The Committee probably does not need reminding that the recent financial crisis, like crises in general, was quite painful, both to the overall economy and to individuals and their families. While it is difficult to disentangle the relative effects of the crisis was the recession that followed the housing bust, we should not forget that unemployment rose to 10 percent in the fall of 2009, with almost 9 million workers losing their jobs in the recent downturn. This has been by almost every measure a weak, slow and painful recovery. It has fallen far short of previous recoveries. The Federal Reserve Bank of Dallas estimates that the cost in lost economic output could be as high as $14 trillion, close to a full year’s GDP.1 The fiscal costs have also been tremendous as both stimulus programs and corporate rescues added to the deficit.
Scholars have found that this pattern is all too common in the wake of financial crises. Adding to that urgency is that financial crises are often followed by poor and sometimes harmful public policy choices. For instance most economists recognize today that many of the New Deal policies implemented in the 1930s slowed the recovery and added to unemployment.2 Although harder to quantify, an important reason to avoid financial crises is to avoid the policy mistakes that sometimes follow in their aftermath. Accordingly I hope we all share the goal of minimizing both the severity and frequency of financial crises. This is not something we ever want to repeat again.
Let me also add that unfortunately financial crises are less rare than is commonly portrayed, at least in the United States. Reinhart and Rogoff estimate that since around the founding of the United States, we have spent approximately 13 percent of the time in a financial crisis or its aftermath.3 That’s a crisis about every 8 years. And of course many members of this Committee recall the Savings and Loan Crisis. We should not fool ourselves into believing that 2008 was just a “hundred flood”.
In 2010 Congress passed the Dodd‐Frank Wall Street Reform and Consumer Protection Act. Many of the stated purposes of Dodd‐Frank, such as a desire “to promote market discipline”, are ones I believe most, if not all, of us share. Stated goals and purposes, however, are not the same thing as achieving said goals and purposes. Good intentions are not sufficient. We must closely evaluate effectiveness and do so setting aside any pre‐existing biases. Before turning to Dodd‐Frank, let me very clearly state that the alternative to Dodd‐Frank was not “doing nothing”. I do not believe the pre‐2010 financial system was either sound or safe. In fact I had spent much of the years preceding the crisis trying to enact measures to avoid it.
While public opinion can be, and quite often is, wrong, it is nevertheless a useful place to start. This is particularly true in finance, where public confidence in our financial regulatory system is a crucial component of its success. A Morning Consult poll conducted earlier this year queried respondents as to whether believed our financial system was stronger today than it was five years ago.4 A plurality (42%) did not believe our financial system was safer. Even among Democrats, slightly more (35% v. 34%) believed our system was less safe. Among Republicans and Independents it was not even close, with an over 20 percentage point gap between those who felt our system was less safe than those who believe it is safer. In sum, the public continues to exhibit considerable skepticism regarding the soundness of our financial system.
Whether any particular remedy is effective depends crucially on the diagnosis and the underlying malady. The weight of the evidence suggests to me that recent crisis was driven largely by a boom and bust in our nation’s property markets, particularly housing, and that our current system of financial regulation linked the performance of the housing and mortgage market to our capital markets in such a manner as to result in a significant disruption when the value of housing and mortgages declined. Key ingredients of this crisis were: exceptionally loose monetary policy, supply rigidities in our property markets, extensive international capital flows into the US (and US mortgage market), extensive policy encouragements for both high corporate and household leverage and extensive moral hazard created by various government guarantees. I do not believe we have sufficiently addressed these distortions. Accordingly we should expect these forces to result in future crises. It is not enough to just “do something” — we must do the correct things.
Moral Hazard and Too‐Big‐To‐Fail
There is perhaps no issue that drove the passage of the Dodd‐Frank Act more than the public perception that certain large institutions enjoyed the backing of the federal government. A situation commonly called “too big to fail” (TBTF). Not by coincidence the first two titles of Dodd‐Frank are aimed at addressing the too‐big‐to‐fail status of our largest financial institutions. Here I raise a number of concerns and observations that merit meeting the claim of ending TBTF with considerable skepticism.
One reason that debates over TBTF are often so heated is that there is no actual explicit subsidy provided on‐budget for such purpose. We can (and should) debate the merits and effects of deposit insurance, for instance, but we don’t debate its existence. It is there. Whether TBTF is real or not is a far trickier question. We are left with no choice but looking for “clues”.
The first clue is the actual text of Dodd‐Frank. There is, of course, language about “eliminating expectations…that the Government will shield” parties from losses “in the event of a failure”. But vague purposes do not constrain explicit authorities. And Dodd‐Frank is quite explicit. Section 204, for instance, is quite clear that the Federal Deposit Insurance Corporation (FDIC) can purchase any debt obligation at par (or even above) of a failing institution. If rescuing a creditor at par is not the very definition of TBTF, I’m not sure what is. Section 201 goes even further by allowing the FDIC to pay “any obligations…” it believes are “necessary and appropriate”. Yes Dodd‐Frank does offer a path for ending TBTF without cost to the taxpayer or the rest of the financial industry. But that path is clearly an optional one. We should not forget that JP Morgan CEO Jamie Dimon called for such a resolution mechanism in 2009.5 It seems unlikely he’d call for something that would undermine the value of a firm he spent so much effort building.
If Congress does not see fit to repeal these titles of Dodd‐Frank, at a minimum they should be amended to limit the ability of regulators to rescue creditors.
One reason I do not believe ending TBTF will be the path taken is that this was also an option with Fannie Mae and Freddie Mac. The Housing and Economic Recovery Act of 2008 created a mechanism similar to Dodd-Frank’s Title II resolution process. It could have been used to keep Fannie and Freddie functioning without a dime of cost to the taxpayer. Instead the taxpayer was tapped. Recall that Freddie is actually smaller and far less complex than CitiBank. Defenders of Dodd‐Frank have yet to offer a reason why Citi would be allowed to fail, when Freddie was not.
Much of the debate over TBTF has revolved around various studies as to the “size” of the subsidy. Such exercises are useful but limited. For instance raw FDIC data suggests that the largest banks enjoy a funding advantage over other banks. Yes that raw advantage has declined since the height of the crisis. One would expect such since bailouts are more likely when policy‐makers are panicking. On the other hand that advantage is bigger than it was pre‐crisis. Compared to the average funding costs over the last three decades, the largest banks enjoy a funding advantage of around 30 basis points, similar to that enjoyed by Fannie and Freddie pre‐crisis. Even if the observed advantage disappeared, such does not mean TBTF is gone. As any first year statistics or science student learns, the absence of a finding is that the same as a finding of absence. During most of the last three decades and even up to 2007, the largest banks actually paid more to borrow than the rest. The data can at best be suggestive.
Recent research from the Federal Reserve Bank of New York suggests even the FDIC’s single‐point‐of‐entry approach has failed to convince market participants that TBTF is over.6
Estimates from the Federal Reserve Bank of Richmond suggest that taxpayer backing of the financial system has greatly expanded. Based upon actions taken during the crisis, the Richmond Fed estimated that 59 percent of US financial system was either explicitly or implicitly backed by the federal government in 2009.7 Their most recent estimates (March 2015) are that 60 percent of US financial system was either explicitly or implicitly backed by the federal government. By the expert judgements of the Richmond Fed, Dodd‐Frank has not resulted in any net reduction in the federal safety net for financial institutions.
It is worth noting a significant expansion of the banking safety net is the result of expanded explicit guarantees in the Dodd‐Frank Act. For instance Dodd‐Frank expands deposit insurance to $250,000 per person per bank. Since the crisis the amount of insured deposits outstanding has risen by over $2 trillion.8 This is another $2 trillion which the taxpayer directly stands behind. It is also another $2 trillion that is immune from market discipline.
According to the Federal Reserve’s Survey of Consumer Finance, the median US household held $4,100 in a checking account.9 For the less than 10 percent that held certificates of deposit, the median holding was $16,000. So let us be crystal clear, a deposit coverage cap of $250,000 has nothing to do with protecting the savings of typical American families. A cap of say $40,000 (that pre-S&L crisis) would more than adequately cover the vast majority of US households while also greatly improving market discipline on US banks. Even the typical (median) retirement account, not all of which are held at banks, is under $60,000.
When it comes to TBTF a special area of concern is the Federal Reserve’s primary dealer system. As the Committee is well, these (currently 22) institutions are the Fed’s primary partners in the conduct of monetary policy. Were a significant number of these entities to become stressed, such could hamper the Fed’s ability to conduct open market operations. Unfortunately such suggests to me that the Fed will protect these entities in order to preserve its own powers. The Fed (and/or this Congress) should examine the functioning of the primary dealer system with the objective of increasing its ability to withstand the failure of a few primary dealers. Ideally the number of participating primary dealers should be expanded sufficiently to protect against any impact on monetary policy effectiveness from the failure of a primary deader.
Dodd-Frank’s push to concentrate derivatives counterparty risk into centralized clearinghouses also runs the risk of creating new TBTF entities, as noted by Treasury’s Office of Financial Research.10
While much of the disgust expressed regarding TBTF comes from its obvious unfairness, treating some institutions more favorably than others, we should recall that the moral hazard created by these both explicit and implicit guarantees reduces the level of private monitoring of banks. Private parties, with their own money at risk, have strong incentives to monitor bank behavior, while regulators, who never seem to lose their jobs regardless of their performance, have relatively weak incentives to monitor banks. The continued crowding out of private monitoring by financial regulators has likely lead to less monitoring of our financial system, not more.
Implied guarantees (like TBTF) are best viewed as a “dial” rather than a switch. The question is have we moved the dial closer to zero. I believe we have not.
It is generally accepted among economists (at least outside the Federal Reserve) that extraordinarily loose monetary policy post‐9/11 contributed to a boom in property prices, particularly in housing.11 Given that just three years of negative real federal funds rates contributed to the massive housing boom in the mid 2000s, one could only imagine what six years of negative real rates is doing to our asset and financial markets today. While one can always debate the presence of bubbles, as we did in the mid 2000s, I am most certain that our current stance of monetary policy is creating sizable distortions that will be quite painful when they unwind. One answer to the question of whether our financial markets are more stable today, is a loud “no” given the actions of the Federal Reserve. Unfortunately Dodd‐Frank not only failed to address failings at the Federal Reserve but also expanded its responsibilities.
Most accounts of the financial crisis of 2008 include a prominent role for the U.S. residential mortgage market. Although other property markets exhibited similar boom and bust patterns, the elevated level of defaults and associated costs borne by the taxpayer have brought a particular emphasis on single‐family mortgage finance policies.
Dodd‐Frank contains a number of provisions impacting the mortgage market. The financial services law firm Davis Polk estimates that Dodd‐Frank will require 49 separate rule‐makings in the area of mortgage reform alone. Of particular importance are those found in titles IX, X and XIV. Despite the extensive expansion of mortgage regulation under Dodd‐Frank, it is unlikely that such will significantly reduce mortgage defaults or mitigate future booms and busts in the housing market.
A goal of the Dodd‐Frank Act is to eliminate certain products and practices from the mortgage market. So at a very basic level the choices facing mortgage borrowers will be reduced, the difficult question is in gauging how much. At least three independent attempts have been made to estimate the impact of QRM and/or QM on mortgage availability. These three analyses were performed by the United States Government Accountability Office (GAO), the Federal Housing Finance Agency (FHFA) and the private firm CoreLogic.
The three studies yield similar conclusions as to the impacts of QRM. The most restrictive provision of the QRM rule would be the ceiling on allowable debt‐to‐income ratios (DTI). Most QRM restrictions are likely to have very modest impacts, as their prevalence in the mortgage market was generally low.121314
The Dodd‐Frank Act is a response to the theory that “bad” mortgage lending and lenders drove borrowers into default, which ultimately drove the housing market into decline leading to a fall in the value of mortgage‐backed securities, resulting in a panic among the holders of mortgage‐backed securities.15 Setting aside that national house prices reached an inflection point almost a year before the inflection point in defaults, one measure of the effectiveness of Dodd-Frank’s mortgage rules will be to what extent does it reduce mortgage defaults.
Despite having the largest impact on the number of loans, the proposed QM/QRM restrictions on DTI appear to have very modest impacts on projected defaults.16 Restrictions on low‐ or no‐documentation loans do appear to have noticeable impacts on defaults in the subprime market. If all but full documentation loans were used, default probabilities, according to GAO’s analysis would fall by -1.08, -1.17, and -1.24 percentage points for fixed rate, long‐term ARM and Hybrid ARM, respectively.
GAO’s default analysis predicts substantial declines in defaults from reductions in LTV, particularly initials moves below a 100 percent closed LTV. For fixed rate non‐prime purchase loans, moving from a LTV of 100 to under 80 percent reduces projected default probabilities by over 3 percentage points. For hybrid non‐prime ARMs, the reduction in projected default probabilities is just over 6 percentage points. Coupled with full documentation and a LTV under 80 percent, one could eliminate over 70 percent of the standardized default risk among hybrid non‐prime ARMs. Academic studies have arrived at similar conclusions when examining the drivers of default among subprime mortgages.
The approach of Dodd-Frank’s mortgage provisions is to focus on loan characteristics, largely ignoring borrower characteristics or housing market impacts. For instance QM/QRM places no restrictions on borrower credit, other than verification. A number of studies, however, find the largest impact on subprime defaults coming from borrower credit, as measured by FICO score. Increasing borrower FICO by one‐standard‐deviation, or about 74 points, decreasing default probability by around seven times as much as switching from an ARM to fixed rate. A 74 point increase in FICO also has over twice the impact of moving from a no/low to full documentation low. Studies also find the impact of housing price changes to be magnitudes higher than the provisions of the QM/QRM rule.
As the down‐payment requirements of the proposed QRM rule were abandoned, the remaining changes are likely to have modest impacts on default probabilities. The biggest impact would be from the full documentation requirements and the cap on DTI. These two changes combined, however, are projected to lower default probabilities by around 1 percentage point.
A study from University of Minnesota Professor Morris Kleiner finds that states with more stringent licensing requirements for mortgage brokers actually witnessed higher levels of mortgage default. The hypothesis is that increased barriers to entry reduce underwriting efforts to such an extent that off‐sets any improvements in broker quality that result from the licensing scheme. Kleiner’s results raise the possibility that Dodd-Frank’s Section 1401 originator requirements, coupled with the SAFE Act, actually increase mortgage defaults rather than reduce them, as the statute intends.
The barely noticeable reduction in projected defaults could be more than off‐set by Dodd-Frank’s impact on the foreclosure process. Dodd-Frank’s Section 1413 allows borrowers an additional delay to the foreclosure process. A longer foreclosure process increases the borrower’s incentive to default. New regulations relating to mortgage servicing that are likely to extend the ultimate time to foreclosure. Researchers, as well as industry experience, confirm the increase in “strategic default” during the recent crisis. Dodd-Frank’s Section 1414(g) notice on anti‐deficiency and the increased delays to foreclosure may well increase strategic defaults more than an amount to off‐set reductions resulting from the QM/QRM provisions. Scholars have found that delays in the foreclosure process largely extend the process, raising the overall level of loans in foreclosure at any one time, without significantly improving final outcomes for the borrower. Dodd‐Frank could very well result in an increase in the level of mortgage defaults during the next housing bust.
If and when those additional defaults occur they will mostly be borne by the taxpayer. Given that Fannie Mae, Freddie Mac and the FHA guarantee the vast majority of today’s mortgage market, as that all of those entities lack any real degree of capital, losses will directly hit the taxpayer. Without substantial reform of our mortgage finance system, the next downturn in the housing market will result in even larger taxpayer losses than the last downturn.
Despite complaints from the mortgage industry, underwriting standards do not appear to have greatly improved. According to FHFA, the typical loan‐to‐value at the peak of the market in 2006 was 76.4, most recently data is 78.5 (May 2015). Over half of FHA’s most recent endorsements are to borrowers with FICOs under 680. Almost half a million single‐family FHA loans are currently non‐performing, or around 6% of insured loans.17 What’s shocking is that this is in an environment of rising home prices. When house prices decline, this number will increase considerably. The two most prominent drivers of mortgage default in the last bust were borrower FICO and LTV. We are essentially back to similar levels for both those metrics.If we continue along this path at least a million families will lose their homes during the next downturn due to loose FHA underwriting. The primary constraint on the mortgage market today appears to be households that are rightly skeptical of betting their financial health on rising home prices. Unfortunately far too many policy‐makers appear intent on reducing that skepticism.
The recent crisis should also leave us with little doubt that house prices do indeed decline. We have not ended the housing cycle. If anything it has likely become more volatile.
I would note the IMF’s recent Financial System Stability Assessment for the United States specifically lists FHA as a concern:“One in five loans originated is insured by the FHA, although it falls short of its capital requirements, which creates fiscal and financial risks due to moral hazard, the distorted competitive landscape, and large subsidies for debt‐financed homeownership.”18
Perhaps one of the few relatively bright areas is the increase in capital ratios among banks. Among all banks Tier 1 risk‐weighted capital was just under 7 percent of assets at the end of 2006. It now stands just over 9 percent. A small part of this increase is due to banks shifting toward lower risk‐weighted assets, which increases risk‐weighted capital without actually raising capital. That said, significant capital has been raised. It is worth noting that effort to increase bank capital began before and independently of Dodd‐Frank. We would likely have higher bank capital today even if Dodd‐Frank had not passed. Perhaps more importantly is the recognition that even the higher levels witnessed today are quite low. Let us not forget that the largest banks, and industry in general, were defined as “well capitalized” during the crisis. At the time of TARP’s passage total risk‐weighted capital as a percent of total risk‐weighted assets was well over 11 percent for industry as a whole. Even Citibank, one of the weakest, remained about 8 percent during the crisis. Either we didn’t need the TARP or something is wrong with our system of capital regulation (or both).
I commend the Committee for calling today’s important hearing. As we so painfully learned over the last decade, financial stability cannot be taken for granted. Although a few modest improvements have been made to increase financial stability, I believe Dodd‐Frank, no net, has reduced financial stability. The reason for such is a combination of both errors of commission and omission. Moral hazard has been increased by Dodd-Frank’s expansion of the financial safety net and increased concentration of risk into fewer entities, while the primary causes of the crisis were largely left untouched. I fear if we continue along our current path, we are almost certain to see another financial crisis sometime in the next decade.
It is also worth observing that one impact of Dodd‐Frank has been a growing concentration within the banking industry. Regulatory costs often fall heavier on smaller entities. The preference should be for measures, such as a flat leverage ratio, whose burdens do not increase disproportionately with decreases in size. We should also recognize that the ultimate burden of financial regulations will fall on those entities will less options. Unfortunately this is often low income families and small businesses.19 If Dodd‐Frank had truly achieved significant financial stability, then perhaps these costs would be worth bearing. Unfortunately Dodd‐Frank imposes significant costs but provides at best modest benefits.
1Assessing the Costs and Consequences of the 2007-09 Financial Crisis and Its Aftermath. http://www.dallasfed.org/research/eclett/2013/el1307.cfm
2See: https://www.cato.org/blog/did-new-deal-help & http://newsroom.ucla.edu/releases/FDR-s-Policies-Prolonged-Depression-5409
3Carmen Reinhart and Kenneth Rogoff. 2009. This Time is Different: Eight Centuries of Financial Folly. Princeton.
4For poll results, see http://morningconsult.com/polls/finance-104/
11See John Taylor, Getting Off Track.
12For instance, Both the QM and QRM rules ban negative amortization features, yet according to GAO’s analysis “almost 100 percent of [subprime] mortgage originations from 2001 to 2007 did not have negative amortization features.” Within the prime market the percent with negative amortization features peaked in 2005 at 9 percent. The average between 2001 and 2010 was closer to 1 percent.
13Dodd‐Frank also places limitations on mortgages with terms in excess of 30 years. In the prime and near‐prime market essentially 100 percent of mortgages were under a 30 year term until about 2005, where longer than 30 year mortgages grew slowly to 4 percent of the market in 2007 before disappearing by 2009. Subprime followed a more unusual situation with nearly 100 percent of subprime being under 30 years until 2005 and 2006, when the share over 30 years peaked at 15 percent of the subprime market.
14Another loan feature restricted by Dodd‐Frank is the use of balloon payments. Final balloon payments are multiples of the monthly payment. Despite the prevalence of balloon loans before the New Deal mortgage reforms of the 1930s, these products were generally rare, even during the height of the recent boom. GAO reports that almost 100 percent of prime, near‐prime and government‐insured mortgages lacked any balloon features between 2001 and 2010. Among subprime loans balloon features were also rare, close to zero until 2005 when they grew to about 10 percent of subprime loans in 2007, after which they have largely disappeared from the subprime market.
15Both the QM and QRM place restrictions upon borrower documentation, particularly in the area of income. A common concern is that no‐ or low‐documentation loans lead to greater levels of fraud and higher losses in the mortgage market than would have occurred otherwise. Whereas the QRM is an obstacle for securitization, the QM standards come with substantial and uncertain liability, so while there is likely to be a market for non‐QRM loans; non‐QM loans will become rare. By GAO’s estimates, the percentage of subprime loans lacking full documentation ranged from 40 percent in 2006 to 20 percent in 2001. A similar, but smaller, trend was witnessed among prime loans, where percent lacking full documentation ranged from around 20 percent in 2006 to almost zero in the early 2000s. The documentation requirements under QM/QRM are likely to impact most self‐employed borrowers. As there are over 15 million self‐employed individuals in the United States, these restrictions could be significant.
16The presence of a DTI in excess of 41 percent increases the probability of default by 0.25, 0.08, and 0.59 for fixed rate, long‐term ARM and Hybrid ARM, respectively. Accordingly to GAO’s analysis, reducing the prevalence of mortgages with a DTI in excess of 41 will have barely notice effects (although statistically significant in all cases).
18http://www.imf.org/external/pubs/ft/survey/so/2015/pol070715a.htm page 32