsumowaniu tej części artykułu należy podkreślić, że rozpatrując piękno nauk ekonomicznych z punktu widzenia apollińskiego odniesionego do przedmiotu badania na wyeksponowanie zasługuje to, że cechą wspólną tych nauk jako całości jest podporządkowanie badań teoretycznych i praktycznych zagadnieniu efektywności. Przejawem piękna nauk ekonomicznych jest więc dążenie do uporządkowania wiedzy odnoszącej się do zachowań ludzkich w szeroko rozumianej sferze gospodarowania czyli do zbudowania teorii opisującej i wyjaśniającej te zachowania.

4. PIĘKNO NAUK EKONOMICZNYCH W UJĘCIU TRADYCYJNYM – ASPEKT METODY6 Gdy rozpatrujemy piękno nauk ekonomicznych przez pryzmat metody badawczej, mamy na myśli przede wszystkim ich fundamenty filozoficzne i metodologiczne oraz etyczne. W przypadku nauk ekonomicznych, scharakteryzowanych powyżej, trudno byłoby mówić o obowiązywaniu jednej szkoły czy podejścia do kwestii fundamentów filozoficznych i metodologicznych prowadzonych badań. Należy raczej zidentyfikować i rozważyć spektrum wchodzących w rachubę możliwości oraz kontekst i częstość ich wykorzystania, co z kolei może pozwolić przynajmniej na pośrednie wnioskowanie o efektywności tychże szkół czy ujęć. Ogólnie można zauważyć, że refleksja na temat podstaw filozoficznych i metodologicznych prowadzenia badań ekonomicznych nie jest ich najmocniejszą stroną, a jednocześnie mniej więcej od początku tego wieku, również w Polsce, daje się zaobserwować wzrost zainteresowania tą problematyką. Przejawem takiej prawidłowości było chociażby wydanie kilku książek ogniskujących uwagę na tych zagadnieniach (np. Hardt, 2013; Gorazda, Hardt, Kwarciński, 2016), a także poświęcenie kwestiom metodologicznym całego zeszytu „Economics and Business Review” (Galbács, 2017a, s. 3–6; 2017b, s. 112–134). W badaniach ekonomicznych, podobnie zresztą jak w innych badaniach naukowych, są przyjmowane określone założenia filozoficzne. Spośród nich trzy wydają się najważniejsze: realność zewnętrznego świata, wielowarstwowa struktura rzeczywistości oraz poznawalność świata (Bunge, 1967, s. 291). Często badacze przyjmują je nie do końca świadomie, nie eksponując tychże założeń, korzystają z nich implicite. Zasadniczo można się zgodzić z poglądem Łukasza Hardta (2013, s. 14–15), że sukcesy współczesnej nauki, w tym także nauk ekonomicznych, są w pewnym sensie konsekwencją przyjmowania przez naukowców stanowiska określanego jako realizm filozoficzny. Stanowisko to jest uzna6 W przygotowaniu

tej części wykorzystano fragmenty opracowania M. Goryni (2018).

154

Marian Gorynia

wane za dominujące w jednej z najbardziej podstawowych kwestii filozoficznych, którą jest postrzeganie relacji między badaczem a światem badanym. Związki pomiędzy tym stanowiskiem a innymi możliwymi podejściami najlepiej oddaje następujący cytat: „Większość ekonomistów implicite (a filozofów ekonomii, tak jak Mäki, explicite) za filozoficzny fundament ekonomii jako nauki uznaje szeroko rozumiany realizm, zarazem podzielając przekonanie, że alternatywę dla niego stanowi w najlepszym razie instrumentalizm (dość specyficznie zresztą interpretowany), w najgorszym zaś «pusty» formalizm (określany zazwyczaj mianem «blackboard economics»). Z tej perspektywy uderzające jest również to, że – szczególnie na tle sporów toczonych w obszarze ogólnej filozofii – filozofię ekonomii wyraźnie wyróżnia nie tylko prymat tego jednego nurtu, ale też właściwie całkowity brak niektórych, możliwych do zajęcia (a w tejże ogólnej filozofii wręcz wiodących) stanowisk” (Scheuer, 2016, s. 69–70). Konsekwencje względnie zgodnego optowania przedstawicieli nauk ekonomicznych za realizmem filozoficznym odnoszą się także do kwestii etycznych, których przejawem jest pogląd, że aktywność naukowa jest dobra, pożyteczna i wartościowa (por. Hardt, 2013, s. 15). Drugą ważną okolicznością jest uznanie przez ekonomistów postulatów biologów na temat hierarchicznej budowy wszechświata (Bertalanffy, 1968; Hammond, 2003). Przyjęcie tej perspektywy skutkuje wśród przedstawicieli nauk ekonomicznych względną zgodą co do tego, że świat badany przez nauki ekonomiczne jest wielopoziomowy. Przedmiotem zainteresowania nauk ekonomicznych są zatem rozmaite jednostki analizy, należące do różnych poziomów ontologicznych (Gorynia, 1993). Wcześniej sygnalizowano jednak, że nie wszystkie dyscypliny nauk ekonomicznych podejmują badania na wszystkich możliwych poziomach analizy. Przedstawiciele nauk ekonomicznych dość rzadko zastanawiają się także nad przyjmowanymi w badaniach założeniami dotyczącymi relacji pomiędzy częścią a całością, która również należy do zasadniczych rozstrzygnięć filozoficznych w prowadzeniu badań naukowych. Dylemat ten dotyczy wyboru jednego z trzech możliwych stanowisk: redukcjonizmu, holizmu lub systemizmu (Bunge, 1979; Sztompka, 1985). Wśród przedstawicieli nauk społecznych najczęściej za efektywne z naukowego punktu widzenia uważane jest stanowisko systemizmu – zarówno w wymiarze ontologicznym, jak i metodologicznym. Kolejnym ważnym aspektem prowadzenia badań w dziedzinie nauk ekonomicznych jest sposób formułowania prawidłowości pretendujących do miana praw nauki lub twierdzeń naukowych. Pod tym względem nauki ekonomiczne są dość silnie zróżnicowane w przekroju poszczególnych dyscyplin, choć jednocześnie są zauważalne pewne podobieństwa. W ramach nauk ekonomicznych mamy zatem do czynienia zarówno z występowaniem podejścia idiograficznego, jak i nomotetycznego. Wydaje się, że sytuację w tym zakresie najlepiej oddają zapisy dokumentów Centralnej Komisji do Spraw Stopni i Tytułów (CK, 2007; 2010). Warto przy tym zwrócić uwagę na częściową odmienność metod stosowanych w dyscyplinie towaroznawstwo – w szerszym zakresie są tam wykorzystywane

O PIĘKNIE NAUK EKONOMICZNYCH

155

metody właściwe naukom eksperymentalnym i przyrodniczym (chemii, fizyce, biologii itp.), a w odniesieniu do oceny poziomu jakości towarów są używane do tego celu własne metody badawcze, między innymi sensoryczne i eksploatacyjno-użytkowe. Ciągłe doskonalenie tych metod jest oparte na instrumentach statystycznych i standardach jakości (por. CK, 2007). Zróżnicowanie metodologiczne i metodyczne nauk ekonomicznych jest, jak się wydaje, ich silną stroną. Świadczy ono niejednokrotnie o uwzględnieniu postulatu interdyscyplinarności w podejściu do badanych problemów, a umiejętność posługiwania się różnymi metodami w celu rozwiązania złożonych zagadnień o charakterze zarówno poznawczo-teoretycznym, jak i praktyczno-aplikacyjnym należy uznać za zaletę (por. Fiedor, 2013). Z metodami stosowanymi w naukach ekonomicznych koresponduje język tych nauk. W tym zakresie można zauważyć współistnienie dwóch tendencji. Pierwsza kładzie nacisk na wspólnotę i konwergencję przynajmniej części języków używanych przez poszczególne dyscypliny. Można ją określić jako tendencję dośrodkową, zmierzającą do zbieżności języków dyscyplin. Druga tendencja, odśrodkowa, przejawia się w rosnącym poziomie hermetyczności języków poszczególnych dyscyplin, szkół, nurtów, paradygmatów itp. Ewolucji nauk ekonomicznych towarzyszy ścieranie się tych dwóch tendencji, przy czym ich wypadkowa pozostaje – jak się wydaje – na w miarę stałym poziomie zależności. Innymi słowy, nie jest zauważalna wyraźna dominacja którejś z zasygnalizowanych tendencji. Osobną wartą zauważenia w tym miejscu kwestią jest rola języka matematyki związana ze stosowaniem metod ilościowych. Także tutaj racjonalnym wyjściem jest podejście kompromisowe zakładające raczej komplementarność języka matematycznego i języka opisowego (literackiego), a nie ich wzajemne wykluczanie się (Landais, 2017; Rodrik, 2015). Język używany przez przedstawicieli nauk ekonomicznych może także być rozpatrywany z punktu widzenia piękna. W moim przekonaniu swoistego piękna nie można odmówić na przykład niektórym określeniom używanym w badaniach, takich jak: dylemat więźnia, paradoks hazardzisty, malejąca krańcowa zmiana użyteczności, błąd faworyta i kandydata o małej szansie, pułapka względności, efekt wabika, efekt mniej znaczy więcej, organizacja jako teatr itp. W konkluzji tej części artykułu należy zwrócić uwagę na zróżnicowanie metodologiczne nauk ekonomicznych. Zazwyczaj za przejaw piękna apollińskiego w dziedzinie metody uznaje się wykorzystanie metod matematycznych. Zakres ich stosowania w poszczególnych dyscyplinach i subdyscyplinach nauk ekonomicznych jest silnie zróżnicowany.

5. ALTERNATYWNE ROZUMIENIE PIĘKNA NAUK EKONOMICZNYCH Pozostając pod wpływem wcześniej zarysowanego rozróżnienia pojęcia piękna apollińskiego i dionizyjskiego, niżej proponuję autorskie rozumienie piękna nauk

156

Marian Gorynia

ekonomicznych z szerszym wyeksponowaniem tych elementów, które w sensie symbolicznym można zaliczyć do idei piękna dionizyjskiego. Przedłożona operacjonalizacja piękna ma charakter po części eksperymentalny i w związku z tym należy sobie zdawać sprawę z jej potencjalnych słabości oraz z tego, że może budzić kontrowersje. W proponowanym tutaj ujęciu piękno może być traktowane jako uosobienie, ekwiwalent, synonim tego co pozytywne. Piękna strona czegoś to to, co pozytywne. Tym, co negatywne, jest wyłom w pięknie, podważenie, odstępstwo od piękna albo wręcz jego zaprzeczenie. Piękno nie jest więc kategorią absolutną; jest kategorią relatywną, obiekty zaś, którym przypisywany jest atrybut piękna, są zazwyczaj piękne pod pewnymi względami, a pod innymi są pozbawione tej cechy. Mamy tutaj zatem do czynienia ze swego rodzaju dwoistością rozumienia piękna czy wręcz z dostrzeganiem antynomii pomiędzy pięknem a niepięknem, czyli brzydotą7. Można się więc pokusić o przeprowadzenie swego rodzaju analizy mocnych i słabych stron nauk ekonomicznych. Powinna ona pokazać, jakie są i na czym polegają silne strony tych nauk, oraz zidentyfikować i opisać słabości, które są ich udziałem. Zaprezentowana analiza ma charakter autorski i subiektywny; jest nie tyle zamknięciem, podsumowaniem dyskusji na ten temat, ile zachętą do permanentnej debaty nad doskonaleniem nauk ekonomicznych. Lista „grzechów” nauk ekonomicznych jest długa. Zarzuty w różnym stopniu odnoszą się do poszczególnych dyscyplin. Poniższa próba uśrednienia i uogólnienia daleka jest więc zapewne od doskonałości. Wśród głównych głosów krytyki w stosunku do nauk ekonomicznych na czoło wysuwa się zarzut o nierealistyczność przyjmowanych założeń, w tym przyjmowanie stereotypu homo oeconomicus i założenia o dążeniu ludzi do maksymalizacji użyteczności oraz założenia o maksymalizacji zysków przez przedsiębiorstwa. Odpowiedź nauk ekonomicznych na zasygnalizowaną krytykę jest dwojaka. Po pierwsze w części nauk ekonomicznych, zaliczanej do tzw. ekonomii głównego nurtu, pojawiły się w odpowiedzi na tę krytykę koncepcje uwzględniające dezyderaty krytyków, czego przykładem może być odejście w niektórych modelach od hipotezy doskonałej racjonalności na rzecz zasady ograniczonej racjonalności. Po drugie ważne jest także wskazanie na to, że w ramach różnorodności nauk ekonomicznych pojawiają się liczne koncepcje alternatywne w stosunku do tych, które są krytykowane; są one szeroko dyskutowane i rozwijane. W tym sensie są one poddawane swoistemu testowi przydatności do wyjaśniania i predykcji zjawisk i procesów ekonomicznych. W moim odczuciu dwa wymienione założenia są wprawdzie niedoskonałymi, ale chyba jednak najlepszymi znanymi, wchodzącymi w rachubę podstawami wyjaśniającymi mechanizmy gospodarczych zachowań ludzi. Wydaje się, że w tej partii ekonomia ortodoksyjna (bo głównie jej dotyczą omawiane zarzuty) jest wysoce realistyczna – bierze jako dane zachowania człowieka podporządkowane tym wartościom, 7 Warto w tym miejscu zaznaczyć, że U. Eco historii piękna poświęcił traktat liczący 437 stron (Eco, 2007). Historia brzydoty została przez tego autora opracowana na 452 stronach (tamże).

O PIĘKNIE NAUK EKONOMICZNYCH

157

wywodząc je z jego (egoistycznej) natury, mimo wielu odstępstw od tego stereotypu w realnym życiu gospodarczym. Wielu ekonomistów (i nie tylko) nie szczędziło trudu na krytykowanie tego podejścia, mimo to jednak nie zdołano zaproponować alternatywnych konceptów użytecznych do ogólnego, teoretycznego wyjaśnienia zachowań człowieka w sferze gospodarki. Warto podkreślić, że ekonomiści głównego nurtu nie twierdzą, że ludzie są doskonale racjonalni. Niemniej przyjmują to założenie jako dobre przybliżenie, które pozwala na budowanie modeli. Nauki ekonomiczne – a ściślej rzecz ujmując, niektóre ich części składowe – są także przedmiotem krytyki za nadmierną aksjomatyzację i oderwanie od życia, co z kolei implikuje możliwość posługiwania się zaawansowanym instrumentarium ilościowym. Zarzut ten koresponduje z poprzednim i dotyczy tych fragmentów nauk ekonomicznych, które dążą do odkrywania harmonii porządku gospodarczego wzorem nauk przyrodniczych poszukujących harmonii porządku panującego w przyrodzie i kosmosie czy wzorem nauk technicznych zmierzających do wykrycia tejże harmonii w doskonałości układów, systemów czy nawet określonych maszyn i urządzeń zbudowanych przez człowieka (por. Fiedor 2016, s. 118). Dążeniu do elegancji opisu matematycznego towarzyszy nierzadko brak realizmu polegający na przyjmowaniu oderwanych od rzeczywistości założeń. W efekcie powstają koncepcje tyleż piękne z formalnego punktu widzenia, co nieprzystające do rzeczywistości i oferujące z tego powodu ograniczone możliwości predykcyjne. Pod adresem dyscypliny ekonomia bywa formułowany zarzut imperializmu, a w szczególności do jej tzw. twardego rdzenia, zaksjomatyzowanego i zmatematyzowanego, pozwalającego na ścisłe rozumowanie. Sformalizowanie ekonomii dodało jej znaczenia i powagi, co doprowadziło niektórych do sformułowania uwagi o imperializmie ekonomicznym (Brzeziński, Gorynia, Hockuba, 2008). Niektórzy autorzy uznali, że doszło do zdominowania przez ekonomię obszaru badawczego nauk społecznych (Fine, 2000; Murak, 2014). Może się to przejawiać dwojako: po pierwsze ekonomia jest dość powszechnie traktowana jako dyscyplina najbardziej „naukowa” spośród nauk społecznych, tj. precyzyjna, rygorystyczna, sformalizowana i wiarygodna – jest to określane jako imperializm w sensie wyższości. Po drugie domena dociekań tej dyscypliny stale się powiększa, wkraczając w sferę moralności, polityki, życia rodzinnego i licznych innych typów zachowania jednostek tradycyjnie będących poza obszarem dociekań ekonomicznych – oznacza to imperializm w sensie inwazyjności. W kontekście trzech wymienionych zarzutów stawianych naukom ekonomicznym warto jednak przytoczyć pogląd laureata Nagrody Nobla w dziedzinie ekonomii Gary’ego Beckera (1990, 1993), iż podejście ekonomiczne nie oznacza, że egoizm i korzyści materialne są jedynymi motywami zachowań ludzkich. Podejście ekonomiczne jest metodą analizy, a nie założeniem dotyczącym jakichś szczególnych motywacji. Ludzkie zachowanie jest uwarunkowane przez szerszy zestaw wartości i preferencji. Analiza ekonomiczna zakłada, że jednostki maksymalizują dobrobyt tak, jak go rozumieją, niezależnie od tego czy są egoistyczne, altru-

158

Marian Gorynia

istyczne, lojalne, złośliwe, czy masochistyczne. Wydaje się, że krytycy nauk ekonomicznych często zapominają o wyłożonych przez Beckera zasadach podejścia ekonomicznego. Osobnym nurtem krytyki nauk ekonomicznych są ich ograniczone możliwości predykcyjne, zwłaszcza w zestawieniu z teoriami i modelami występującymi na przykład w naukach przyrodniczych. O ile niektóre teorie ekonomiczne (na przykład prace związane z nurtem równowagi oraz prace z zakresu teorii wzrostu) dorównują pod względem elegancji opisu i stopnia zaawansowania instrumentarium matematycznego pracom fizyków, astronomów, biologów itp., o tyle często podkreśla się, że zdolności nauk ekonomicznych do przewidywania rzeczywistości są zdecydowanie mniej zaawansowane, co ma wskazywać na ich istotną słabość. Zarzut ten nie jest odnoszony li tylko do ekonomii czy makroekonomii; stosunkowo małe możliwości przewidywania przyszłych zdarzeń przypisywane są także mikroekonomii, finansom, naukom o zarządzaniu, a nawet towaroznawstwu. Na czele tej krytyki znajduje się zwłaszcza zarzut, że nauki ekonomiczne nie potrafią przewidywać kryzysów i w konsekwencji im zapobiegać. Odnosi to się nie tylko to światowego kryzysu gospodarczego zapoczątkowanego w 2008 r., ale także do problemów w funkcjonowaniu przedsiębiorstw, związanych z trudnościami finansowymi, niewypłacalnością, chybionymi kampaniami marketingowymi, nierzadko skutkującymi upadłościami. Z jednej strony standardowe wyjaśnienie tej niedoskonałości nauk ekonomicznych odwołuje się do argumentu, że w procesach gospodarczych największe znaczenie mają zachowania i decyzje podejmowane przez ludzi. Mechanizmy tych zachowań i decyzji są na tyle skomplikowane, że nie poddają się niezawodnej predykcji, jak na przykład funkcjonowanie systemów technicznych. Głębsza refleksja nad tym zagadnieniem pozwala jednak zauważyć, że nawet nauki, które powszechnie uważa się za bardziej zaawansowane aniżeli nauki ekonomiczne, także miewają z tym problemy (np. medycyna, geologia, meteorologia – lekarz, który nie potrafi przewidzieć choroby pacjenta; geolog, który nie potrafi przewidzieć trzęsienia ziemi; meteorolog, który nie potrafi prognozować załamania pogody). Co więcej, w kwestii zdolności do wyjaśniania i predykcji nauki ekonomiczne nie muszą wcale mieć kompleksów w stosunku do nauk uważanych za bardziej zaawansowane. Jak zauważa Mark Blaug (1995, s. 42–46), gdy przyjrzymy się dwom słynnym teoriom – grawitacji i ewolucji, wówczas okaże się, że ta pierwsza nic nie wyjaśnia w sensie znalezienia mechanizmu przyczynowego, ale wszystko doskonale przewiduje; ta druga zaś świetnie wyjaśnia (podaje opis mechanizmu odpowiedzialnego za proces ewolucji), ale nie daje absolutnie podstaw do budowania prognoz. Podobne uwagi można zresztą odnieść na przykład do psychologii głębi Sigmunda Freuda czy teorii samobójstwa Emila Durkheima. Przywoływany wcześniej Wilkin (2016, s. 63–64) przytacza także za Thomasem Carlyle’em i Kennethem Arrowem pogląd o posępności ekonomii, co można tutaj potraktować szerzej i odnieść do całości nauk ekonomicznych. Posępność ta miałaby wynikać z tego, że ekonomiści niekiedy muszą w zgodzie z realiami

O PIĘKNIE NAUK EKONOMICZNYCH

159

uświadamiać swojemu otoczeniu, że zasoby, jakimi dysponują ludzie, są rzadkie i że nie wszystko daje się zrobić od razu. Przeciwstawiając się takiemu stanowisku, można argumentować, że to dzięki ekonomistom ludzie mają świadomość ograniczonych możliwości zaspokajania swoich potrzeb oraz konieczności rozkładania ich w czasie. Wręcz przeciwnie, należy zauważyć, że dzięki wiedzy ekonomicznej ludzkim zachowaniom można nadawać charakter racjonalny, a tym samym realistyczny. Dla mnie jako ekonomisty nauki ekonomiczne są piękne, dlatego że z wdziękiem, ale nie bez dyskursu i debaty, opisują świat gospodarczej aktywności człowieka, przyczyniając się do uczynienia tej działalności w tendencji bardziej racjonalną i w ten sposób prowadząc do podniesienia efektywności działalności gospodarczej, a tym samym poziomu dobrobytu ludzkości. Rzekoma posępność nauk ekonomicznych po głębszym namyśle powinna być więc traktowana nie jako słabość nauk ekonomicznych, ale jako ich największa zaleta. Nauki ekonomiczne poprzez główny przedmiot swoich zainteresowań, czyli efektywność, nadają ludzkim zachowaniom wymiar racjonalny, co nie wiąże się wcale z nieuzasadnionym uświadamianiem ograniczoności zasobów, ale prowadzi do ustalenia hierarchii zaspokajania (nieograniczonych) ludzkich potrzeb przy wykorzystaniu rzadkich środków. Konsekwencją stosowania podejścia ekonomicznego jest więc określony ład i porządek w tymże zaspokajaniu potrzeb ludzkich. Tak więc z zarzutu o posępność ekonomii czy całej dziedziny nauk ekonomicznych można wyprowadzić ich najważniejszą silną stronę. Ta silna strona może być wyjaśniona dwojako. Z punktu widzenia funkcjonalnego nauki ekonomiczne są przydatne i pożyteczne, gdyż ich rolą (funkcją) pełnioną w społeczeństwie jest przyczynianie się do racjonalizacji gospodarczych zachowań człowieka, dzięki czemu sprawiają, że możliwy i realny staje się wyższy poziom dobrobytu. Podobnie może być wyjaśnione miejsce nauk ekonomicznych z punktu widzenia teleologicznego. Jeśli w uproszczeniu przyjąć, że celem wspólnym przyświecającym ludziom jest dobrobyt (albo jego wzrost), to naukom ekonomicznym zawdzięczamy chociaż w części, że cel ten przynajmniej w jakimś stopniu jest realizowany. W ten sposób wyprowadzona rola nauk ekonomicznych współdeterminuje poziom rozwoju cywilizacyjnego ludzkości. Drugą jasną stroną nauk ekonomicznych, służebną w stosunku do pierwszej, najważniejszej, jest ich zdolność do ewolucji, do korekt i do wchłaniania oraz uwzględniania nowych zjawisk i procesów. Zwrócono na to uwagę wcześniej, omawiając piękno przedmiotu i metody nauk ekonomicznych. Świadczy o tym różnorodność ich zainteresowań, ich wieloparadygmatowość oraz bogactwo metodologiczne. Jednocześnie przedstawiciele tych nauk potrafią wyciągać wnioski z niedoskonałości wypracowanych koncepcji i zmieniać je w takim kierunku, by lepiej wyjaśniały zachodzące zmiany oraz pozwalały formułować coraz lepsze (niestety niedoskonałe) predykcje. Trzecia silna strona nauk ekonomicznych może być kontrargumentem dla wcześniej podniesionego zarzutu o imperializm. Tą siłą jest zdolność i gotowość nauk ekonomicznych do współpracy z innymi dyscyplinami, dziedzinami i obsza-

160

Marian Gorynia

rami wiedzy, w tym także interdyscyplinarność tychże nauk. Interdyscyplinarność odniesioną do nauk ekonomicznych można rozumieć co najmniej na cztery sposoby: 11

prowadzone w danej dyscyplinie badania odwołują się wspomagająco do innych dyscyplin zlokalizowanych w innym obszarze nauki (na przykład odwołania ekonomii oraz nauk o zarządzaniu do ogólnej teorii systemów mającej korzenie biologiczne; inny przykład to nawiązania ekonomii i nauk o zarządzaniu do teorii ewolucji – ewolucyjna teoria firmy);

prowadzone w danej dyscyplinie badania odwołują się wspomagająco do innych dyscyplin zlokalizowanych w innej dziedzinie nauki, ale należącej do tego samego obszaru nauki (na przykład ujęcie firmy w ramach ekonomicznej teorii praw własności nawiązuje do dyscypliny „prawo” należącej do dziedziny nauk prawnych; inny przykład to nawiązanie w ekonomii i naukach o zarządzaniu do znanej z socjologii koncepcji zakorzenienia; kolejny przykład to czerpanie z dorobku psychologii przez naukowców uprawiających finanse, co doprowadziło do wykształcenia się finansów behawioralnych; co więcej – wykorzystywanie przez ekonomistów dorobku psychologii doprowadziło nie tylko do powstania finansów behawioralnych, ale co ważniejsze – całej szkoły w ekonomii „ekonomia behawioralna”);

prowadzone w danej dyscyplinie badania odwołują się wspomagająco do dyscyplin zlokalizowanych w tej samej dziedzinie (na przykład teorie firmy rozwijane w ramach ekonomii oraz nauk o zarządzaniu często odwołują się do koncepcji wypracowanych przez dyscyplinę „finanse”);

prowadzone w danej subdyscyplinie badania nawiązują wspomagająco do badań należących do innych subdyscyplin w ramach danej dyscypliny (ten przypadek interdyscyplinarności daje się łatwo zakwestionować z czystej pozycji logicznej, ale przez niektórych bywa on respektowany) (Gorynia, 2016).

6. CO MOŻEMY ZROBIĆ, BY NAUKI EKONOMICZNE BYŁY JESZCZE PIĘKNIEJSZE?8 Przedstawione silne i słabe strony nauk ekonomicznych powinny być punktem wyjścia do ich doskonalenia. W tym miejscu można podjąć próbę sformułowania na podstawie przeprowadzonych rozważań kilku rekomendacji pod adresem naukowców uprawiających nauki ekonomiczne z myślą o zwiększeniu pożytków płynących z ich pracy dla społeczeństwa. Pierwszy postulat to otwarcie się szeroko rozumianego środowiska przedstawicieli nauk ekonomicznych na argumenty inne aniżeli ściśle ekonomiczne. Eleganckie modele budowane przez ekonomistów nie zawsze przystają do świata, 8 W przygotowaniu

tej części wykorzystano fragmenty opracowania Goryni (2018).

O PIĘKNIE NAUK EKONOMICZNYCH

161

który mają opisywać. Oznacza to konieczność rozbudowywania modeli i uwzględniania w nich zmiennych, które dotychczas były nieobecne w badaniach ekonomicznych, przy zachowaniu poczucia realizmu. Po drugie, przedstawiciele nauk ekonomicznych powinni pokryć swoimi zainteresowaniami te pola, które nie zostały jeszcze przez nich wystarczająco zbadane, oraz uwzględnić zmienne, które tradycyjnie nie były przedmiotem ich zainteresowań. Spełnieniem tego postulatu jest zauważone już powstanie i rozwój takich pól badawczych, jak ekonomia szczęścia, ekonomia oszustwa itp. Na podobnej zasadzie można postulować rozwijanie ekonomii dobroczynności. Przy tym należy zaznaczyć, że nie chodzi tutaj o dominację, supremację ani imperializm ekonomii w innych sferach nauki czy realnego życia. Po trzecie, ważne jest odejście od arogancji właściwej niektórym grupom przedstawicieli nauk ekonomicznych. Należy zwłaszcza przestrzegać przed przywiązywaniem zbyt dużej wagi do argumentów wynikających z niektórych modeli formalnych, opartych na specyficznych założeniach. Oderwanie od realnego świata nie może być traktowane jako synonim naukowości. Po czwarte, należy zalecać pokorę w stosunku do świata rzeczywistego i do danych empirycznych, które go opisują. Wydaje się, że rozwój badań empirycznych jest uzasadniony, konieczny i nieunikniony. Unaocznił to dobitnie światowy kryzys gospodarczy zapoczątkowany w 2008 roku. Konfrontacja (w sensie testowania) koncepcji teoretycznych z realiami życia gospodarczego nie ma alternatywy. Po piąte, warto także, aby przedstawiciele nauk ekonomicznych przywiązywali odpowiednią wagę do poczucia tożsamości nauk, które uprawiają i reprezentują, oraz do zrozumienia i eksponowania tego, co wspólne. Tutaj na ponowne wyeksponowanie zasługują koncepcje efektywności i rzadkości zasobów. Po szóste, uznając potrzebę i konsekwencje specjalizacji we współczesnych naukach ekonomicznych, nie należy fetyszyzować podziałów formalnych. Budowanie granic między dyscyplinami może bowiem prowadzić do zaściankowości. Świat rzeczywisty jest z natury interdyscyplinarny i w badaniach naukowych, zwłaszcza tzw. badaniach stosowanych, trzeba to respektować. Po siódme, wskazana jest, często dzisiaj nieobecna w uzasadnionym zakresie, refleksja nad statusem nauk ekonomicznych z punktu widzenia filozofii nauki. Fundamenty filozoficzne są ważne i brzemienne w skutki dla prowadzonych badań niezależnie od tego, czy zdajemy sobie z tego sprawę, czy nie. Rozwijanie i uświadamianie podstaw filozoficznych będzie mieć kluczowe znaczenie dla jakości rozwoju nauk ekonomicznych. Po ósme, należy zalecać otwarte podejście do wieloparadygmatowości i zrozumienie różnorodności nauk ekonomicznych, unikanie uprzedzeń. Wskazana jest umiejętność czerpania korzyści z wieloparadygmatowości dziedziny nauk ekonomicznych i poszczególnych dyscyplin, a rynek paradygmatów powinien ewoluować zgodnie z postulatem twórczej destrukcji. Po dziewiąte, wiele wskazuje na to, że co do zasady lepsze wyniki daje uprawianie nauki w zespołach badawczych. Pojawia się zatem rekomendacja prowa-

162

Marian Gorynia

dzenia badań w skali większych zespołów dopasowanych do rozmiarów i doniosłości problemów badawczych. Wydaje się bowiem, że będzie wzrastać zapotrzebowanie na prace syntetyczne, integrujące w teorie wyższego szczebla wyniki prac powstałych w nurcie analitycznym, wyspecjalizowanym i sfragmentaryzowanym. Po dziesiąte, współpraca z otoczeniem międzynarodowym jest absolutnym priorytetem. Era naszej samowystarczalności dydaktycznej odchodzi w niepamięć, a brak nadążania za trendami światowymi w uprawianiu nauki spowoduje, że nasi studenci wyjadą za granicę.

ZAKOŃCZENIE Podsumowując, można stwierdzić, że zwłaszcza dionizyjskie pojmowanie piękna daje potencjalnie interesujące wyniki, jeśli chodzi o rozpatrywanie tego atrybutu nauk ekonomicznych. W tym ujęciu piękno nauk ekonomicznych należy wiązać przede wszystkim z ich funkcją jako katalizatora postępu gospodarczego, pełnioną w rozwoju cywilizacyjnym, przyczyniającego się do wzrostu dobrobytu ludzkości. W tym aspekcie – jak starano się wykazać – nauki ekonomiczne są naprawdę piękne, aczkolwiek niedoskonałe.

BIBLIOGRAFIA Ardalan K. (2008), On the Role of Paradigms in Finance, Ashgate, Burlington. Baker G., Gibbons R., Murphy K.J. (2002), Relational Contracts and the Theory of the Firm, “Quarterly Journal of Economics”, 117(1), ss. 39–84. Bałtowski M. (2016), Wstęp, w: Ekonomia przyszłości. Wokół nowego pragmatyzmu Grzegorza W. Kołodko, M. Bałtowski (red.), Wydawnictwo Naukowe PWN, Warszawa. Becker G.S. (1990), The Economic Approach to Human Behavior, The University of Chicago Press, Chicago. Becker G.S. (1993), The Economic Way of Looking at Life, “Journal of Political Economy”, 101(3), ss. 383–409. Bertalanffy L. von (1968), General System Theory. Foundations, Development, Applications, New York. Blaug M. (1995), Metodologia ekonomii, Wydawnictwo Naukowe PWN, Warszawa. Bochenek M. (2017), Historia rozwoju ekonomii. Współczesne szkoły ekonomiczne, Wydawnictwo Naukowe UMK, Toruń. Bolton P., Scharfstein D.S. (1998), Corporate Finance, the Theory of the Firm, and Organizations, “Journal of Economic Perspectives”, 12(4), ss. 95–114. Brzeziński M., Gorynia M., Hockuba Z. (2008), Ekonomia a inne nauki społeczne na początku XXI w. Między imperializmem a kooperacją, „Ekonomista”, 2, ss. 201–232. Bunge M. (1967), Scientific Realism. The Search for Systems, Springer, New York– –Heidelberg–Berlin.

O PIĘKNIE NAUK EKONOMICZNYCH

163

Bunge M. (1979), A System Concept of Society. Beyond Individualism and Holism, “Theory and Decision”, 10(1–4), ss. 13–30. CK (2007), Uchwała Prezydium Centralnej Komisji ds. Stopni i Tytułów z dnia 29 maja 2007 roku przyjmująca dokument zatytułowany „Zakres merytoryczny dyscypliny naukowej «towaroznawstwo» w ramach dziedziny «nauki ekonomiczne»” (maszynopis), Centralna Komisja ds. Stopni i Tytułów, Warszawa. CK (2010), Zakres dyscyplin „ekonomia”, „finanse” i „nauki o zarządzaniu” w ramach dziedziny „nauki ekonomiczne” (maszynopis), Centralna Komisja ds. Stopni i Tytułów, Warszawa. Csaba L. (2016), Nieortodoksja, heterodoksja a ekonomia globalna, w: Ekonomia przyszłości. Wokół nowego pragmatyzmu Grzegorza W. Kołodko, M. Bałtowski (red.), Wydawnictwo Naukowe PWN, Warszawa. Csaba L. (2017), Comparative Economics and the Mainstream, “Economics and Business Review”, 3(17), ss. 32–51. Eco U. (2007), Historia brzydoty, Dom Wydawniczy Rebis, Poznań. Eco U. (2007), Historia piękna, Dom Wydawniczy Rebis, Poznań. Eco U. (2010), Historia piękna, Dom Wydawniczy Rebis, Poznań. Fiedor B. (2013), Uwagi o potrzebie równowagi metodologicznej w ekonomii, „Studia Ekonomiczne”, 1(76), ss. 101–118. Fiedor B. (2016), Ekonomia a piękno – kilka refleksji, w: Ekonomia i środowisko. Księga jubileuszowa Profesora Bogusława Fiedora, S. Czaja, A. Graczyk (red.), Wydawnictwo Uniwersytetu Ekonomicznego we Wrocławiu, Wrocław. Fine B. (2000), Economics Imperialism and Intellectual Progress: The Present as History of Economic Thought? “History of Economics Review”, 32, ss. 10–35. Flejterski S. (2007), Metodologia finansów, Wydawnictwo Naukowe PWN, Warszawa. Galbács P. (2017a), Editorial. The Methodology of Contemporary Macroeconomics, “Economics and Business Review”, 17(3), ss. 3–6. Galbács P. (2017b), Methodology…?! Why? Some Methodological Aspects of the Controversy Between Mainstream Economics and Institutionalism, “Economics and Business Review”, 17(3), ss. 112–134. Giza W. (2016). Nowy pragmatyzm Grzegorza W. Kołodki – o poszukiwaniach ekonomii przyszłości, w: Ekonomia przyszłości. Wokół nowego pragmatyzmu Grzegorza W. Kołodko, M. Bałtowski (red.), Wydawnictwo Naukowe PWN, Warszawa. Gorynia M. (1993), Poziomy analizy w naukach ekonomicznych, „Ekonomista”, 4, ss. 501–506. Gorynia M. (1998), Zachowania przedsiębiorstw w okresie transformacji. Mikroekonomia przejścia, Wydawnictwo AE w Poznaniu, Poznań. Gorynia M. (1999a), Przedsiębiorstwo w różnych ujęciach teoretycznych, „Ekonomista”, 4, ss. 531–546. Gorynia M. (1999b), Przedsiębiorstwo w nowej ekonomii instytucjonalnej, „Ekonomista”, 6, ss. 777–790. Gorynia M. (2000), Teoria przedsiębiorstwa w okresie transformacji, „Ekonomista”, 2, ss. 181–193. Gorynia M., Jankowska B., Owczarzak R. (2005), Zarządzanie strategiczne jako próba syntezy teorii przedsiębiorstwa, „Ekonomista”, 5, ss. 595–619.

164

Marian Gorynia

Gorynia M., Kowalski T. (2013), Nauki ekonomiczne i ich klasyfikacja a wyzwania współczesnej gospodarki, „Ekonomista”, 4, ss. 457–474. Gorynia, M. (2016), Nauki ekonomiczne a postulat interdyscyplinarności, w: Ekonomia i środowisko, S. Czaja, A. Graczyk (red.), Wydawnictwo Uniwersytetu Ekonomicz nego we Wrocławiu, Wrocław. Gorynia M. (2018), Współczesne nauki ekonomiczne – tożsamość, ewolucja, klasyfikacje, Referat na konferencję pt. „Ewolucja nauk ekonomicznych. Jedność a różnorodność, relacje do innych nauk, problemy klasyfikacyjne”, Komitet Nauk Ekonomicznych PAN, Polskie Towarzystwo Ekonomiczne, Warszawa. Griffin R.W. (1999), Management, Houghton Mifflin Company, New York. Hammond D. (2003), The Science of Synthesis. Exploring the Social Implications of General Systems Theory, University Press of Colorado, Colorado. Hardt Ł. (2013), Studia z realistycznej filozofii ekonomii, C.H. Beck, Warszawa. Heller M. (2010), Dowód na istnienie Boga. Z ks. Prof. Michałem Hellerem rozmawiają Jerzy Baczyński i Adam Szostkiewicz, „Polityka – Niezbędnik Inteli genta” (6). Koenig G. (1993), Théories de la Firme, Economica, Paris. Kołodko G. (2014), Nowy Pragmatyzm, czyli ekonomia i polityka dla przyszłości, „Ekonomista”, 2, ss. 161–180. Koontz H. (1961), The Management Theory Jungle, “Journal of the Academy of Management”, 4(3), ss. 174–188. Krzakiewicz K., Cyfert S. (2013), Teoretyczne problemy zarządzania organizacjami. Wydawnictwo UEP, Poznań. Kuhn T. (1968), Struktura rewolucji naukowych, PWN, Warszawa. Landais B. (2017), Pour une science économique efficace et une recherche équilibrée, “Revue Internationale des Economistes de Langue Française”, 2(2), ss. 7–19. Metaekonomia. Zagadnienia z filozofii ekonomii (2016)., M. Gorazda, Ł. Hardt, T. Kwarciński (red.), Copernicus Center Press, Kraków. Murak J. (2014), Imperializm ekonomii, „Ekonomia XXI wieku”, 3(3), ss. 113–126. Niekipielow A. (2016), Nowy pragmatyzm Grzegorza W. Kołodki – alternatywa czy uzupełnienie czystej teorii ekonomii? w: Ekonomia przyszłości. Wokół nowego pragmatyzmu Grzegorza W. Kołodko, M. Bałtowski (red.), Wydawnictwo Naukowe PWN, Warszawa. Nietzsche F.W. (2011), Narodziny tragedii, Wydawnictwo Uniwersytetu Jagiellońskiego, Kraków. Noga A. (2009), Teorie przedsiębiorstw, PWE, Warszawa. Nowa ekonomia instytucjonalna a nauki o zarządzaniu (2016), S. Rudolf (red.), Wyższa Szkoła Bankowa w Gdańsku, Gdańsk. Platon (2007), Dialogi, Unia Wydawnicza „Verum”, Warszawa. Rodrik D. (2015), Economics Rules: Why Economics Works, When it Fails, and How to Tell the Difference, Oxford University Press, Oxford. Scheuer B. (2016), Konstruktywizm w ekonomii, w: Metaekonomia. Zagadnienia z filozofii ekonomii, M. Gorazda, Ł. Hardt, T. Kwarciński (red.), Copernicus Center Press, Kraków. Słownik historii myśli ekonomicznej, (2004), E. Kundera (red.), Oficyna Ekonomiczna, Kraków.

O PIĘKNIE NAUK EKONOMICZNYCH

165

Sztompka P. (1985), Wstęp, w: J. Turner, Struktura teorii socjologicznej, przekł. J. Szmatka. PWN, Warszawa. Szyszka A. (2009), Finanse behawioralne. Nowe podejście do inwestowania na rynku kapitałowym, Wydawnictwo UEP, Poznań. Tatarkiewicz W. (1970), Wielkość i upadek pojęcia piękna, „Studia Filozoficzne”, 1, ss. 11–16. The Common Sense of Political Economy, and Selected Papers on Economic Theory (1933), L. Robbins (ed.), Philip Wicksteed, G. Routledge and Sons, London. Tischner J. (2000), Myśli wyszukane, Wydawnictwo Znak, Kraków. Wilkin J. (2009), Czy ekonomia może być piękna? Rozważania o przedmiocie i metodzie ekonomii, „Ekonomista”, 3, ss. 295–313. Wilkin J. (2016), Instytucjonalne i kulturowe podstawy gospodarowania. Humanistyczna perspektywa ekonomii, Wydawnictwo Naukowe SCHOLAR, Warszawa. Współczesne teorie organizacji (1983), A.K. Koźmiński (red.), PWN, Warszawa. Współczesne koncepcje zarządzania (1987), A.K. Koźmiński (red.), PWN, Warszawa. Wojtyna A. (2017), Czy w wyniku obecnego kryzysu światowego zmienił się zakres i sposób uprawiania ekonomii? Prezentacja na posiedzeniu Komitetu Nauk Ekonomicznych PAN pt. „Klasyfikacja nauk ekonomicznych – dywergencja czy konwergencja?”, Warszawa, 13 marca. Zarządzanie, organizacje i organizowanie. Przegląd perspektyw teoretycznych (2016), K. Klincewicz (red.), Wydawnictwo Naukowe Wydziału Zarządzania Uniwersytetu Warszawskiego, Warszawa. Zingales L. (2000), In Search of New Foundations, “Journal of Finance”, 55(4), ss. 1623–1653.

ABOUT THE BEAUTY OF ECONOMIC SCIENCES ABSTRACT The aim of the article is to identify the manifestations (aspects) of the beauty of economic sciences. By economic sciences, in this text, we mean four disciplines of Polish classification of disciplines which was in force until September 2018. In the field of economic sciences this classification included (in alphabetical order): economics, finance, commodity science and management sciences. The basis for the conceptual search for the beauty of economic sciences is to be found in philosophy, and more specifically in its fragment, which constitutes aesthetics. The text essentially uses two ways of comprehending beauty: Apollonian beauty and Dionysian beauty, the first way being treated as traditional and the other as an alternative. Both of these ways were used to identify the manifestations of the beauty of economic sciences due to the subject of the study and due to the applied research method.

166

Marian Gorynia

The basic research method used in the preparation of this study was the analysis of the literature on the subject. Keywords: economic sciences, beauty as aÂ philosophical category, beauty of the subject of economic sciences, beauty of the method of economic sciences. JEL Classification: A10, A11, A12, D00, E00, G00, M00

144

Barbara Błaszczyk

Instytut Nauk Ekonomicznych Polskiej Akademii Nauk Pałac Staszica ul. Nowy Świat 72 00-330 Warszawa www.inepan.pl studia.ekonomiczne@inepan.waw.pl Cena 50,00 zł (w tym 5% VAT) Nakład 200 egz.