Housing Markets

• “Did Affordable Housing Legislation Contribute to the Subprime Securities Boom?” by Ruben Hernandez-Murillo, Andrea C. Ghent, and Michael T. Owyang. March 2012. SSRN #2022461.
• “Predatory Lending and the Subprime Crisis,” by Sumit Agarwal, Gene Amromin, Itzhak Ben-David, Souphala Chomsisengphet, and Douglas D. Evanoff. April 2012. SSRN #2055889.
• “Why Did So Many People Make So Many Ex Post Bad Decisions? The Causes of the Foreclosure Crisis,” by Christopher L. Foote, Kristopher S. Gerardi, and Paul S. Willen. May 2012. NBER #18082.

When a public crisis occurs, popular commentary usually converges fairly quickly on an “explanation” of how the crisis came about. Subsequent scholarship often undermines those explanations, but too late to dislodge the conventional wisdom and the policies that result. The housing bubble and subsequent financial crisis are becoming a case in point.

In my “Working Papers” column in the last issue, I discussed how many commentators, including former Federal Reserve chairman Paul Volcker, have blamed the 2008 financial crisis and subsequent recession on the 1999 Gramm-Leach-Bliley Act, which eliminated the 1933 Glass-Steagall banking act’s legal barriers between investment and commercial (i.e., traditional deposits and loans) banking and insurance. As part of the column, I reviewed a paper by New York University economist Lawrence White that argued that the high-leveraged lending to the housing sector that fueled the financial crisis would not have been prohibited by either Glass-Steagall or its proposed present-day reincarnation, the Volcker rule. Despite the fact that Gramm-Leach-Bliley had no role in the financial crisis, the need to reimpose Glass-Steagall-like regulation through the so-called Volcker rule is now conventional wisdom, especially since the trading losses of JP Morgan Chase were announced in May of this year and former Citigroup chief executive Sanford Weill announced in July that the separation of investment and commercial banking should be reinstated.

Another prominent component of the conventional wisdom about the housing bubble, particularly for conservatives, is that affordable housing goals imposed by Congress on banks by the Community Reinvestment Act enacted in 1977 and imposed on Fannie Mae and Freddie Mac in 1992 led to the increase in high-risk mortgages offered and their subsequent default. Those goals mandated mortgage lending in census tracts with large minority or low-income populations and to minority and low-income people regardless of their residential location.

Hernandez-Murillo, Ghent, and Owyang do not dispute the role played by private label mortgage-backed securities (PLMBS) in the downfall of Fannie and Freddie. The two mortgage giants bought PLMBS heavily at the height of the housing boom, believing they would yield large long-term profits. Instead, though PLMBS accounted for only one-third of Fannie Mae’s business, they produced more than 70 percent of its losses through 2010. Rather, the authors’ goal is to test whether borrowers and census tracts that fulfilled affordability goals disproportionately received loans that then subsequently defaulted. They do this using the arbitrary legal divisions that distinguish loans that achieve affordability goals from loans that do not. Loans that “just” qualify as achieving affordability goals are compared to loans that “just fail” to qualify. If affordability goals affected the behavior of the agencies and loan suppliers, then the population of loans that just qualify should be larger than the population of loans that just fail to qualify.

The authors identify two CRA and five Fannie and Freddie goals:

• Loans to borrowers living in census tracts with a median income of 80 percent or less than median Metropolitan Statistical Area (MSA) income.
• Loans to borrowers with incomes of 80 percent or less of the median MSA income.
• Loans to borrowers living in census tracts with a minority population of 30 percent or more and median income of 120 percent or less than MSA median income.
• Loans to borrowers living in census tracts with a median income 90 percent or less than MSA median income.
• Loans to borrowers with incomes of 60 percent or less than the median MSA income.
• Loans to borrowers with incomes of 80 percent or less than the median MSA income and who live in census tracts with a median income 80 percent less than MSA median income.
• Loans to borrowers with incomes of 100 percent or less than the median MSA income.

If any of the affordable housing goals affected loan decisions, we would expect to see discontinuities in originations, interest rates, or default rates related to the median income of the census tract relative to the MSA median, the minority population in the census tract, or the ratio of individual borrower income to median MSA income. For each of the goals, the authors conduct several tests: Are there more loans per capita in census tracts that just qualify relative to tracts that just fail to qualify? Is there a discontinuity in interest rates in census tracts that just qualify relative to tracts that just do not? Is there a discontinuity in the default rates in census tracts that just qualify relative to tracts that just fail?

The data for the tests consist of all 722,000 subprime securitized mortgages in metropolitan California and Florida in 2004–2006. Seventy percent of the loans satisfy some affordable housing goal. The average loan amount was $296,000. The average borrower had 173 percent of median MSA income and lived in a 47 percent minority census tract of below-average income.

The authors could find no statistically significant discontinuities. This result is robust to those loans with full documentation and different bandwidths—i.e., 1 percent, 2 percent, and 5 percent above and below the relevant cutoffs. Affordable housing goals appear not to be responsible for the PLMBS that Fannie and Freddie purchased.

For those on the political left, the conventional wisdom is that the housing bubble was the result of “predatory lending”—mortgages issued on terms that seemed attractive but ultimately were unfair and financially crippling to borrowers. Agarwal et al. analyze the effects of such predatory lending in Chicago. In 2005, the Illinois legislature enacted legislation to curtail those lending practices. The program was initially conceived as a four-year pilot program primarily for the South Side neighborhoods of Chicago, and it mandated counseling for all those borrowers with FICO scores (a measure of creditworthiness) less than 621 and those between 621and 650 if they chose high-risk products. The law also mandated counseling regardless of FICO score if the mortgage product included negative amortization, prepayment penalties, or closing costs higher than 5 percent.

The intervention had large effects on lenders and throughput. Purchase loan applications decreased by 18.6 percent and refinancing applications by 46.6 percent. Actual purchase mortgages declined 29.2 percent and mortgage refinancing declined by 48.5 percent. Of the 49 lenders offering loans in the area before the program, only 25 remained during the program. (Because of the drastic effects on applicants and lenders, community activists and lenders succeeded in suspending the program after only 20 weeks of operation on January 17, 2007.)

But the large effect on throughput had no effect on the subsequent default rate 18 months after origination. Relative to a control area, the default rate on the treated group declined a statistically insignificant 3.6 percent.

Both of these papers add to the evidence presented in my Spring 2011 “Working Papers” column that neither affordable housing goals nor subprime loans were important explanations of the housing bubble.

Another important component of conventional explanations of the foreclosure crisis involves informed insiders selling overpriced houses and associated financial products to uninformed outsiders. Foote, Gerardi, and Willen present 12 facts that refute this narrative. They argue that the explanation most consistent with the facts is one in which overly optimistic beliefs about ever-climbing housing prices were held by investors, borrowers, and lenders alike. If the insider/​outsider story were true, then better regulation and information might prevent future bubbles, but because collective beliefs are the problem, prevention is much more difficult.

The most interesting of their 12 arguments are as follows:

The exploding Adjustable Rate Mortgage (ARM) theory has been central to many narratives, including that given by Shelia Blair of the Federal Deposit Insurance Corporation. But data about various vintages of “2/28” ARMs (loans with an initial low interest rate that “resets” to a market rate two years later) and default rates show no relationship between reset and foreclosure. Only 12 percent of borrowers involved in foreclosures between 2007 and 2010 were making payments higher than the initial rate when they defaulted.

Many argue that “mortgage innovation”—that is, the recent development of “exotic” types of mortgages—was the problem. But these products weren’t all that new or exotic. The option ARM, for example, was invented in 1981 and accounted for one-third of all originations in California by 1996. “NoLos,” in which the borrower had to provide little if any evidence of creditworthiness, were also old-hat; by 1990, 35 percent of loans were no or low documentation. Low down payments mortgages were started long ago by the Veterans Administration and Federal Housing Administration, and accounted for half the market in the 1950s. Mortgage-backed securities (MBSs) and collateralized debt obligations (CDOs) that were used to finance mortgages, as well as other complex financial products, had been around for decades.

The originate-to-distribute model of mortgages, in which banks initially make the loan but then promptly sold off the asset to investors, was likewise not new—the authors quote from a 1959 discussion of the model. Its use grew dramatically in the 1970s when unanticipated inflation decimated the traditional originate-and-hold model for fixed-rate mortgages. By the late 1980s, 50 percent of savings-and-loan mortgages were sold rather than held by the originating institution.

It’s difficult to argue that the financial bust was the product of savvy industry insiders taking advantage of gullible and naïve outside investors. Mortgage investors had lots of information about the assets available to purchase. In fact, much of the information we have on loans comes from the disclosures given to investors before they bought. Lehman and UBS prospectuses illustrated the expected performance in a housing price meltdown scenario and have been quite accurate.

Moreover, mortgage market insiders were the biggest losers from the bust. Bear Stearns executives were the major investors in their two hedge funds that went bankrupt. The biggest winners were outsiders who made large bets against mortgages in the credit default swap (CDS) market. Ironically, most analysts at the time were very bullish on subprime mortgages and the negative CDS bets were made on the basis of a regression-to-the-mean trend analysis rather than any insights about underwriting defects.

It should be noted that, in the financial collapse, AAA-rated bonds did not turn out to be toxic. Only 10 percent of those bonds suffered losses. However, AAA-rated CDOs were toxic; investors suffered losses on 90 percent of AAA-rated CDO tranches. Both CDOs and MBSs were rated by the same firms, so why the different outcomes? MBSs were rated using structural models to estimate how correlated defaults would be if housing prices changed; CDOs were not. The analysts who studied the CDOs were bond analysts who simply used historical correlations, which for corporate bonds have proved to be excellent predictors of default even today. Thus the information asymmetries were not between outsiders and insiders, but between two different groups of analysts within the investment firms

Many believe that the rising popularity of lower down payment loans last decade stimulated house prices. This belief conflicts with a fundamental theorem of finance: that financial innovation improves risk sharing, decreases precautionary savings, and lowers asset prices. The Achilles heel of all rational financial innovation models is that temporary innovations cannot cause prices to change. Thus the models must assume that permanent changes occur and then have surprising exogenous reversals to account for booms and busts and satisfy their own internal logic.

A recent rational financial innovation model to explain the housing boom and bust assumes a down payment collapse from 25 to 1 percent to generate its results. But down payments have been low for some time, so the applicability of this model is suspect. And housing prices did not collapse because financial innovations were reversed beforehand. Rather, financial innovations ceased to be used because housing prices collapsed.

Oil Markets

• “The Incidence of an Oil Glut: Who Benefits from Cheap Crude Oil in the Midwest?” by Severin Borenstein and Ryan Kellog. June 2012. NBER #18127.

The unexpected boom in oil production in the Midwest from shale rock has created strains on the oil pipeline transportation system. The long-term decline in domestic production over the last 50 years has resulted in a pipeline system that is designed to transport imported crude oil from the Texas Gulf Coast inland to Cushing, Okla., where various pipelines meet and tank farm inventory facilities are available. The same system cannot ship crude oil to the Gulf Coast, however. Thus the ironic temporary result of increased domestic production in the shale areas in the United States and the tar sands area in Canada is that once the oil reaches Cushing, it then cannot reach the world market. It can only be shipped to inland refineries. The result is a mismatch between too much domestic supply and too little domestic refinery demand, and thus a price discount relative to crude oil prices in the rest of the world market.

Even though West Texas Intermediate (WTI) crude oil prices are lower than world prices, Midwestern gasoline prices are not lower because refined product pipelines are not capacity-constrained and the output of Midwestern refineries is not sufficient to meet Midwest product demand. Thus the marginal source of gasoline in Midwestern markets uses crude oil priced at world levels rather than the lower-priced WTI. This means that though crude prices in the Midwest are low because of the Cushing blockage, gas prices are not.

Projects are underway to reverse the flow of crude pipelines so that the excess oil in Cushing can travel to the Gulf Coast and find its way onto world markets. The result will be an increase of WTI prices to world levels, but no change in the price of gasoline, whose price already reflects world crude prices. The refineries that use WTI will see a reduction in their profit margins. The expansion of the Keystone Pipeline from Canada, if it occurs, will result in an analogous increase in the price paid for Canadian crude.

Chrysler Bankruptcy

• “Was the Chrysler Reorganization Different?” by Mark J. Roe and Joo-Hee Chung. July 2012. SSRN #2103276.

In the 2009 federal government–sponsored reorganization of Chrysler, secured creditors of the company received only 29 cents on the dollar while unsecured but government-favored creditors (mostly pension and health benefits) received full payment. Some commentators said this violated the rule of law, but others said that the debtor-in-possession (in this case the government) gets to set the rules and that this case “didn’t turn anything upside down.”

The authors argue that priority mischief among creditors in a bankruptcy proceeding can occur when some debt is carried through the reorganization rather than a straight cash sale for assets with all debts extinguished. Thus the percentage of debt that passes through the reorganization process rather than being extinguished is a measure of the differential treatment of some debtors relative to others. The authors gathered data on this ratio for 63 large ($100 million or more in assets) bankruptcies prior to Chrysler and compared various descriptive statistics between Chrysler and the others.

In the Chrysler case, more than half of the preexisting liabilities (health and pension funds) were carried through to the new entity. For the 63 other bankruptcies, the modal and median values of debt carried through were zero. In the Chrysler case, the ratio of total debt assumed to purchase price was 90 percent. For the other 63 the mode was 0, the median was 3 percent, and the average was 23 percent. Even among the subset of the 63 cases that involved high pension obligations, Chrysler was different.

So was the Chrysler reorganization different? The answer appears to be an emphatic yes.