The late Hyman Minsky developed theories of financial crises as macroeconomic events. The economic logic he focused on starts with unrealistically high asset prices and buildups of leverage based on momentum effects, myopic expectations and widespread overleveraging of consumers and firms. When asset prices collapse, the negative wealth effect on aggregate demand is amplified by a “financial accelerator”; that is, collapsing credit feeds and feeds on falling aggregate demand credit. A severe economic decline is the outcome. Many bloggers refer to this as a "Minsky moment" (see Minsky 1975 for the real thing.)
I am sympathetic to the view that “Minsky moments” can happen (indeed, I have written numerous studies that give some support to that claim). But in my view, the correct application of the Minsky model to the current data indicates that we are not facing a Minsky moment – at least not yet. This column, which draws on a much longer that analysis I have posted at the AEI, summarises my reasoning.
At the moment, it is not obvious that housing or other asset prices are collapsing, or that leverage is unsustainably large for most firms or consumers. That is not to say that the economy will avoid a slowdown, or possibly even a recession. My main focus is not on forecasting changes in housing prices or consumption, per se, which are very hard to predict. I am interested in assessing the likelihood that financial weakness will substantially magnify aggregate demand shocks through a “financial accelerator” (otherwise known as a credit crunch).
The current liquidity shock
We are currently experiencing a liquidity shock to the financial system, initiated by problems in the subprime mortgage market, which spread to securitisation products more generally - that is, mortgage-backed securities, asset-backed securities, and asset-backed commercial paper. Banks are being asked to increase the amount of risk that they absorb (by moving off-balance sheet assets onto the balance sheet), but the related losses that the banks have suffered are limiting somewhat the capacity of banks to absorb those risky assets. The result is a reduction in aggregate risk capacity in the financial system as losses force those who are used to absorbing risk to sell off or close out their positions.
The financing of many risky activities unrelated to the core mortgage market shock has been reduced relative to their pre-shock levels. There are, at least temporarily, lots of "innocent bystanders" that are affected due to the aggregate scarcity of equity capital in financial intermediaries relative to the risk that needs reallocating.
The housing finance sector shock that started the current problems was small relative to the economy and financial system (estimated losses on subprime mortgages range from $200 billion to $400 billion). It was magnified because of the increased and imprudent use that has been made of subprime mortgage-backed securities in the creation of other securitisation conduits, and because of the connection of the instruments issued by those conduits to short-term asset-backed commercial paper.
From 2000 to 2005, the percentage of non-conforming mortgages that became securitised increased from 35% to 60%, and the volume of non-conforming origination also rose dramatically. Subprime mortgage originations rose from $160 billion in 2001 to $600 billion in 2006. And many of these securitised mortgages became re-securitised as backing for collateralised debt obligations (CDOs). As of October 2006, 39.5% of existing CDO pools covered by Moody’s consisted of MBS, of which 70% were subprime or second-lien mortgages. Why did subprime issuance boom from 2002 to 2006? Foreclosure rates for subprime mortgages actually peaked in 2002, but remarkably, that experience led to a sharp acceleration in the volume of subprime originations because the 2002-2003 foreclosures did not produce large losses. Losses from foreclosure were low in the liquid and appreciating housing market, and ratings agencies wrongly concluded that the forward-looking risks associated with subprime foreclosure were low. Instead, ratings should have recognised that this was an unusual environment, and that there was substantial risk implied by high foreclosure rates.
Despite CDOs’ increasing reliance on subprime mortgage-backed securities, and the observably low quality of these assets (i.e., high subprime foreclosure rates), CDO pools issued large amounts of highly rated debts backed by these assets. The CDO problem became magnified by the creation of additional layers of securitisation involving the leveraging of the “super-senior” tranches of CDOs (the AAA-rated tranches issued by CDO conduits). These so-called leveraged super-senior conduits, or “LSS trades,” were financed in the asset-backed commercial paper (ABCP) market. Some banks structured securitisations that levered up their holdings of these super-senior tranches of CDOs by more than 10 times, so that the ABCP issued by the LSS conduits was based on underlying organiser equity of only one-tenth the amount of the ABCP borrowings, with additional credit and liquidity enhancements offered to assure ABCP holders and ratings agencies. When CDO super-senior tranches turned out not to be of AAA quality, the leveraging of the CDOs multiplied the consequences of the ratings error, which was a major concern to ABCP holders of LSS conduits.
We have learned from the recent turmoil that mistakes in the pricing of fundamental risks in one market can have large consequences for the global financial system. In some ways, the global dimension of the shock is a sign of progress. Over the last two decades, securitisation had produced great progress in the sharing of risk and the reduction of the amount of financial system equity capital needed to absorb risk, by establishing mechanisms for transferring risk from banks’ and finance companies’ balance sheets to the market, and by establishing those mechanisms in creative ways that reduced adverse selection and moral hazard costs associated with more traditional securities markets.
That progress was real and these technological innovations will persist. Mistakes were made as part of what could be called a process of ‘learning by losing’ (the history of the last two decades has seen many temporary disruptions to the process of financial innovation in securitisation, as discussed in Calomiris and Mason 2004, of which the current liquidity shock is clearly the most severe). Securitisations have had a bumpy ride for two decades, which is inherent in innovation, but overall the gains from reshaping risk, sharing risk, and creating mechanisms that reduce the amount of equity needed per unit of risk (through improved risk measurement and management) have been large and will remain large, even if there is a substantial permanent shrinkage in securitised assets.
Risk reallocation has already produced a decline in the supply of available credit for some purposes, and this will not be fixed overnight. The financial system was devoting too little equity to intermediating risk in the mortgage securitisation market. There is likely to be a long-term reduction in the amount of credit that can be supplied per unit of equity capital in the financial system.
Furthermore, the shock occurred at a time when credit spreads seemed unreasonably low to many of us, reflecting the unusually high level of liquidity in the marketplace and the willingness of investors consequently not to charge sufficiently for bearing risk. In this sense, it is quite possible that credit spreads, once disturbed from those unrealistically low levels, will remain somewhat elevated after the shock dissipates.
But these adjustments, at least for now, do not a financial crisis make. It is possible that the financial system and economy could follow the patterns of 1970, 1987, and 1998 and recover from financial disturbances quickly without experiencing a recession, even without any further monetary policy stimulus by the Fed.1
Reasons to be cheerful
My view of the limited fallout rests on eight empirical observations:
1. Housing prices may not be falling by as much as some economists say they are.
Too much weight is being attached to the Case-Shiller index as a measure of the value of the US housing stock. Stanley Longhofer and I, along with many others, have noted (Calomiris and Longhofer 2007) that the Case-Shiller index has important flaws. Most obviously, it does not cover the entire US market, and the omitted parts of the US market seem to be doing better than the included parts. A comparison between the Case-Shiller and OFHEO housing price indexes shows that the Case-Shiller index provides a strikingly different, and less representative, picture of the US housing stock than OFHEO’s index. According to the OFHEO index, as shown in Figure 2, housing prices continued to rise on average through June 2007.
2. Although the inventory of homes for sale has risen, housing construction activity has fallen substantially.
The reduced supply of new housing should be a positive influence on housing prices going forward. Single-family housing starts dropped 7.1% in August relative to July and are down 27.1% on a year-to-year basis. Building permits for single-family homes slumped 8.1% in August (the largest decline since March of 2002) and are down 27.9% on the year. This decline in residential investment responded to an apparent excess supply problem; homeowner vacancy rates, which had averaged 1.7% from 1985 to 2005, jumped to 2.8% in 2006. The decline thus far in residential investment by the household sector as a share of GDP has been comparable by historical standards to the declines in the 1950s, 1960s, 1970s and 1980s (most, but not all, of which preceded recessions), as shown in Figure 8.
Note: Recessions are shaded.
Sources: Federal Reserve Statistical Release Z.1, Table F.6
(http://www.federalreserve.gov/releases/z1/Current/data.html); National Bureau of Economic Research, Business Cycle Expansions and Contractions (http://www.nber.org/cycles.html/)
3. The shock to the availability of credit has been concentrated primarily in securitisations rather than in credit markets defined more broadly (for example, in asset-backed commercial paper but not generally in the commercial paper market).
As Figure 9 shows, almost the entire decline in commercial paper in recent months has come from a contraction of asset-backed commercial paper, while financial commercial paper has contributed somewhat to the decline, and nonfinancial commercial paper has remained virtually unchanged.
This shows that the fallout from the shock has mainly to do with the loss in confidence in the architecture of securitisation per se, and secondarily with rising adverse-selection costs for financial institutions, but has not produced a decline in credit availability generally.
4. Aggregate financial market indicators improved substantially in September and subsequently. Stock prices have recovered, treasury yields rose in September as the flight to quality subsided, and bond credit spreads have fallen relative to their levels during the flight to quality (although Tbill yields remain low relative to other money market instruments).
5. As Figure 15 shows, nonfinancial firms are highly liquid and not overleveraged. Thus, many firms have the capacity to invest using their own resources, even if bank credit supply were to contract.
Note: Gross corporate leverage is defined as liabilities divided by assets. Net corporate leverage is defined as liabilities, less cash, divided by assets. Cash is defined as total financial assets, less trade receivables, consumer credit, and miscellaneous assets.
Sources: Federal Reserve Statistical Release Z.1, Table B.102
6. As David Malpass (2007) has emphasised, households’ wealth is at an all-time high and continues to grow. So long as employment remains strong, consumption may continue to grow despite housing sector problems.
7. Of central importance is the healthy condition of banks. As Fed Chairman Ben Bernanke noted from the outset of the recent difficulties, financial institutions’ balance sheets remain strong, for the most part, even under reasonable worst-case scenarios about financial sector losses associated with the subprime fallout. Bank lending has been growing rapidly, which is accommodating the transfer of securitised assets back onto bank balance sheets. The high capital ratios of banks at the onset of the turmoil is allowing substantial reintermediation to take place without posing a threat to the maintenance of sufficient minimum capital-to-asset ratios.
8. Banks hold much more diversified portfolios today than they used to. They are less exposed to real estate risk than in the 1980s, and much less exposed to local real estate risk, although US banks’ exposure to residential real estate has been rising since 2000 (Wheelock 2006).
In previous episodes of real estate decline (the 1920s, 1930s, and 1980s) much of the distress experienced by the banking sector resulted from its exposure to regional shocks, because of the absence of nationwide branch banking. In the 1980s, shocks associated with commercial real estate investments in the northeast, and oil-related real estate problems in the southwest, were particularly significant sources of banking distress. During the last two decades, however, banks have become much more diversified regionally, owing to state-level and federal reforms of branching laws. Banks also have a more diverse income stream due to the expansion of bank powers, which culminated in the 1999 Gramm-Leach-Bliley Act.
I conclude from this evidence that the consequences of the recent shocks for the supply of bank credit may turn out to be modest.
The current financial market turmoil resulted from a moderate shock to the housing and mortgage markets, which was magnified by the uses of subprime mortgages in a variety of securitisations vehicles, which produced a collapse of confidence in the architecture of securitisation and led to a sudden need to reallocate and reduce risk in the financial system. The liquidity risks inherent in maturity mismatched asset-backed commercial paper conduits substantially aggravated the short-term problem. Despite these disruptions, the fallout thus far in the financial system has been limited and appears to have been contained by a combination of market discipline and short-term central bank intervention. It is hard to know whether new financial shocks will occur (e.g., large housing price declines, or substantial increases in defaults on other consumer loans), or whether consumption demand will decline independent of financial system problems, but there is little reason to believe that a substantial decline in credit supply under the current circumstances will magnify the shocks and turn them into a recession. We have not (yet) arrived at a Minsky moment.
Of course, if housing prices fell by 50% nationwide (as some have argued is “entirely possible”) there is no question that the impact on consumers would be severe, both directly (via the decline in wealth) and indirectly (through its effects on the financial system). Judging from previous episodes of real estate price collapses, it would take years to sort out the losses. Real estate in liquidation is notoriously illiquid and hard to value; in the 1980s and early 1990s, banks and Savings and Loans that were stuck with large inventories of real estate took years to liquidate it, and given the valuation challenges associated with that real estate, found it costly to raise equity capital in the meantime. A real estate collapse would not only cause a decline in consumption via a wealth effect, it could produce a major financial accelerator effect.
Calomiris, Charles W., and Joseph R. Mason (2004). “Credit Card Securitization and Regulatory Arbitrage,” Journal of Financial Services Research, Vol. 26, 5-27.
Calomiris, Charles W., and Stanley D. Longhofer (2007). “Why the Housing Market Crisis Isn’t,” Working Paper.
Malpass, David (2007). “Running on Empty?” The American, September/October, 72-73.
Minsky, Hyman P. (1975). John Maynard Keynes, Columbia University Press.
Wheelock, David C. (2006). “What Happens to Banks When House Prices Fall? U.S. Regional Housing Busts of the 1980s and 1990s,” Federal Reserve Bank of St. Louis Review, September/October, 413-28.
1 Recent Fed actions through the discount window and the fed funds rate (discussed below) are comparable to the Fed actions in 1970, 1987, and 1998 – episodes during which Fed loosening was confined to fed funds rate declines that averaged 1.1% over the three episodes. To be specific, in 1970 the fed funds rate fell from 7.80% on June 17 to 6.34% on August 26; in 1987 it fell from 7.59% on October 14 to 6.43% on November 4; in 1998 it fell from 5.50% on September 29 to 4.75% on November 17.