At a Federal Reserve meeting I attended shortly after he took office, Chairman Bernanke remarked that the plural of anecdote was data. In that spirit, a working paper last month by economists affiliated with the Federal Reserve Bank of Minneapolis sought to shed some empirical evidence on what they characterized as four myths about the current financial crisis.(1) This paper was quickly followed by a working paper from the Federal Reserve Bank of Boston, which argued that, if one looked behind the data, the so-called myths weren’t myths at all.(2) Such a difference of opinion and open debate between researchers at Federal Reserve banks is both unusual and interesting, and it is testimony to the difficult times in which we find ourselves. Which of these works is more likely to be correct, and what did the authors do? In what follows, I attempt both to summarize the findings of the two camps and, in the spirit of improving the quality of the debate, to provide a disinterested third-party view of the issues.
The Four Myths
The FRB Minneapolis economists, hereafter referred to as CCK, argued there were four myths about the current crisis, which were that:
- Lending to nonfinancial corporations and individuals has declined sharply
- Interbank lending is essentially nonexistent
- Commercial paper issuance by nonfinancial corporations has declined sharply and
- Banks play a large role in channeling funds from savers to borrowers.
They employed the most current, publicly available data from the Federal Reserve and Bloomberg and showed that there was little support in the aggregate data that bank lending in total had declined; and in fact lending to both businesses and consumers actually appeared to increase, especially during the crisis period.(3) Similarly, the authors also found that, while volatile, interbank lending was substantially above what it was in mid-2005-2006. (In fact, further examination shows that it was up by nearly $150 billion or 50% greater). CCK also argue that the data also fail to support the idea that commercial paper issuance by nonfinancial corporations has declined or that this market has been closed to qualified issuers (such as some of those being accommodated under the Federal Reserve’s new facility); and this remains true even into November, where the data released subsequent to the study show that outstanding nonfinancial paper is above year-end 2007 levels and has not exhibited sharp declines. Even in the case of financial commercial paper, monthly aggregate outstanding volumes exceeded year-end levels for both 2006 or 2007 until September and October. And even here most of the drop in outstandings was to foreign financial issuers. The picture for either nonfinancial or financial issuers can’t compare with the declines in asset-backed commercial paper issuance, most of which was related to mortgage-backed paper. Finally, using data from the Federal Reserve’s Flow of Funds, CCK argue that nearly 80% of the sources of funding for businesses are outside the banking system, so concerns that bank troubles would destroy the ability of business to obtain funding were largely unfounded.
CCK’s attention then turns to credit spreads, which have typically played an important role in discussions of the supposed freezing up of credit markets. They argue that the focus on spreads may be appropriate in normal times, but in the current crisis, while spreads have widened, the level of rates remains low, as does the real cost of borrowing.
What About the Myth-Busters Busters Case?
The Boston Fed authors, hereafter referred to as BF, attack the cases for the four myths. First, while they concede that the aggregate data don’t show a decline in lending to consumers or nonfinancial businesses, but they argue that going behind the data to look at the detail shows a sharp weakening of credit conditions. Here their evidence and its relevance isn’t clear at all. For example, BF provide data on the decline of the issuance of asset-backed commercial paper. Most of this commercial paper, however, was related to mortgages, which would be expected to decline, given what has happened in the mortgage market and was primarily issued by financial firms. Thus, focusing on asset-backed paper is largely irrelevant to the issue of what has happened to lending to consumers and businesses. BF then go on to argue, by presenting data on the volume of unused loan commitments and commitments relative to loans, that this somehow sheds light on lending conditions. Unused commitments aren’t loans, they are simply commitments. That the ratio of commitments to loans has declined could be due to substitution of bank loans for other sources of credit, hence the reason that loan volumes haven’t shrunk. Or it might be simply due to a reduction in demand. The data provided by BF don’t enable one to examine these hypotheses and, again, are largely irrelevant to the issues. It is interesting, however, that recent reports from the National Federation of Independent Business suggest that shrinkage in loan demand may be part of the explanation and that, while credit conditions have become tighter, they are consistent with what one would see at the end of an expansion and haven’t risen to credit-crunch levels.
As for the second myth – that interbank lending has essentially dried up – the BF authors offer four arguments. The first is that by looking at the interbank market CCK ignore lending to and by brokers and dealers which, they assert, without empirical support, is the same size as the interbank market and has declined. Their second point is that anecdotal evidence suggests that a large portion of the interbank market has become of overnight maturity and isn’t working as usual. But again, anecdotes aren’t data and no data are provided. The third piece of evidence is data on the comparative cash holdings of large versus small banks. Cash holdings of large banks have ballooned, whereas there is no discernable increase in the cash holdings of small banks. BF interpret this to suggest that if interbank lending markets were functioning normally, then banks would not have to carry such large cash holdings because of the significant opportunity cost of holding cash. Finally, the authors argue that because interbank demands for funds are highly inelastic, rates, spreads provide a better indication of the health of the market than do volume statistics.
Of the four pieces of evidence offered by BF, only with respect to cash holdings is there hard data, while the other evidence is either anecdotal or in the form of assertions without empirical support. Interestingly, the data on cash holdings suggest that the so-called liquidity problem is really a large-bank problem, which is consistent with the difficulties experienced by the primary dealers. In other words, the interbank market problems, to the extent they exist, result from large banks being unwilling to transact with each other, rather than a feature of the entire banking system. Interestingly, virtually all of the Federal Reserve lending programs have been targeted to this problem and are directed toward the primary dealers. These include the Primary Dealer Credit Facility, Transitional Credit Extensions, Securities Lending Program, Term Securities Lending Facility, and Term Securities Lending Facility Options Program.
BF’s discussion of the third myth – that commercial paper issuance by nonfinancial companies has dried up – again suggests the need to disaggregate the commercial paper issuance data by rating classes. They assert that CCK’s conclusion rests primarily upon data on outstanding aggregate non financial paper and on rates for AA –90-day nonfinancial commercial paper. However, BF argue that disaggregation by ratings shows that not only has the issuance of A2P2 paper declined but that rates for this lower-quality paper have increased. As a result, they conclude that the third myth is false. Data on issuance of A2P2 nonfinancial paper show it has clearly slacked off. However, it is also the case that the monthly average volume, except for the partial data for November, of this lower quality paper for 2008 exceeded the annual average for 2006 (which predates the crisis period) and except for October and the partial month of November issuance also exceeded the annual average issuance for 2007. Clearly, there was a decline in the ability of lower-quality issuers to tap the market, and it is also clear that spreads for this segment increased. But most of this decline post dates the period when the claim was that the commercial paper market had frozen up for nonfinancial issuers. Even for those issuers unable to place paper, it is still possible that they were able to take down other credit lines. It is also the case that the level of rates to A2P2 borrowers declined and in absolute terms were below those that prevailed in early 2008 and throughout 2007. In this respect, then, pinning a conclusion on spreads may not reflect the real cost of borrowing, a point of emphasis by CCK.
Finally, with respect to CCK’s evidence that banks play only about a 20% role in channeling funds from households to businesses, the only counterarguments advanced by BF are that (1) we don’t know if the relationship is stable and (2) data show that outstanding consumer credit in the month of August declined. Neither of these are particularly persuasive, especially since we don’t know whether this decline represents a reduction in supply or a reduced quantity of outstanding credit because of reduced demand.
Are the Myths Busted?
This short tour of the issues and evidence suggests several conclusions. First, none of the evidence provided by BF appears to be fatally damaging to CCK’s case, as they would like to suggest, and in this sense the myths remain busted. The aggregate data provided by CCK don’t support the dire descriptions of the financial disruptions that have characterized many of the statements of government officials and the press. However, the analysis provided by CCK is also incomplete, since it should have considered evidence on both outstanding stocks and fund flows. In addition, a complete understanding of why quantities might have increased or decreased requires a decomposition into demand as well as supply effects. (4)
What we do know is there are clearly problems being experienced by large financial institutions – both banks and investment banks. Many have reported losses, some for the very first time. Second, those that relied upon short-term funding in the commercial paper market to support an originate-and-distribute business model have either failed or have been merged out of existence. Third, many of these institutions were primary dealers and had been relied upon as a major channel as the Federal Reserve attempted to conduct ordinary monetary policy activities. That conduit appears to have broken down and is need of restructuring and repair. The special treatment these institutions have been afforded hasn’t worked. Fourth, the FDIC maintains that the vast majority of banks are sound and well-capitalized. They will be even more so as they avail themselves of the government subsidies inherent in the TARP preferred-capital injection program. Finally, this is a debate and discussion that is worth continuing, so as to better understand the problem and to design appropriate remedies.
(1) See “Facts and Myths about the Financial Crisis of 2008,” by Chari, Christiano and Kehoe, WP 666, Federal Reserve Bank of Minneapolis, October 2008.
(2) “Looking Behind the Aggregates: A Reply to ‘Facts and Myths about the Financial Crisis of 2008,’” by Cohen-Cole, Duygan-Bump, Fillat and Montoriol-Garriga, Federal Reserve Bank of Boston Working Paper No. QAU08-5. November 3, 2008.
(3) Subsequent extensive documentation of the data sources were released in “Technical Notes on Facts and Myths about the Financial Crisis of 2008,” by Maxim Troshkin, WP 667, Federal Reserve Bank of Minneapolis, October 2008.
(4) To their credit BF do recognize the need to consider both supply and demand effects.