144 THE CHANGING WEALTH OF NATIONS
very close to one. Still, to ensure comparability between both investment series, the investment estimates derived from the accounting identity were used only if the country-specific median of these ratios, for the period 1960–2006, was close to one. For the rest of the countries for which complete investment series are still not available, data on gross fixed capital formation are used for the missing years. Complete investment series for the 19 years preceding 1995, calculated using the three methods listed above, were available for 107 countries, while complete data series for 125 countries were used to calculate produced capital for the years 2000 and 2005. For 20 countries for the year 1995 and two countries for 2000, all still missing complete investment series, produced capital is estimated after adjusting the values obtained using a lifetime assumption of 14–19 years (as the case may be). The adjustment made is that values obtained using less than 20 years are multiplied with the median of the ratio of capital obtained from 20 years to that obtained from less than 20 years (over 1960–2006). For the remaining countries, we tried to overcome the data limitations by using a quite conservative approach. We extended the investment series by regressing the logarithm of the investment output ratio on time, as did Larson et al. (2000). However, we did not extrapolate output, limiting the extension of the investment series to cases in which a corresponding output observation was available. In particular, the 20-year service lifetime assumption was used to estimate capital stocks from investment series predicted from a regression of the ratio of the log of investment to GDP on time. Produced capital estimates for 6, 27, and 34 additional countries were obtained using this method for the years 1995, 2000, and 2005, respectively.
Urban Land In the calculation of the value of a country’s physical capital stock, the final physical capital estimates include the value of structures, machinery, and equipment, since the value of the stocks is derived (using the Perpetual Inventory Method) from gross capital formation data that account for these elements. In the investment figures, however, only land improvements are captured. Thus, our final capital estimates do not entirely reflect the value of urban land. Drawing on Kunte et al. (1998), we valued urban land as a fixed proportion of the value of physical capital. Ideally, this proportion would be country-specific. In practice, detailed national balance sheet information with which to compute these ratios was not available. Thus, like Kunte et al. (1998), we used a constant proportion equal to 24 percent: 5 Ut 0.24Kt.
(A.3)