Real Estate Research provides analysis of topical research and current issues in the fields of housing and real estate economics. Authors for the blog include the Atlanta Fed's Jessica Dill, Kristopher Gerardi, Carl Hudson, and analysts, as well as the Boston Fed's Christopher Foote and Paul Willen.
Comments are moderated and will not appear until the moderator has approved them.
Please submit appropriate comments. Inappropriate comments include content that is abusive, harassing, or threatening; obscene, vulgar, or profane; an attack of a personal nature; or overtly political.
In addition, no off-topic remarks or spam is permitted.
November 24, 2015
The Pass-Through of Monetary Policy
In the wake of the Great Recession, the Federal Reserve instituted three rounds of large-scale asset purchases (LSAPs) in 2008, 2010 and 2012, more commonly known as "quantitative easing 1" (QE1), "QE2" and "QE3." The objective of these interventions was to keep interest rates low in an attempt to stimulate household consumption and business investment.1
In the United States, housing is the single largest asset on the household balance sheet, accounting for 73 percent of nonfinancial assets for the average U.S. household and an even higher share for homeowners.2 Mortgage payments represent the largest class of household debt obligation.
Evidence of the effectiveness of the asset purchase programs on real economic activity has until recently been limited due to the lack of data and a credible identification strategy (by which we mean a way to separate the causal impact of the LSAPs on the economy from other government programs and market factors that were occurring at the same time). When we chart a timeline of the three LSAPs against the primary mortgage rate, we can see that the primary mortgage market rate effectively dropped below 6 percent when the Fed began buying $600 billion in mortgage-backed securities during QE1. Indeed, the rate dropped following each of the subsequent LSAPs.
Since 2009, a number of papers have been published that evaluate the effectiveness of the policy interventions through different transmission channels. One such paper (Keys, Piskorski, Seru, and Yao, 2014) reports on borrowers with adjustable-rate mortgages (ARMs) who automatically receive the benefits of lower interest rates with no frictions or transaction costs, unlike borrowers with fixed-rate mortgages (FRMs) who must refinance in order to take advantage of lower interest rates. The paper provides new evidence on the effectiveness of the LSAPs.
Our strategy is to compare the change in the household balance sheets of 7/1 ARM borrowers to those of 5/1 ARM borrowers, using credit bureau data linked to mortgages. These two ARMs are the most popular ARM products among prime borrowers with very similar credit quality and risk preferences, yet they differ only in years 6 and 7, when the 5/1 ARM is eligible for a rate reset and the 7/1 is still locked (that is, the rate is still fixed). This creates a natural experiment that allows us to isolate other factors that might affect the mortgage rate.
By controlling for borrower characteristics and economic environments, we estimate that mortgage rates in the treatment group (5/1 ARMs) dropped in the first year by 1.14 percentage points, from 5.1 percent, and that payments dropped by $150 per month, or about a 20 percent reduction on average. The average borrower had a cumulative two-year savings of $3,456.3 We also subsequently found that borrowers spent 18 percent of the money saved on paying off credit card balances and that there was an 11 percent increase in new car purchases for the group. As a result, the leverage of U.S. households' dropped considerably from its peak during the financial crisis.4
We also find significant heterogeneity for these effects across different populations. Less creditworthy and more liquidity-constrained borrowers appear to have benefited the most from LSAPs as they experienced significantly greater reductions in mortgage rates and payments and larger improvements in mortgage and credit card performance. In terms of how they spent the extra liquidity received, highly leveraged borrowers (high credit utilization) spent 40 to 50 percent of the extra liquidity received during the first year, or $814 out of $1,740, to repay their revolving debts, then spent 20 percent of the extra liquidity received during the second year. Borrowers in the top quartile of credit utilization rates allocated about 70 percent of the extra liquidity toward repaying their credit card debt. We found similar effects among borrowers in the bottom quartile of credit scores. (The low-wealth borrowers with low credit utilization experienced a much larger increase in auto debt or new car purchases.) In other words, the LSAP programs effectively stimulated household investment and consumption.
We also find, as a result of the estimated effects at the micro level, a significant impact on local (nontradable) employment growth, consumer spending, and house price recovery in regions that were more exposed to ARMs. For example, a 10 percentage point increase in the ARM share, which is associated with about a 20-basis-point average reduction in ZIP code mortgage rates, is associated with about a 0.25 percentage point increase in quarterly home price growth, or about 1 percent annual appreciation, a very meaningful increase.
By Vincent Yao, visiting scholar at the Federal Reserve Bank of Atlanta and associate professor in the Real Estate Department in the J. Mack Robinson College of Business at Georgia State University.
Di Maggio, Marco; Amir Kermani; and Rodney Ramcharan. 2014. "Monetary Pass-Through: Household Consumption and Voluntary Deleveraging," Working Paper.
Hancock, Diana and Wayne Passmore. 2014. "How the Federal Reserve's Large-Scale Asset Purchases (LSAPs) Influence Mortgage-Backed Securities (MBS) Yields and U.S. Mortgage Rates," Finance and Economics Discussion Series 2014–12. Board of Governors of the Federal Reserve System.
Keys, Benjamin J.; Tomasz Piskorski; Amit Seru; and Vincent Yao. 2014. "Mortgage Rates, Household Balance Sheets, and the Real Economy," NBER Working Paper No. 20561.
1 The LSAPs involved purchases of long-term securities issued by the U.S. Treasury, agency debts, and agency mortgage-backed securities (MBS). They ultimately affected the yields of the MBS as well as the mortgage rates offered to borrowers in the primary mortgage market through several potential transmission channels: (1) the signaling of the Fed's commitment to keeping rates low, (2) a portfolio rebalance between assets and deposits and among different durations, and (3) increasing the liquidity value of MBS (Hancock and Passmore, 2014).
2 Survey for Consumer Finance, Federal Reserve Board of Governors, 2013.
3 Di Maggio, Kermani. and Ramcharan (2014) found much bigger savings for subprime and Alt-A borrowers based on a similar approach.
4 It is notable that in the United States the majority of prime borrowers take out fixed rate mortgages while most subprime borrowers take out adjustable rate mortgages.
April 20, 2015
Income Growth, Credit Growth, and Lending Standards: Revisiting the Evidence
Almost a decade has passed since the peak of the housing boom, and a handful of economics papers have emerged as fundamental influences on the way that economists think about the boom—and the ensuing bust. One example is a paper by Atif Mian and Amir Sufi that appeared in the Quarterly Journal of Economics in 2009 (MS2009 hereafter). A key part of this paper is an analysis of income growth and mortgage-credit growth in individual U.S. ZIP codes. The authors find that from 2002 to 2005, ZIP codes with relatively low growth in incomes experienced high growth in mortgage credit; that is, income growth and credit growth were negatively correlated during this period.
Economists often cite this negative correlation as evidence of improper lending practices during the housing boom. The thinking is that prudent lenders would have generated a positive correlation between area-level growth in income and mortgage credit, because borrowers in ZIP codes with high income growth would be in the best position to repay their loans. A negative correlation suggests that lenders instead channeled credit to borrowers who couldn't repay.
Some of the MS2009 results are now being reexamined in a new paper by Manuel Adelino, Antoinette Schoar, and Felipe Severino (A2S hereafter). The A2S paper argues that the statistical evidence in MS2009 is not robust and that using borrower-level data, rather than data aggregated up to the ZIP-code level, is the best way to investigate lending patterns. The A2S paper has already received a lot of attention, which has centered primarily on the quality of the alternative individual-level data that A2S sometimes employ.1 To understand the relevant issues in this debate, it's helpful to go back to MS2009's original statistical work that uses data aggregated to the ZIP-code level to get a sense of what it does and doesn't show.
Chart 1 summarizes the central MS2009 result. We generated this chart from information we found in either MS2009 or its supplementary online appendix. The dark blue bars depict the coefficients from separate regressions of ZIP-code level growth in new purchase mortgages on growth in ZIP-code level incomes.2 (These regressions also include county fixed effects, which we discuss further below.) Each regression corresponds to a different sample period. The first regression projects ZIP-level changes in credit between 1991 and 1998 on ZIP-level changes in income between these two years. The second uses growth between 1998 and 2001, and so on.3 During the three earliest periods, ZIP-level income growth enters positively in the regressions, but in 2002–04 and 2004–05, the coefficients become negative. A key claim of MS2009 is that this flip signals an important and unwelcome change in the behavior of lenders. Moreover, the abstract points out that the negative coefficients are anomalous: "2002 to 2005 is the only period in the past eighteen years in which income and mortgage credit growth are negatively correlated."
There are, however, at least three reasons to doubt that the MS2009 coefficients tell us anything about lending standards. First of all, the coefficients for the 2005–06 and 2006–07 regressions are positive—for the latter period, strongly so. By MS2009's logic, these positive coefficients indicate that lending standards improved after 2005, but in fact loans made in 2006 and 2007 were among the worst-performing loans in modern U.S. history. Chart 2 depicts the share of active loans that are 90-plus days delinquent or in foreclosure as a share of currently active loans, using data from Black Knight Financial Services. To be sure, loans made in 2005 did not perform well during the housing crisis, but the performance of loans made in 2006 and 2007 was even worse.4 This poor performance is not consistent with the improvement in lending standards implied by MS2009's methodology.
A second reason that sign changes among the MS2009 coefficients may not be informative is that these coefficients are not really comparable. The 1991–98 regression is based on growth in income and credit across seven years, while later regressions are based on growth over shorter intervals. This difference in time horizon matters, because area-level income and credit no doubt fluctuate from year to year while they also trend over longer periods. A "high-frequency" correlation calculated from year-to-year growth rates may therefore turn out to be very different from a "low-frequency" correlation calculated by comparing growth rates across more-distant years. One thing we can't do is think of a low-frequency correlation as an "average" of high-frequency correlations. Note that MS2009 also run a regression with growth rates calculated over the entire 2002–05 period, obtaining a coefficient of -0.662. This estimate, not pictured in our graph, is much larger in absolute value than either of the coefficients generated in the subperiods 2002–04 and 2004–05, which are pictured.
A third and perhaps more fundamental problem with the MS2009 exercise is that the authors do not report correlations between income growth and credit growth but rather regression coefficients.5 And while a correlation coefficient of 0.5 indicates that income growth and credit growth move closely together, a regression coefficient of the same magnitude could be generated with much less comovement. MS2009 supply the data needed to convert their regression coefficients into correlation coefficients, and we depict those correlations as green bars in chart 1.6 Most of the correlations are near 0.1 in absolute value or smaller. To calculate how much comovement these correlations imply, recall that the R-squared of a regression of one variable on another is equal to the square of their correlation coefficient. A correlation coefficient of 0.1 therefore indicates that a regression of credit growth deviated from county-level means on similarly transformed income growth would have an R-squared in the neighborhood of 1 percent. The reported R-squareds from the MS2009 regressions are much larger, but that is because the authors ran their regressions without demeaning the data first, letting the county fixed effects do the demeaning automatically. While this is standard practice, this specification forces the reported R-squared to encompass the explanatory power of the fixed effects. The correlation coefficients that we have calculated indicate that the explanatory power of within-county income growth for within-county credit growth is extremely low.7 Consequently, changes in the sign of this correlation are not very informative.
How do these arguments relate to A2S's paper? Part of that paper provides further evidence that the negative coefficients in the MS2009 regressions do not tell us much about lending standards. For example, A2S extend a point acknowledged in MS2009: expanding the sample of ZIP codes used for the regressions weakens the evidence of a negative correlation. The baseline income-credit regressions in MS2009 use less than 10 percent of the ZIP codes in the United States (approximately 3,000 out of more than 40,000 total U.S. ZIP codes). Omitted from the main sample are ZIP codes that do not have price-index data or that lack credit-bureau data.8 MS2009 acknowledge that if one relaxes the restriction related to house-price data, the negative correlations weaken. Our chart 1 conveys this information with the correlation coefficients depicted in red, which are even closer to zero. A2S go farther to show that if the data set also includes ZIP codes that lack credit-bureau data, the negative correlation and regression coefficients become positive.
But perhaps a deeper contribution of A2S is to remind the researchers that outstanding questions about the housing boom should be attacked with individual-level data. No one doubts that credit expanded during the boom, especially to subprime borrowers. But how much of the aggregate increase in credit went to subprime borrowers, and how did factors like income, credit scores, and expected house-price appreciation affect both borrowing and lending decisions? Even under the best of circumstances, it is hard to study these questions with aggregate data, as MS2009 did. People who take out new-purchase mortgages typically move across ZIP-code boundaries. Their incomes and credit scores may be different than those of the people who lived in their new neighborhoods one, two, or seven years before. A2S therefore argue for the use of HMDA individual-level income data so that credit allocation can be studied at the individual level. This use has been criticized by Mian and Sufi, who believe that fraud undermines the quality of the individual-level income data that appear in HMDA records. We should take these criticisms seriously. But the debate over whether lending standards are best studied with aggregate or individual-level data should take place with the understanding that aggregate data on incomes and credit may not be as informative as previously believed.
2 Data on new-purchase mortgage originations come from records generated by the Home Mortgage Disclosure Act (HMDA). Average income at the ZIP-code level is tabulated in the selected years by the Internal Revenue Service.
3 Growth rates used in the regressions are annualized. The uneven lengths of the sample periods are necessitated by the sporadic availability of the IRS income data, especially early on. The 1991 data are no longer available because IRS officials have concerns about their quality.
4 Chart 2 includes data for both prime and subprime loans. The representativeness of the Black Knight/LPS data improves markedly in 2005, so LPS loans originated before that year may not be representative of the universe of mortgages made at the same time. For other evidence specific to the performance of subprime loans made in 2006 and 2007, see Figure 2 of Christopher Mayer, Karen Pence, and Shane M. Sherlund, "The Rise in Mortgage Defaults," Journal of Economic Perspectives (2009), and Figure 1 of Yuliya Demyanyk and Otto Van Hemert, "Understanding the Subprime Mortgage Crisis," Review of Financial Studies (2009). For data on the performance of GSE loans made in 2006 and 2007, see Figure 8 of W. Scott Frame, Kristopher Gerardi, and Paul S. Willen, "The Failure of Supervisory Stress Testing: Fannie Mae, Freddie Mac, and OFHEO," Atlanta Fed Working Paper (2015).
5 MS2009 often refer to their regression coefficients as "correlations" in the text as well as in the relevant tables and figures, but these statistics are indeed regression coefficients. Note that in the fourth table of the supplemental online appendix, one of the "correlations" exceeds 1, which is impossible for an actual correlation coefficient.
6 Because a regression coefficient from a univariate regression is Cov(X,Y)/Var(X), multiplying this coefficient times StdDev(X)/StdDev(Y) gives Cov(X,Y)/StdDev(X)*StdDev(Y), which is the correlation coefficient. Here, the Y variable is ZIP-code–level credit growth, demeaned from county-level averages, while X is similarly demeaned income growth. As measures of the standard deviations, we use the within-county standard deviations displayed in Table I of MS2009. Specifically, we use the within-county standard deviation of "mortgage origination for home purchase annual growth" calculated over the 1996–02 and 2002–05 periods (0.067 and 0.15, respectively) and the within-county standard deviation of "income annualized growth" over the 1991–98, 1998–2002, 2002–05, and 2005–06 periods (0.022, 0.017, 0.031, and 0.04, respectively). Unfortunately, the time periods over which the standard deviations were calculated do not line up exactly with the time periods over which the regression coefficients were calculated, so our conversion to correlation coefficients is an approximation.
7 It is true that the regression coefficients in the MS2009 coefficients often have large t-statistics, so one may argue that ZIP-level income growth has sometimes been a statistically significant determinant of ZIP-level credit growth. But the low correlation coefficients indicate that income growth has never been economically significant determinant of credit allocation within counties. It is therefore hard to know what is driving the income-credit correlation featured in MS2009, or what may be causing its sign to fluctuate.
8 Though house prices and credit bureau data are not required to calculate a correlation between income growth and mortgage-credit growth, the authors use house prices and credit bureau data in other parts of their paper.
January 14, 2015
The Effectiveness of Restrictions of Mortgage Equity Withdrawal in Curtailing Default: The Case of Texas
As an economist who has studied the causes of the recent mortgage default and foreclosure crisis, I am often asked how to design policies that will minimize the likelihood of another crisis. My typical response to such a question is that one of the most effective ways of lowering mortgage defaults would be to limit borrower leverage by either increasing down payment requirements at the time of purchase or limiting home equity withdrawal subsequent to purchase.
The reason behind my belief is twofold. First, economic theory tells us that being in a situation of negative equity (where the remaining balance of the mortgage is greater than the market value of the property) is a necessary condition for default and foreclosure. Homeowners with positive equity will almost always have a financial incentive to sell their homes instead of suffering through the foreclosure process, while borrowers who are “under water” have a difficult time refinancing or selling (since they would need to have enough cash at closing to cover the difference between the outstanding balance of the mortgage and the sale price/appraisal of the house) and have less of a financial incentive to continue paying the mortgage. Second, numerous empirical studies in the literature have confirmed the theory by documenting a strong positive correlation between the extent of negative equity and the propensity to default on one’s mortgage.
New evidence on preventing defaults
An important new paper by Anil Kumar, an economist at the Federal Reserve Bank of Dallas, provides new evidence that shows just how effective restricting leverage can be in preventing mortgage defaults. His paper confirms many of the findings in previous studies that have shown a positive relationship between negative equity and default. However, it goes a step further by using plausibly random variation in home equity positions created by a government policy that placed explicit restrictions on home equity withdrawal.
Kumar's paper is a significant contribution to the literature because it seems to overcome a serious identification issue that has plagued most empirical studies on the topic. The major challenge is that a homeowner can partially control his or her equity position through decisions about initial down payments on purchase mortgages and decisions about cash-out refinancing and home equity loans or lines of credit subsequent to purchase. As a result, it's unclear whether homeowners with more negative equity are more likely to default because of their worse equity positions or because of other reasons (unobserved by the researcher) that happen to be correlated with the decision to put less money down at purchase or to extract more equity over time.
Both theory and empirical evidence tell us that more impatient individuals tend to borrow more and are more likely to default on their debts. Thus, it might simply be the case that more impatient borrowers who are less likely to repay any type of debt choose to put less money down and extract more equity over time, creating the observed correlation between negative equity and the propensity to default. To put it in the language of econometrics, there are both selection and treatment effects that could be driving the correlation that we see in the data, and the policy implications of restricting borrower leverage are likely very different depending on which cause is more important.
Do home equity restrictions cause lower default rate?
The paper focuses on a policy enacted in the state of Texas that placed severe restrictions on the extent of home equity withdrawal. The Texas constitution, enacted in 1876, actually prohibited home equity withdrawal. The prohibition was eventually lifted in 1997 and the restrictions were further relaxed in 2003, but even in the post-2003 period, Texas law placed serious limits on equity withdrawal, which remain in effect today.1 Subsequent to purchase, a borrower cannot take out more than 50 percent of the appraised value of the home, nor exceed 80 percent of total loan-to-value (LTV). For example, if a borrower owned a home worth $200,000 and had an outstanding mortgage balance of $140,000, the borrower would be allowed to take out only $20,000 in a cash-out refinance. It is important to note that this LTV restriction does not bind at the time of purchase, so a homebuyer in Texas could take out a zero-down-payment loan, and thus begin the homeownership tenure with an LTV ratio of 100 percent (we will come back to this issue later).
Here's a nice quote in the April 4, 2010, issue of the Washington Post crediting the cash-out restriction for Texas weathering the foreclosure crisis better than many areas of the country.
But there is a broader secret to Texas's success, and Washington reformers ought to be paying very close attention. If there's one thing that Congress can do to help protect borrowers from the worst lending excesses that fueled the mortgage and financial crises, it's to follow the Lone Star State's lead and put the brakes on "cash-out" refinancing and home-equity lending.
At first glance, the data suggest that such a sentiment may be correct. In the figure below, we display subprime mortgage serious delinquency rates (defined as loans that are at least 90 days delinquent) for Texas and its neighbors (Arkansas, Louisiana, New Mexico, and Oklahoma). We focus on the subprime segment of the market because these are the borrowers who are more likely to be credit-constrained and thus more likely to extract home equity at any given time. It is apparent from the figure that Texas had the lowest subprime mortgage delinquency rates over most of the sample period. While the paper uses a slightly different data set, a similar pattern holds (see Figure 1 in the paper). The figure is certainly compelling and suggests that the home equity withdrawal restrictions in Texas had an important effect on default behavior, but a simple comparison of aggregate default rates across states really doesn’t tell us whether the policy had a causal impact on behavior. There could be other differences between Texas and its neighboring states that are driving the differences in default rates. For example, house price volatility over the course of the boom and bust was significantly lower in Texas compared to the rest of the country, which could also explain the differences in default rates that we see in the figure.
The paper uses a relatively sophisticated econometric technique called "regression discontinuity" to try to isolate the causal impact of the Texas policy on mortgage default rates. We won't get into the gory details of the methodology in this post, so for anyone who wants more details, this paper provides a nice general overview of the technique. Essentially, the regression discontinuity approach implemented in the paper compares default rates over the 1999–2011 period in Texas counties and non-Texas counties close to the Texas borders with Louisiana, New Mexico, Arkansas, and Oklahoma while controlling for potential (nonlinear) trends in default rates that occur as a function of distance on each side of the Texas border. The paper also controls for other differences across counties that are likely correlated with mortgage default rates (such as average house price appreciation, average credit score, and more). The idea is to precisely identify a discontinuity in default rates at the Texas border caused by the restrictions on home equity withdrawal in Texas. This strikes us as a pretty convincing identification strategy, especially in light of the fact that information on actual home equity withdrawal is not available in the data set used in the paper.
The estimation results of the regression discontinuity specification show that the equity restriction policy in Texas lowered overall mortgage default rates over the 13-year period by 0.4 to 1.8 percentage points depending on assumptions about sample restrictions (including counties within 25, 50, 75, or 100 miles of the border) and functional form assumptions for the “control function” (that is, whether distance to the border is assumed to be a linear, quadratic, or cubic polynomial). At first glance, this may not seem like a large effect, but keep in mind that the average mortgage default rate over the entire sample period was only slightly above 3 percentage points in Texas and 4 percentage points in the neighboring states. The paper also restricts the sample to subprime mortgages only and finds significantly larger effects (2 to 4 percentage points), which makes sense. We expect subprime mortgage borrowers to be affected more by the equity restriction since they are more likely to withdraw home equity.2 The paper implements a battery of robustness checks to make sure that the results aren’t overly sensitive to functional form assumptions and adds controls for other types of state-level policy differences. Based on the results of those tests, the findings appear to be quite stable.
But is it a good policy?
So the paper appears to confirm what previous research on the relationship between equity and mortgage default has found, although it uses methods that aren’t quite as clean as the regression discontinuity approach employed in this analysis. However, it doesn’t mean that such a law change is necessarily good policy. While it seems to be effective in reducing defaults, it may also have some real costs. The most obvious one is the decrease in the volume of low-cost secured credit that many borrowers used to improve their circumstances during the housing boom. An unintended consequence of the policy might have been to push financially distressed households into higher-cost credit markets like credit cards or payday loans. A second drawback of the policy may have been that it increased homeowner leverage at the time of purchase. As there were no restrictions on LTV ratios at the time of purchase, many homebuyers may have decided to make lower down payments, knowing that their access to equity would be restricted in the future. It’s also possible that this may have resulted in a larger volume of subprime mortgage lending in Texas. Households with relatively high credit scores who could have obtained a prime mortgage with significant down payments (say, 20 percent), may have turned to the subprime segment of the market, where they could obtain loans with low down payments but with much more onerous contract terms.
While it’s not clear whether the actual Texas policy of restricting home equity extraction is welfare-improving, it might seem from the research that restricting borrower leverage is an effective way to reduce mortgage default rates. But limiting borrower leverage is very unpopular. In fact, it probably isn’t too much of an exaggeration to say that the vast majority of market participants are adamantly opposed to such policies. After all, it is perhaps the only policy upon which both the Center for Responsible Lending (CRL) and the Mortgage Bankers Association (MBA) share the same negative view.3 Thus, while such policies have been adopted in other countries, don’t expect to see them adopted in the United States any time soon.4 To the contrary, policy is more likely to go in the opposite direction as evidenced by the Federal Housing Finance Agency’s announcement to relax down payment requirements for Fannie Mae and Freddie Mac.
By Kris Gerardi, financial economist and associate policy adviser at the Federal Reserve Bank of Atlanta
1 Before 1998, both home equity lending (loans and lines of credit) and cash-out refinancing were explicitly prohibited in Texas. A 1997 constitutional amendment relaxed this ban by allowing for closed-end home equity loans and cash-out refinancing as long as the combined LTV ratio did not exceed 80 percent of the appraised value (among a few other limitations that are discussed in the paper). In 2003, another constitutional amendment passed that further allowed home equity lines of credit for up to 50 percent of the property’s appraised value, although still subject to a cap on the combined LTV ratio of 80 percent.
2 The effects are actually smaller for the subprime sample when compared to the average default rate over the entire sample period, since the average rate is significantly higher in the subprime segment of the market (10 percent subprime default rate compared to the 3 percent overall default rate in Texas).
April 25, 2014
Two Views of the Involvement of Credit Rating Agencies in the Mortgage Crisis
A lot of people have blamed credit rating agencies (CRAs) for helping to cause the mortgage crisis. The report of the Financial Crisis Inquiry Commission (FCIC) labelled CRAs as "key enablers of the crisis," because the exploding mortgage-backed bonds that caused so much trouble could not have been sold without stamps of approval from the CRAs. Commentators often link CRA failings to the fact that they are paid by the issuers of the securities they rate, with the implication that CRAs are thus given incentive to award good ratings to securities that do not deserve them. Indeed, two recent articles by academic economists on this topic come to the same conclusion: financial markets would work better if we scrapped the issuer-pays model in favor of some other way to pay CRAs for their evaluations. But the two articles disagree on why this is so, and understanding the source of this disagreement sheds some harsh light on claims that CRAs should be even partly blamed for the financial crisis in the first place.
Grade inflation in the student-pays model
The first article is a Wall Street Journal op-ed piece by Princeton economist Alan Blinder. Blinder likens the awarding of credit ratings to mortgage-backed securities to his own awarding of letter grades to his Princeton students. "Suppose I proposed to grade my students by a 'student pays' model," Blinder suggests. Such a setup would encourage him to give easy As in hopes of attracting more students and higher pay, and the information in the grades would suffer as a result. "Yet that description comes pretty close to mimicking the way we pay rating agencies," Blinder writes. "Looking back, is it any wonder that so many securities were blessed with undeserved triple-A ratings?"
One interpretation of Blinder's analogy is that college grading works better than securities rating because universities have not adopted the student-pays model. That argument will seem curious to many college instructors, because this model approximates their own compensation arrangements pretty well. Students may not write checks to professors, but they (or their parents) write checks to colleges, who then pay the professors. Instructors whose grades are overly harsh in relation to other courses are likely to see their class sizes dwindle, to the dismay of department chairs facing hard budget constraints. Even if an instructor has no problem attracting students, she may not want grading disparities among courses to distort student decisions on what to study, so she might ease up in her own grading as well. Given the incentives of professors, it is not surprising that grade inflation is debated at many universities, even the very best ones. A December 2013 article in Harvard University's student newspaper, the Crimson, described a faculty meeting at which a professor bemoaned the fact that the most frequently awarded grade at Harvard College is an A-minus. A university dean corrected him: "The median grade in Harvard College is indeed an A-minus," the dean is quoted as saying. "The most frequently awarded grade in Harvard College is actually a straight A." (Disclosure: Harvard's grading policy is of personal interest to two authors of this blog post, who teach intermediate macroeconomics courses there in their spare time.)
Rational employers, rational investors
If the student-pays model leads to grade inflation, then don't we have even more ammunition that the bad incentives inherent in the investor-pays model for CRAs is partly responsible for the mortgage crisis? Not necessarily. For bad CRA incentives to have caused the crisis, two things must be true: one, the incentives must have caused inflated ratings, and two, the investors had to believe the inflated ratings. The second step in this causal chain is open to question. If the investors knew that the issuer-pays model gave incent to the rating agencies to inflate ratings, then rational investors would have taken that information into account when making investment decisions.
The college-grading analogy is again useful here. Consider an employer who is thinking about hiring a recent graduate who received a B-minus in a course that is highly relevant to what the firm does. How should the employer use this information? One option would be for the employer to look up how the student's official university documents define a B-minus—the documents are likely to define a grade in the B range as indicating a better-than-average understanding of the material. But a rational employer who knows the incentives facing American professors would also know that instructors are given cause to inflate grades. The firm could thus surmise that an applicant on the border between a B and a C may actually have a lower-than-average mastery of the subject. In the same way, rational mortgage investors who knew that CRAs had incentive to inflate ratings would have taken those ratings with a grain of salt when evaluating mortgage-backed investments.
Investor rationality plays a prominent role in a second recent piece on CRA incentives, a formal paper by the economists Anil Kashyap and Natalia Kovrijnykh (KK). Because this article is part of the academic economics literature, the authors adopt the fundamental assumption that all actors in the model are rational. As we might expect from our analogy of the job applicant, the rationality assumption makes a big difference when analyzing CRA payment regimes. Consider a situation in which CRAs are paid by the issues of securities, as they are today. Further assume that CRAs receive more money for good ratings than for bad ones. Rational investors in the KK model would realize the ratings are likely to be inflated under this set of incentives and would deflate the ratings accordingly. But if the CRAs are unable to fool investors who know both the CRAs' preferences and their opportunities, then the CRAs might as well tell the truth. KK therefore constrain their attention to equilibria where rating agencies are always truthful.
The revelation principle
In assuming truth-telling, KK are following a long tradition in the modeling of imperfect information. In fact, the assumption that actors with private information tell the truth shows up so often in models of imperfect information that it has a special name: the revelation principle. This principle is useful for modelers because it allows them to focus on equilibria in which the agent with private information has no reason to lie. To be clear, in this situation, the revelation principle does not mean that rating agencies never lie. Rather, it states that any equilibrium in which rating agencies lie is equivalent to one in which they tell the truth. The lying doesn't affect the actions of investors who know the incentives and opportunities of the CRAs, just as inflation of our B-minus student's grade does not lead the employer into an inappropriate hire. Because lying does not encourage agents to take inappropriate actions, it can safely be ignored when thinking through the fundamental aspects of the problem.
The appropriateness of the revelation principle in this context hinges on the ability of mortgage investors to analyze CRA incentives and opportunities and thereby back out the truth. Is this realistic? Ironically, the critics of CRAs provide evidence in support of this view. When Barney Frank alleged that CRA incentives led them to inflate ratings, he was doing exactly the sort of reverse engineering that lies behind the revelation principle. And if legislators could figure out that rating agencies had distorted incentives, why couldn't investors, who were putting up their own money? Indeed, investors should have had much better information about agency incentives than Barney Frank. It turns out that financial intermediaries lost enormous sums on the mortgage-related securities that they purchased and held on their balance sheets (more details on this in the next post). At the same time, they were also large issuers of these securities. Who would know better about the potential for corruption of rating agencies than the financial intermediaries that supposedly corrupted them?
Of course, if the KK model holds that rating agencies always tell the truth, then the model cannot rationalize arguments that CRAs helped cause the crisis by misleading investors. Indeed, the revelation principle makes it hard to rescue any story about untruthful CRAs. What if credit rating agencies had private information about their incentives, in addition to private information about their effort and the quality of the securities that they rated? Setting aside the fact that the issuer-pays model of credit ratings was common knowledge in the market, this change to the model has no effect on its outcome. Here again, the revelation principle would imply that CRAs truthfully reveal the private information about their incentives. For investors to be misled, they cannot simply be confused about incentives. Rather, they must believe that the CRAs' incentives were better aligned than they actually were. In our view, that is unlikely.
CRA payment arrangements
We began this post by noting that both of the recent articles on CRA incentives argued against the issuer-pays model. How can KK make this argument if investors in their model are not fooled? The reason involves some subtle implications of exactly how CRAs are paid in different states of the world. In all contracts in KK's issuer-pays regime, CRA pay is contingent on the outcome of the security. That means that if an AAA-rated security defaults, the CRA gets paid less than if the security pays off. To induce effort by the CRA, the spread between the payoffs must be large (that is, the CRA must be paid a lot more when the AAA security is successful compared to when it defaults). Because of limited liability, the CRA's compensation is bounded below by zero when a bond defaults—that is, investors can't demand payment from the CRAs in the default state—so high-powered incentives, which require high average pay, imply that compensation to the CRA in the good state has to be very high. As a result, paying the CRA for high effort can be prohibitively expensive for the issuer, causing the issuer to settle for low-powered incentives instead and thus receiving low effort from the CRA. Even in the low-effort equilibrium, however, CRAs increase the information set of investors and are socially useful.
Going farther, KK show that having the investor rather than the issuer pay the CRA solves the limited-liability problem and thereby raises social welfare. Particularly surprising about this finding is that the investor-pays model is not only good for society, but it is also good for the CRAs! The reason once again involves the revelation principle. In equilibrium, everyone knows both the amount and usefulness of the effort expended by the CRAs in evaluating securities. The larger the CRA's social benefit, the more the CRA gets paid. If KK's model is accurate, then CRAs themselves may lead the way to a better social outcome by encouraging the adoption of the investor-pays model.
While KK's paper includes many specific lessons about potential CRA payment arrangements, the bottom line to emerge from a comparison of the Blinder op-ed and the KK model involves their differing assumptions regarding investor rationality. The KK model illustrates how the revelation principle, which follows from investor rationality, works against the argument that CRAs helped cause the crisis by misleading investors. As long as investors understand the basic structure of the market, then standard models of asymmetric information—of which the KK model is an example— do not predict that investors will experience large and unexpected losses.
You can read the Harvard Crimson article on the magazine's website.
Chris Foote, senior economist and policy adviser at the Federal Reserve Bank of Boston,
Kris Gerardi, financial economist and associate policy adviser at the Federal Reserve Bank of Atlanta, and
By Paul Willen, senior economist and policy adviser at the Federal Reserve Bank of Boston
Real Estate Research Search
- Affordable housing goals
- Credit conditions
- Expansion of mortgage credit
- Federal Housing Authority
- Financial crisis
- Foreclosure contagion
- Foreclosure laws
- Governmentsponsored enterprises
- Homebuyer tax credit
- House price indexes
- Household formations
- Housing boom
- Housing crisis
- Housing demand
- Housing prices
- Income segregation
- Individual Development Account
- Loan modifications
- Monetary policy
- Mortgage crisis
- Mortgage default
- Mortgage interest tax deduction
- Mortgage supply
- Multifamily housing
- Negative equity
- Positive demand shock
- Positive externalities
- Rental homes
- Subprime MBS
- Subprime mortgages
- Supply elasticity
- Upward mobility
- Urban growth