ESTIMATING THE VALUE OF THE LEGACY ASSETS
WHAT WE DO AND DON’T KNOW ABOUT THE ULTIMATE COST TO CREDIT UNIONS OF THE CORPORATE STABILIZATION FUND
A White Paper Prepared by the Research and Policy Department of the Credit Union National Association
Intended Solely for the Internal Use of CUNA Members.
TABLE OF CONTENTS
Page Executive Summary
The genesis of the problem
What we know for sure is quite limited
Rigorous analysis narrows the range
Corporate credit unions
MEMBERS Capital Advisors
Speculation based on rigorous analysis
MCA: US Central and WesCorp Credit Union Structured Securities Impairment Analysis
Executive Summary. The financial crisis of 2007 to 2009 wreaked havoc on portions of the portfolios of a number of corporate credit unions. Most of the troubled securities were held in five corporates, which have since been conserved. Their troubled assets are being managed by NCUA under the Corporate Stabilization Fund. With this Fund, credit unions will eventually be required to pay, through assessments to NCUA, for the actual losses that are realized from these portfolios. Almost $7 billion in losses have already been covered by previous Corporate Stabilization assessments by credit unions ($1.3 billion) and depleted capital in the five conserved corporates ($5.6 billion). The remaining losses will have to be paid by credit unions in assessments over the remaining eleven years of the Corporate Stabilization Fund. The crucial question then becomes, how large are the remaining losses to be paid? This White Paper collects and describes the publicly available information on the “legacy assets” and provides an opinion on the range into which the losses are likely to fall. Specifically the White Paper: • • • • •
Generally describes the nature and amount of the legacy asset portfolios of the five conserved corporates. Describes how three concepts of valuation of a portfolio of troubled securities might vary. Explains that what we know for sure is very limited. Explains at a high level how rigorous portfolio valuation models work, and reports on the results of publicly available valuations using such valuation models. Reports on an important analysis of a subset of the legacy assets (those acquired by US Central and WesCorp) by MEMBERS Capital Advisors, a subsidiary of the CUNA Mutual Group. The MCA analysis is being released with this White Paper for the first time.
With the announcement of the Legacy Assets plan last September, NCUA released a range of total loss estimates of $14 to $16 billion (midpoint $15 billion). Subtracting the $7 billion already paid, that would leave $7 to $9 billion (midpoint $8 billion) to be collected in credit union assessments. Based on our analysis of all information available thus far, we believe the ultimate losses will more likely be closer to $12 billion than $15 billion. That would leave the total assessments for credit unions to pay closer to $5 billion than $8 billion. Of course, no one yet knows what the ultimate losses will be. That depends on the pace of the economic recovery, particularly the outlook for unemployment and home prices. More important than any specific valuation of the legacy assets is a basic understanding by credit unions of the issue. That is what this White Paper seeks to address. The publicly available information on the legacy assets will likely increase in the future, making it easier for credit unions to track and understand this important process.
Introduction. In September 2009, NCUA conserved three corporate credit unions, bringing to total to five, and announced a plan to deal with the “legacy” assets of the five conserved corporates. The plan establishes a mechanism whereby credit unions will pay the actual losses on these assets, with the payments to be spread over the next eleven years. Because most of the losses on the legacy assets will occur in the future, they are unfortunately 2
unknown and unknowable. However, much has been said about what the level of those losses will be. This White Paper describes estimates that others have presented on the potential losses on the legacy assets, and tries to make some sense out of the range of estimates that have been provided. The purpose of the White Paper is to help credit unions understand the factors that will drive the assessments that will be necessary to pay the Corporate Stabilization Fund. The legacy assets plan entails placing the assets in a trust and issuing NCUA Guaranteed Notes to fund them until they either amortize and mature, or default. The Notes are designed to have repayment schedules that approximate the cash flows of the assets in the trust. At the time of the announcement, the face value of the legacy assets was roughly $50 billion, and NCUA planned to issue about $35 billion in notes. The $15 billion difference approximates the expected losses on the portfolios1. This is the estimated amount that credit unions will eventually have to pay. A little under half ($6.9 billion) has already been paid in the form of depleted capital in the five conserved corporates ($5.6 billion) and Corporate Stabilization assessments paid by credit union in 2009 and 2010 (a total of $1.3 billion.) Using the $15 billion eventual loss estimate, that leaves $8.1 billion to be paid over the remaining 11 year term of the Corporate Stabilization Fund. NCUA adopted this plan to avoid having to sell the securities at market values that were substantially less than their expected, realizable values. Because the Notes are being issued with the backing of the US Treasury, their interest rates will be fairly low, especially when compared to the yields on the underlying legacy assets. Under this approach, the ultimate result for credit unions is that they will have to pay whatever the actual losses on the securities end up being. Based on NCUA’s latest estimate of roughly $15 billion in total losses, credit unions would have to pay an average of around 7.5 basis points of insured shares over the next eleven years. However, if the losses turn out to be more than $15 billion, assessments will have to be greater. On the other hand, if the losses end up being less than $15 billion, assessments will be reduced accordingly. Under NCUA’s plan, the die is cast, and the ultimate cost to credit unions now depends almost exclusively on the actual amount of future losses on the securities. The $15 billion figure that approximates NCUA’s estimate of losses as of last September is not a certain, known amount. In fact, it is just one of a number of estimates of the value of the portfolios that has been performed. The remainder of this White Paper attempts to describe the publicly available information on the securities and their likely value, to help credit unions form opinions about the future costs of the corporate stabilization.
The genesis of the problem. Over several years leading up to 2007, a number of corporate credit unions (primarily the five that have since been conserved)2 accumulated a large amount of “structured” securities. In general, a structured security is one whose value depends in some way on other securities or indices. There are many possible types of structured securities, but those purchased by corporates were “asset backed.” They were comprised of pieces of other loans or bonds, primarily private-label residential mortgage backed securities (RMBS), sometimes known as non-agency RMBS. These RMBS included first and second liens, with loans that ranged from prime 1
When the legacy asset plan was announced last September, NCUA projected the range of losses to be between $13.9 billion and $16.1 billion. We use the midpoint, or $15 billion, as the expected loss amount. 2 US Central and WesCorp were conserved in March of 2009. Southwest, Members United and Constitution were conserved in September 2010.
through Alt-A to subprime.3 The portfolios also included other asset-backed securities such as commercial mortgage backed securities (CMBS), securities backed by other consumer loans, and collateralized debt obligations (CDOs) made up of pieces of other assets, many of which were structured securities themselves. Most of the structured securities were bought in 2005, 2006, and 2007. At their peak they totaled about $65 billion out of total corporate assets of around $150 billion.4 The vast majority of the securities appear to have been first lien RMBS, mostly variants of Alt-A loans, and Option ARMs. Toward the end of 2007 the markets for these securities began to deteriorate, and gaps opened up between the book values of the securities, their market values, and their probable values if held to maturity (their “realizable” values). As a group, these securities have variously been referred to with such names as “legacy assets” or “toxic assets” or “troubled securities.” Differences between “market” and “realizable” values exposed a primary problem with these sorts of securities, and also explain why it is so difficult to determine the actual losses that will eventually be incurred as a result of holding them. Simply put, the value of any asset-backed security ultimately depends on the value of the underlying assets that back it. There are, however, two possible measures of that value. One is the simple market value, or what the security would fetch in an open market. The second is the realizable or economic value, or the amount the holder of the security can reasonably expect to receive over the security’s life as it amortizes or matures. More specifically, the realizable value is the present value of those expected cash flows. In theory, if the underlying assets were all “good,” i.e., if the originally planned cash flows were fully expected to occur, and market participants generally knew this to be the case, the present value of those cash flows would be very close to the market value of the security. In such a situation, markets would function smoothly (there would be no market “dislocations”) and the market and realizable values would differ from the original face value only if interest rates had changed from their issue levels. In a hypothetical case where some defaults are expected on the underlying assets, but where all potential investors have complete and free access to information about the likely amount of those defaults and the resulting losses, the market value would still be very close to the present value of expected future cash flows, i.e., the realizable value. In this case, however the market and realizable values would be lower than the face value due to the expected losses. However, if a security or many like it is in doubt, and detailed information about the underlying assets is expensive or hard to come by, buyers would understandably be nervous about buying it, and the market would not function smoothly. The market value would fall to something below the realizable value because of the uncertainty on the part of investors. One of the problems with asset-backed securities is that they are not standard; they are all different; each is made up specific and different underlying assets. For example, to accurately estimate the future cash flows for any given residential mortgage backed security, one would need to know specific loan-level information about every loan in the security, and each mortgage backed security is 3
Prime loans have full documentation and are made to borrowers with strong credit histories. Subprime loans were made to borrowers with tarnished credit histories. Alt-A loans were typically made without the complete documentation of prime loans, or to borrowers with slight credit history blemishes. 4 The $150 billion total assets included investments by retail corporate credit unions in US Central. Netting out these deposits, the total assets of corporates were around $110 billion, meaning the troubled securities amounted to about 60% of external investments by corporates.
made up of pieces of different underlying loans. Such information is difficult or expensive to come by on the vast quantity of such bonds. Therefore, since buyers do not have low-cost access to such information, the discounts for uncertainty imbedded in market values will be quite high. Thus there can be significant differences between face value, likely realizable values, and market values. This is why declines in market values tend to overstate the actual deterioration in a non-standard asset-backed security’s realizable value. This is not to say that the losses in the corporates’ portfolios are not very large. However, they were certainly not as huge as suggested by declines in market values. For example, had the entire portfolios of troubled securities held by the corporates been liquidated late in 2008, the market losses would likely have been in excess of $30 billion. As we will see below, most loss estimates based on evaluating the present value of future cash flows range from somewhere between $9 billion and $16 billion, depending on the future outlook for the economy, interest rates, housing prices and mortgage markets. These are of course huge losses, but only between a third and a half of the losses suggested by market value declines.
What we know for sure is quite limited. There will be substantial losses. Not all the original roughly $65 billion in troubled securities will fully pay off. So far, actual incurred losses on the securities themselves have only amounted to a little over $2 billion. Without doubt, that number will rise. The current face value of the troubled securities is a little shy of $50 billion (a bit more than $15 billion of the original $65 billion has amortized). So, all we can really say absolutely for sure is that the final costs will be between $2 and $50 billion. Of course, we know the losses will be considerably greater than $2 billion and considerably less than $50 billion. We turn now to a summary of the rigorous analysis that has attempted to determine a narrower range for the ultimate losses.
Rigorous analysis narrows the range. Moving beyond what we know for sure, there have been a number of attempts at rigorous analysis of the troubled securities held by the corporates. By “rigorous” we mean valuations of the portfolios, or portions of the portfolios, using detailed models of the likelihood of future payment based on analysis of the underlying assets behind the structured securities. For instance, for residential mortgage backed securities, this approach typically gathers information on the specific loans in a security, including information on local real estate markets, borrower characteristics and collateral, and then models or estimates future payment probabilities under a variety of economic scenarios regarding unemployment rates, interest rates and home price movements. This involves estimating a probability of default, and an estimated loss in event of default, for each loan. The analysis also takes into account the structures of the securities, i.e., if one or more of the underlying loans defaults, where does the security stand in line compared to all other structured securities that may include pieces of that loan? The more “senior” a structured security, the less will it be affected by defaulting loans in the pool. Finally, the value of a security in this approach is the present value of the predicted future cash flows. That is, how much would one have to invest today to generate the predicted cash flows using a relevant “discount” or interest rate? These rigorous valuation approaches typically produce a range of estimated values of a portfolio, rather than a single number. There are two reasons for this. Most important is the effect of evaluating different economic scenarios. The performance of a pool of mortgages in a period of falling unemployment and rising home 5
values (optimistic case) will naturally be much better than in a period of rising unemployment and falling home prices (pessimistic case). Valuation attempts therefore typically produce estimates under a variety of possible scenarios, with “optimistic” and “pessimistic” cases bracketing the “most likely” or “base” scenario. Second, within a given scenario, valuation estimates are usually presented as a range because even if we knew with certainty the future course of the economy (unemployment rates, home prices, interest rates, etc.), the models would only give a reasonably accurate but not exact prediction of the performance of the underlying loans. In the current context, for most analysts an “optimistic” scenario would probably approximate a “V” shaped recovery, i.e., a rapid and strong recovery at roughly the speed and strength of the downturn in 2008 and 2009. Such an optimistic scenario would also likely envision rising home prices, although certainly not at the rate of home price increases from 2000 to 2006. A “most likely” scenario would refer to a “U” or “bowl” shaped recovery, with the modest but positive growth and falling unemployment, taking three to four years for the economy to fully recover to its pre-recession level, with flat or modest home price increases. A pessimistic scenario would likely refer to a double-dip recession, with declines in GDP and further increases in unemployment, and falling home prices. We are aware of three groups of rigorous analysis of the portfolios: for NCUA, for the corporates themselves, and by CUNA Mutual Group’s MEMBERS Capital Advisors. NCUA. In the process of creating the Temporary Corporate Credit Union Share Guarantee Program and the Temporary Corporate Credit Union Stabilization Fund (henceforth, Corporate Stabilization Fund), NCUA required valuation estimates of the troubled corporate portfolios for accounting and other reasons. The Agency contracted with a number of outside vendors to value the corporate portfolios over the past two years. In March of 2009, NCUA announced loss estimates based on its analysis of valuation estimates produced by PIMCO. Specifically, according to NCUA Letter 09-CU-06 “The results confirm NCUA’s analysis that potential credit losses on all securities could approach upwards of $16 billion, with a most reasonable estimate in the current environment of $10.8 billion.” Based on these estimates, NCUA announced that the reserve for the Temporary Guarantee Program would need to be $5.9 billion. The Agency was not particularly forthcoming with details behind this information, but one assumes the $5.9 billion figure was an actuarial estimate based on the probability of failure of each corporate and estimates of loss in the event of failure. One further assumes that the two most important pieces of data in the determination of the $5.9 billion reserve figure were the “most likely” valuation of the portfolios (the above mentioned $10.8 billion loss estimate) and the available capital in the corporates in question to cover losses before the Temporary Corporate Guarantee would have to step in. That capital totaled around $6 billion. The Agency reported the “most likely” loss estimate as $10.8 billion, and the upper bound of the estimates to be $16 billion, which one assumes is the top of the range for the “pessimistic” case. However, they did not reveal the results for an “optimistic” case. By symmetry (if the $10.8 billion most likely figure were the midpoint between optimistic and pessimistic) we might assume the optimistic estimate to be around $6 billion. However, these things are not always symmetrical, so we’ll be conservative and assume that the “optimistic” case was no higher than $9 billion. 6
In summary, inferring as best we can from the information revealed by NCUA in March of 2009, at that time, the estimate for losses in the corporate portfolios in the “most likely” scenario was around $11 billion (rounding). The “pessimistic” case was $16 billion, and the “optimistic” case was at worst (no more than) $9 billion, although probably somewhat less than that. The next mention by NCUA of any data related to estimated losses in the corporate portfolios was in May of 2010. In addition to the previous work done by PIMCO, some of the corporates had retained Clayton Holdings to evaluate some of the securities. In a memo to the NCUA Board for its May 20 Board meeting, Deputy Executive Director Larry Fazio wrote that the Stabilization Reserve now stood at $6.4 billion. That was $0.5 billion greater than the $5.9 billion Insurance Reserve established in March of 2009. Because the first was an “Insurance” reserve and the second a “Stabilization” reserve, the difference may have been due to factors other than changes in the valuation of the troubled assets. We cannot be sure because in May of 2010, the Agency did not provide any updates on the previously announced total loss estimate of $10.8 billion. Again inferring from the limited information available, the $6.4 billion Stabilization Reserve would still put the total portfolio loss estimate at an amount that rounds to $11 billion in the “most likely” scenario.5 Finally, when NCUA announced the legacy assets plan and placed three more corporates into conservatorship in September of 2010, they revealed a substantial amount of useful and relevant information about the value of the portfolios.6 A new valuation of the portfolios of the five conserved corporates was performed by Barclays Capital. The range of projected losses produced by Barclays was $13.9 billion to $16.1 billion (midpoint: $15 billion).7 There are two striking differences between these estimates and the range previously reported in March of 2009, which was from $10.8 billion (most likely) to $16 billion (pessimistic) (and our inferred optimistic estimate of around $9 billion). First, the estimates NCUA reported based on Barclays’ analysis are in a much narrower range ($14 billion to $16 billion). Second, and more significant, the $15 billion midpoint is much higher than the previous most likely estimate (around $11 billion). There are two possible interpretations of this set of data: •
With the passage of time, uncertainty about the future estimates has been substantially reduced (the range has narrowed), and the outlook for the securities has worsened dramatically (the midpoint has risen), or The purposes of the previous and latest estimates are different, and therefore are not directly comparable. The original estimates were for accounting purposes, and thus represented “most likely” outcomes. The more recent estimates were for underwriting purposes, and therefore are more likely to represent pessimistic outcomes.
Regarding the first interpretation, the passage of time will indeed narrow the range of estimates of future events, but likely not from a range as wide as $7 billion ($9 billion to $16 billion) to one as narrow as $2.2 5
If the expected losses on the portfolios had indeed increased by the full $0.5 billion, the new estimate would be $11.3 billion. 6 This information is available from the link: www.ncua.gov/Resources/CorporateCU/CSRMain.aspx. 7 This range includes some fees and other expenses of the stabilization fund, but the vast majority of the amounts represent expected bond defaults.
billion ($13.9 billion to $16.1 billion) in just a year. Second, most other indications over the period from 2009 to 2010 suggest that the market for mortgage backed securities had at least stabilized. There certainly is no evidence that the RMBS market had significantly deteriorated during that period. So, a dramatic increase in the “most likely” estimates of losses in the portfolios over the period is very unlikely. Considering the second interpretation, the purpose of the valuation for the legacy assets plan was to determine how much the Stabilization Fund could borrow in the form of NCUA Guaranteed Notes on a basis that matched the expected cash flows from the troubled assets. At the time the plan was drawn up, the remaining face value of the troubled securities was about $50 billion. There are two possible outcomes for that $50 billion. Some of it will amortize as originally expected, and some of it will not be collected, i.e., there will be losses. The Stabilization Fund was to issue Notes that would essentially be collateralized by the troubled securities. The goal was for the proceeds from the troubled assets (the good payments) to cover the repayment of the Notes. The balance, the losses, is what credit unions will have to pay in the form of the depleted capital of corporate credit unions and Corporate Stabilization assessments. An extremely important consideration in all of this is the Treasury’s guarantee of the NCUA Guaranteed Notes. The Agency has received permission to state that the Notes have the full faith and credit of the US Government. Treasury has essentially co-signed the Notes. As a result, NCUA is issuing the Notes through Barclays at attractive rates, lowering the ultimate cost of the Stabilization to credit unions. Recall that the actual losses on the underlying securities are unknown. They will depend on the performance of the underlying loans which depends on the future course of the economy and housing markets. Therefore, the NCUA can’t be sure as to just how much of the $50 billion will be repaid, and hence should be funded with Notes, and how much will result in losses that must be paid by credit unions through assessments. In these circumstances, NCUA may well have used a loss estimate for underwriting purposes that would leave little doubt that the proceeds from the legacy assets would be sufficient to repay the Notes. Erring on the pessimistic side would simply require refunding a portion of credit union assessments (or ending the assessments early.) Erring on the optimistic side would require having to increase assessments in the future to cover a shortage of funds necessary to repay the Notes backed by the US Treasury. It is likely then that an underwriting estimate would lean more to the pessimistic side than would a “most likely” estimate for accounting purposes. Of course, we cannot be certain about any of this without access to the full modeling and assumptions that Barclays used in the evaluation of the portfolios, which NCUA has not released. However, it is consistent with the other estimates described in this White Paper to assume that the Barclays range would lie toward the pessimistic end of the spectrum. Corporate Credit Unions. Financial institutions with troubled assets are required by accounting rules to generate estimates of future expected losses on those assets for financial statement preparation purposes. Consider a troubled security with an uncertain future. The book value of that security (the price the institution originally paid) likely overstates its “actual” value because it is “troubled.” As previously discussed, reductions in market values can overstate likely eventual losses. Therefore, rather than reporting market values, financial institutions report values net of “other than temporary impairments.” These so-called OTTI amounts represent the difference between the book value of a security and the expected “realizable” value of the 8
security. The expected realizable value of the security is typically estimated using standard security valuation models as described above. In other words, the OTTI taken on a security is an approximation for the amount of expected losses a security will suffer if held through amortization or until maturity. When NCUA released the details of the legacy asset plan in September of 2010, it reported that a total of $11.7 billion of OTTI charges had been taken by the five conserved corporates as of June 30, 2010 as follows: WesCorp, $6.9 billion; US Central, $3.6 billion; Members United, $600 million; Southwest, $496 million; and Constitution, $122 million. The two largest of these corporates, US Central and WesCorp had been under conservatorship and hence NCUA management for over a year. The other three were not conserved until September of 2010, although they had been under substantially increased NCUA scrutiny since early 2009. Although most security valuation models produce ranges of possible values based on different scenarios, for financial statement purposes the amount of an OTTI charge must be a specific number, and is usually based on the application of a “most likely” scenario. The determination of this number can be topic of considerable debate between a financial institution and its auditing firm. Simply put, a troubled financial institution has an incentive to estimate potential losses on the low side so as to appear as healthy as possible. Similarly an outside accounting firm has an incentive to estimate potential losses on the high side just because accounting tends to be more comfortable with conservative assumptions and estimates. Assuming for the moment the debate between the institution and outside auditors produces a “reasonable” estimate of expected losses (one that a disinterested analyst would not violently disagree with), there is another feature of OTTI accounting that tends to produce loss estimates that through time will err on the high side. This is sometimes referred to as the “ratchet” effect. In any accounting period during which a security becomes more impaired (new information becomes available, increasing the expected future loss on the security) the amount of that impairment must be charged, i.e., the value of the security must be written down by the amount of the increase in expected loss. However, if in a later accounting period new information becomes available that the expected loss on a security is less than previously estimated (for instance, if delinquency rates on underlying loans improve, or home prices rise) the charge cannot be reversed. The security would then continue to be held at less than its expected future value. In this case, the previously recorded losses would only be recovered as the security amortizes, matures or is sold at a price above the impaired amount. Therefore, the total amount of OTTI charges taken by a financial institution represents the sum of all periodic impairments to the securities it owns, ignoring any subsequent recoveries in expected realizable values.8 This means that through time, an estimate of losses measured by OTTI charges will not necessarily be the same as the latest estimate of losses. It is only when a group of securities are holding stable or continually getting worse that OTTI charges will be the same as the latest estimate of portfolio losses. If any securities in a portfolio have improved, OTTI losses will overstate the most current estimate of losses. This will be the case even if exactly the same valuation technique is used to estimate the losses under the two approaches. Hence the term “ratchet effect.” Therefore, unless their auditing firms and the NCUA were allowing them to grossly understate losses, the $11.7 billion sum of OTTI charges at the five corporates as of last June likely overstates somewhat the “most likely” estimate of losses as of that time. 8
Further evidence that accounting tends to be more comfortable with conservative assumptions and estimates.
MEMBERS Capital Advisors. In 2010, MEMBERS Capital Advisors (MCA), a subsidiary of the CUNA Mutual Group, performed a valuation on the structured securities of two of the five corporates now under conservatorship: WesCorp and US Central. A copy of the report on their analysis, which was completed in December, 2010 is attached. The MCA report will be valuable for credit unions to better understand issues surrounding the legacy assets. It should be read thoroughly before continuing with this White Paper because it: • • • •
provides important details on the composition of the portfolios of structured securities at US Central and WesCorp, both in terms of type and vintage of the securities (Pages 13 – 15). provides general and specific explanations of how the estimates were generated (Pages 9 and 10). provides a clear and concise description of the four scenarios that drove the analysis (Page 8). and most important, provides estimates of the values of the US Central and WesCorp portfolios under the four scenarios (Pages 7, 8, and 12).
The MCA analysis was performed as of June 2010, at which time the face value of the troubled asset portfolios at the two corporates was $34.5 billion. That represents approximately 70% of the total amount of legacy assets (about $50 billion) at the five combined corporates covered under the Corporate Stabilization Fund. MCA first estimated losses under three scenarios: optimistic (recovery), base (extended slow recovery), and pessimistic (recession). Using these three cases, MCA estimated total losses on the structured securities of the two corporates in the range of $7.3 billion to $10.3 billion. They also provided estimates for a fourth scenario, very pessimistic (severe recession), which ranged from $12.3 billion to $13.1 billion. Thus, over all four of the scenarios, MCA’s loss estimates ranged from a low of $7.3 billion to a high of $13.1 billion. The reader should refer to Page 8 of the MCA analysis for a complete description of the four scenarios, and the estimated loss ranges for each scenario. The table on Page 8 provides excellent insight into the factors that will influence the future performance of the US Central and WesCorp portfolios. Most important to the results are the future paths of unemployment (which drives the expected loan default rate) and home price appreciation (which drives losses in event of default). Three quarters after the effective date of the analysis, the actual performance of the economy appears to most closely conform to the base scenario: extended slow recovery. Actually, home prices have performed worse than expected, but unemployment has improved more than anticipated. The base scenario envisaged no change in home prices in the first year. However, from June to December, various measures of home price appreciation have registered declines of 3% to 5%. On the other hand, the base scenario assumed no change in unemployment for a year, then a gradual decline to 5.5% over five years. However, over the first 8 months, the unemployment rate has dropped from 9.5% to 8.9%. On balance, this information suggests that defaults might run a little below the base case estimates (due to falling unemployment), but losses in event of default might be slightly higher than expected (due to falling home values).
Speculation based on the rigorous analysis. The following analysis is simply speculation by CUNA’s economists and is not part of MCA’s analysis. Quoting from page 4 of MCA’s analysis: Scope limited to USC and WesCorp. We did not review holdings of other institutions within the corporate system. 10
Having said that, we will attempt to infer9 from information provided by MCA on the subset of the total legacy assets they evaluated AND the previously described estimates for the total portfolios what an overall estimate of the likely total losses on the portfolios of all five conserved corporates might be. As noted previously, the face value of the troubled securities at US Central and WesCorp amount to about 70% of the total at the five conserved corporates. The most simplistic assumption is that the average condition of the non-MCA analyzed portfolios (from the other three conserved corporates) is equivalent, on average, to the portfolios at US Central and WesCorp. In that case, the likely losses at US Central and WesCorp as estimated by MCA would represent about 70% of the losses on the total portfolio, i.e., the losses on the combined portfolio would range from a low of $10.4 billion in the optimistic case to a high of $14.7 billion in the pessimistic case. The range for the base case would be $11.6 billion to $12.2 billion, with a midpoint of $11.9 billion, which rounds to $12 billion. Throwing in the “most pessimistic” case provides a top estimate of $18.7 billion. To repeat, these are NOT MCA’s estimates of the likely losses on the total portfolios. Rather they are a simple extrapolation of what the total likely losses might be IF the portfolios of the other three conserved corporates (which MCA did NOT analyze) perform on average as the portfolios at US Central and WesCorp will. Of course, the portfolios of the other three corporates are different from those at US Central and WesCorp. They will perform differently. The question then becomes, what reasons do we have to believe that the portfolios at the other three corporates might be better than, similar to, or worse than those at US Central and WesCorp? The answer is not much, except for some very revealing information on OTTI charges at the five corporates as of last June. As of that date, NCUA reported that the total OTTI charges at just US Central and WesCorp amounted to 90% (actually 89.6%) of the total of $11.7 billion of OTTI charges at the five corporates.10 This is to be compared with the fact that US Central and WesCorp accounted for only 70% of the combined troubled portfolios of the five conserved corporates. There are three possible reasons for this discrepancy: 1. US Central and/or WesCorp grossly over-reported OTTI charges. 2. As a group, the other three conserved corporates grossly under-reported OTTI charges. 3. As a group, the portfolios of the three other conserved corporates were not as troubled as those of US Central and WesCorp combined. We’ll reject the first explanation out of hand. By the way, if it were correct, the total costs of the Corporate Stabilization Fund would likely come in lower than any of the estimates described in this White Paper. The second explanation is also unlikely, but could be true, although almost certainly not enough to account for the full difference. As of last June when the OTTI numbers were reported, NCUA was “in charge” at US Central and WesCorp, so under-reporting of OTTI charges there was very unlikely. The then managements at the other three corporates would have had some incentive to report as low an OTTI as they could reasonably 9
To infer is to conclude or surmise or make a judgment based on available, but usually incomplete information. Perhaps a better definition is to make an informed and educated guess, but a guess nonetheless. 10 As previously reported, as of June 2010, total OTTI charges were: WesCorp, $6.9 billion; US Central, $3.6 billion; Members United, $600 million; Southwest, $496 million; and Constitution, $122 million.
justify, but they had to deal with their auditing firms, and were at the time under close supervision by NCUA. We have no way of knowing how different their OTTI charges might have been, if at all, had they been under full conservatorship. But assuming for the moment that their OTTI estimates were merely half of what they should have been (that they would have been twice as much if under conservatorship), that would still leave US Central and WesCorp with 80% (instead of 90%) of the total OTTI charges, and only 70% of the portfolios.11 This lends substantial credence to the third explanation, that the troubled portfolios at the other three conserved corporates were not quite as troubled as those at US Central and WesCorp combined. This provides some justification to applying no more than the same expected loss rate at US Central and WesCorp to the portfolios of the other three corporates, suggesting an overall loss estimate of around $12 billion. In fact, based on OTTI comparisons it is most likely that the losses at the other three corporates will occur at a lower rate than at US Central and WesCorp. Thus, the total expected losses could indeed be something less than $12 billion.
Summary. Putting this all together, we have the following sets of estimates of losses under most likely scenarios on the legacy assets, which will ultimately be paid by credit unions through the Corporate Stabilization Fund. Initial NCUA estimate. About $11 billion, as reported in 2009 with the establishment of the Temporary Corporate Share Guarantee Program. Later NCUA estimate. Again about $11 billion, as reported in 2010 with the establishment for Corporate Stabilization Fund. Latest NCUA estimate. About $15 billion, as reported in September 2010 for underwriting purposes with the legacy assets plan. However, this appears to have been a “pessimistic” rather than a “most likely” estimate. MCA estimate for just US Central and WesCorp. About $8.5 billion, as reported in December 2010, estimates as of June 2010. This covers only US Central and WesCorp, which account for about 70% of the troubled assets of the five conserved corporates. CUNA’s Simple Estimate Based on MCA analysis and OTTI data. Somewhat less than $12 billion. This is based on reviewing all of the other estimates summarized above. Assuming for the moment this $12 billion figure is reasonable, that would mean that credit union future assessments to pay the Corporate Stabilization Fund would be about $3 billion less than those built into the Corporate Stabilization Fund. Those current estimates are based on the roughly $15 billion NCUA estimate of losses as of last September less the $6.9 billion credit unions have already paid in two assessments totaling 11
We are NOT suggesting that there was any actual under-reporting of OTTI charges at the three as yet not conserved corporates. We are merely showing that even if there were, it would be very unlikely to be enough to change the conclusion that the portfolios at those corporates were at least no worse than those of US Central and WesCorp.
$1.3 billion and depleted corporate capital of $5.6 billion, meaning there is $8.1 billion to be paid through the Corporate Stabilization Fund. If instead the remaining losses are closer to $12 billion than $15 billion, the remaining amount to be paid by credit unions over the next eleven years will be closer to $5 billion than $8 billion. It bears repeating that NO ONE KNOWS what the final losses will actually be. They will depend on the performance of almost $50 billion worth of securities over the next several years, which in turn will depend on the performances of the economy and housing markets, and how borrowers, lenders and servicers responds to events. Nevertheless, it is at least helpful that most of the loss estimates are clustering in the lower end of the range between $12 billion and $15 billion. If indeed the losses end up being closer to $12 billion than $15 billion, the most likely consequence for credit unions is that their assessments will end before the eleven year period of the Corporate Stabilization Fund. It is of course entirely possible, though not very likely, that the losses could be even more than $15 billion. In that event, credit unions could expect to pay Stabilization assessments for the full eleven years, with increased amounts later in the period. It is unfortunate that much of this analysis has had to resemble reading tea leaves. It would have been preferable if detailed information on the trouble portfolios had been more widely available. However, credit unions can look forward to much greater transparency in the future. Once the last NCUA Guaranteed Notes are issued, information on all of the securities will be publicly available, so that a number of analysts will be able to provide estimates of likely future losses. In addition, NCUA officials have recently reported that they intend to provide periodic updates on the performance of the portfolios once the Note issuance is complete.
Bill Hampel, Chief Economist Credit Union National Association April 2011.
MEMBERS Capital Advisors US Central and WesCorp Corporate Credit Union Structured Securities Impairment Analysis
This presentation is intended only for the exclusive benefit and use of Credit Union National Association, Inc and it’s members. This presentation was prepared to demonstrate a comparative, estimated range of the relative value of one or more specific bond portfolios in accordance with MEMBERS Capital Advisors, Inc.’s (“MCA”) ordinary, internal valuation practices. Neither this presentation nor any of its contents may be used for any other purpose without the prior written consent of MCA. Any public use of this presentation and/or its contents may also be subject to the prior approval of the National Credit Union Administration (“NCUA”). The information in this presentation reflects prevailing market conditions and MCA’s judgment as of this date, which are subject to change. In preparing this presentation, MCA has relied upon and assumed without independent verification, the accuracy and completeness of all information available from public sources and the NCUA. MCA considers the information in this presentation to be accurate, but does not represent that it is complete or should be relied upon as the sole source of information. The scope of this presentation is limited to USC and WesCorp. The analysis was performed prior to the September 2010 conservatorship of Members United, Southwest, and Constitution Corporate FCU. The analysis was also performed prior to the September 2010 announcement of the NCUA-Guaranteed Notes securitization transaction.
Contents • • • • • •
Analysis Qualifiers What is the purpose of this review? Portfolio Statistics What are the results? How were these results generated? What are the assumptions behind this analysis?
• Appendix: – Structured Portfolio Composition – Deal Vintages – OTTI Loss Estimates by Corporate CU
Analysis Qualifiers • Scope limited to USC and WesCorp. We did not review holdings of other institutions within the corporate system. • Scope limited to non-agency Residential Mortgage Backed Securities, Commercial Mortgage Backed Securities, and Collateralized Debt Obligations. • Loss estimates calculated as of 6/30/2010. • Corporate Stabilization Assessments will not directly equate to MCA’s estimates of expected investment losses, although the two are highly correlated. This is because the Stabilization Assessments will cover other factors and costs, including securities at corporates other than USC and WesCorp. • Loss estimates are based on portfolio data received from USC and WesCorp. MCA has not verified the accuracy or completeness of this information. • Losses discussed in this review are estimates of future events, some of which will not occur for several years. As such, these estimates are subject to a number of uncertainties that could materially impact actual portfolio performance. • The losses are estimated by generating expected cash flows under a number of scenarios, calculating the present value of those expected cash flows, and comparing the sum of those present values to the current face value of the securities. These estimated losses will be slightly lower than estimated principal losses because of the value of the coupon (interest) payments received prior to any potential bond default.
What is the purpose of this review? â€˘ Dual Objectives â€“ Provide independent estimates of potential loss ranges, under different economic scenarios. The potential losses are presented as ranges because of the uncertainty of future events. â€“ Provide credit unions information to better understand the key drivers of these potential losses. Although the ultimate actual losses cannot be known now, there is much that can be said about how those losses will depend on how the economy performs over the next several years.
Portfolio Statistics Combined USC & WesCorp Analysis limited to Non-Agency RMBS, CMBS, and CDO securities ($Billions)
Total Original Face Value
The total par value of the securities; if purchased at par, this would be the original cost to USC and WesCorp.
Less: Estimated Realized Prepayments
All principal payments (scheduled or prepayments) received by USC and WesCorp since issuance.
Less: Estimated Realized Losses
= Total Current Face
What the book value would be if there were NO expected future credit losses, i.e., no Other than Temporary Impairment (OTTI) charges.
Less: Estimated Book Value*
The book value of the securities as of June. This is lower than the face value because of realized expected future credit losses.
= Estimated NCUA OTTI loss**
Calculated as the difference between Face and Book Values. It is an estimate of the conservatorâ€™s assessment of future credit losses (OTTI charges).
Actual credit losses already taken on the securities.
* Data provided was as of 3/31/10. To calculate our loss estimate we incorporated the factor changes from 3/31/10 to 6/30/10. We applied the original book prices to the 6/30/10 current face amount to calculate the book value as of 6/30/10. ** Adjusted for FASB 133.
What are the results? • The range of estimated effective losses under various economic scenarios is $7.3B to $13.1B. – Most likely (base case): $8.1B to $8.5B
– Including optimistic and pessimistic scenarios: $7.3B to $10.3B
– Including a very pessimistic scenario: $7.3B to $13.1B
Four Economic Scenarios Name
Annual Home Price Appreciation
Interest Rates (across yield curve)
Estimated OTTI Loss Range ($ billions)
Yr 1: 0% Yr 2 on: 3%
Falls from current level to 5.5% over five years.
Yr 1: 0% Yr 2: 0% Yr 3 on: 3%
Flat for a year, then falls to 5.5% over five years.
Extended Slow Recovery
Yr 1: -8% Yrs 2 â€“ 3: 0% Yr 4 on: 3%
Rises to 12% in one year, then falls to 5.5% over six years.
Fall by 50 bp
$9.9 - $10.3
Yr 1: -10% Yr 2: -5% Yrs 3 â€“ 4: 0% Yr 5 on: 3%
Rises to 14% in 18 months, then falls to 5.5% over seven years.
Fall by 75 bp
$12.3 - $13.1
Rise by 50 bp
$7.3 - $7.7
$8.1 - $8.5
How were these results generated? General Overview • Residential Mortgage Based Securities (RMBS) – Analyzed each mortgage based on various factors, including location, loan terms, and borrower characteristics. – Developed assumptions on mortgage defaults and loss severity in event of default, under four economic scenarios (home price changes, employment, interest rates, etc.) – Produced estimates of the performance of the loans based on the assumptions. – Projected the performance of each security, based on the estimated cashflows from the pool of mortgages. • Commercial Mortgage Backed Securities (CMBS) – Analyzed each mortgage with data provided by third-party vendors.. – Produced estimates of the performance of the loans based on assumptions about rent growth, vacancy rates, capitalization rates, etc. • Residential Second Mortgages (2nd Lien / NIM) – Used a third party methodology to project losses on the underlying loans. • Collateralized Debt Obligations (CDO’s) – Individually reviewed collateral pool of each CDO – Loss projections vary based on scenario
How were these results generated? Technical Overview • RMBS – Use Intex to generate deal cashflows • • • •
Industry standard At deal inception, Intex models individual deal structures and payment waterfalls Requires prepayment, default, loss severity and delinquency assumptions Also requires assumptions regarding wrapped bonds (i.e. will the monoline insurers ultimately pay on their obligations)
Assumptions provided by Citibank Mortgage Research group • • •
Uses an econometric model to analyze collateral Analyzes each deal by collateral grouping Allows variability in model assumptions, with the major factors being Home Price Appreciation (HPA), Employment and Interest Rates
• CMBS – Use Intex and Trepp to generate cash flows – Analytics are performed at the individual loan level, using a proprietary model. Inputs include commercial real estate market projections from REIS, and individual loan loss projections from RealPoint. Major assumptions include market rent growth, vacancy rates, cap rates, and refinance constraints. • 2nd Lien / NIM – Utilizes Moody’s default model to project losses on the underlying collateral pool – Main drivers are deal vintage, trailing default rates, and deal type (HELOC vs. Closed-end 2nd). • CDO’s – Individually review collateral pool of each CDO – Loss projections vary based on scenario
OTTI Loss Estimates- USC & WesCorp, as of June 30, 2010 Analysis limited to Non-Agency ABS, CMBS, and CDO securities Scenario
Combined Estimated OTTI Loss Range ($B)
2.2 – 2.4
5.1 – 5.3
7.3 – 7.7
Extended Slow Recovery
2.6 – 2.8
5.5 – 5.7
8.1 – 8.5
Recession / Stress
3.7 – 3.9
6.2 – 6.4
9.9 – 10.3
5.2 – 5.6
7.1 – 7.5
12.3 – 13.1
Estimated NCUA OTTI Loss* *See slide 6 for additional details; adjusted for FASB 133.
Appendix: USC and WesCorp Structured Portfolio Composition at 6/30/10
Structured Portfolio Composition, Combined USC & Wescorp
RMBS: Subprime 28%
Structured Portfolio Composition by Corporate CU ($B) $9.0 $8.0 $7.0 $6.0
RMBS: 2nd Lien / NIM 3%
RMBS: Prime 6%
RMBS: Alt-A 46%
RMBS: 2nd Lien / NIM
Range of Potential Loss Severities by Asset Class Combined USC & WesCorp, as of June 30, 2010 Analysis limited to Non-Agency ABS, CMBS, and CDO securities
Percent of Portfolio
Estimated Loss Severity Range*
RMBS (Subprime, Alt-A, & Prime)
2nd Lien RMBS (HELOC, 2nd Lien and NIM)
*Loss Severities = Projected OTTI impairment divided by current face value 14
Bond Vintages* Combined USC & WesCorp Deal Vintage (as % of portfolio) 45% % of Total Portfolio
40% 35% 30% 25% 20% 15% 10% 5% 0% pre-2004
* “Vintage” refers to the year the security was created.