Some Observations on Our Mortgage Market
Policy options should be informed by facts. A few facts, which I believe are directly relevant to the state of our mortgage markets, particularly the trend in foreclosures and delinquencies are as follows:
- The vast majority of underwater borrowers are current on their mortgages. Even the majority of deeply underwater borrowers are current. For prime borrowers with loan-to-values (LTV) over 125%, over 75% are current. Over half of deeply underwater subprime borrowers are current. (Fitch)
- GSE underwater borrowers are also preforming, with almost 80 percent current. The GSEs’ book of underwater loans has actually seen the percent current increasing over the last year.
- GSE loans display a smaller percentage (9.9%) underwater than loans in private label securities (35.5% underwater).
- According to Fannie Mae’s National Housing Survey only about 10% surveyed believed it was appropriate for underwater borrowers to simply “walk away”. While higher than I would prefer, this does indicate that the risk of widespread strategic default is limited.
- Credit quality of the borrower continues to be the primary predictor of default. For borrowers with FICOs in excess of 770, of those deeply underwater (125% LTV) 85% are still current. (Fitch)
- About a fifth of subprime borrowers who have significant equity (LTV
- Total delinquencies are down over 25% from the peak in January 2010, having declined from 10.97% to 7.97% in January 2012. (LPS)
- Over 40% of loans in foreclosure are over 2 years past due. These loans will likely never cure. Only 19% of loans in foreclosure are less than 8 months past due. No one can say, with a straight face, that foreclosures, in general, are happening “too fast”.
- Almost half of loans, currently entering foreclosure, were previously in foreclosure, that is they are “repeat foreclosures”. (LPS)
- The rate of new problem loans, those newly seriously delinquent that were current 6 months previous, peaked in Spring 2009, when the economy was hitting bottom, and have been steadily declining since.
- Including distressed transactions, the peak-to-current change in the national HPI (from April 2006 to January 2012) was ‑34.0 percent. Excluding distressed transactions, the peak-to-current change in the HPI for the same period was ‑24.2 percent. (CoreLogic)
The last point is particularly relevant, as the number of underwater borrowers greatly depends upon current home values. If home values are based upon distressed transactions, then the number of underwater borrowers would be far greater than if one excludes distressed sales. There is some reason to believe the distressed sales are not representative of the overall market, for instance they are likely to have seen greater physical deterioration.
State of the Housing Market
The U.S. housing market remains weak, with both homes sales and construction activity considerably below trend. Despite sustained low mortgage rates, housing activity has remained sluggish in 2011. Although construction activity picked up in 2001, housing starts are still below half the levels seen in 2007. In fact I believe it will be at least until 2015 until we see construction levels approach those of the boom. In addition to the 4.7 percent decline in existing home prices in 2011, we are likely to see additional, but small, declines in 2012. Consensus estimates run around a 3 percent decline in home prices for 2012.
Housing permits, on an annualized basis, increased 0.7 percent from December 2010 to January 2011 (671,000 to 676,000). Permits for both single family units and permits for larger multifamily properties (5+ units) increased slightly, but permits for smaller multifamily units fell 4.2%. Single family permits increased from 441,000 to 445,000 in December. Permits for 2–4 unit properties fell (24,000 to 23,000) in January. Permits for 5+ units climbed to 206,000 in January from 204,000 in December.
According to the Census Bureau, January 2012 housing starts were at a seasonally adjusted annual rate of 699,000, up slightly from the December level of 689,000. Overall starts are up, on an annualized level, from 2011’s 610,700 units. This increase, however, is mostly driven by a jump in multifamily starts, as single-family starts decreased slightly. Total residential starts continue to hover at levels around a third of those witnessed during the bubble years of 2003 to 2004.
As in any market, prices and quantities sold in the housing market are driven by the fundamentals of supply and demand. The housing market faces a significant oversupply of housing, which will continue to weigh on both prices and construction activity. The Federal Reserve Bank of New York estimates that oversupply to be approximately 3 million units. Given that annual single family starts averaged about 1.3 million over the last decade, it should be clear that despite the historically low current level of housing starts, we still face a glut of housing. NAHB estimates that about 2 million of this glut is the result of “pent-up” demand, leaving at least a million units in excess of potential demandi. Add to that another 1.6 million mortgages that are at least 90 days late. My rough estimate is about a fourth of those are more than two years late and will most likely never become current.
The nation’s oversupply of housing is usefully documented in the Census Bureau’s Housing Vacancy Survey. The boom and bust of our housing market has increased the number of vacant housing units from 15.6 million in 2005 to a current level of 18.4 million. The rental vacancy rate for the 4th quarter of 2011 declined to 9.4% after increasing to 9.8% the previous quarter, although this remains considerably above the historic average. The decline in rental vacancy rates over the past year has been driven largely by declines in suburban rental markets. The vacancy rate for newly constructed rental units is approaching the rate for old construction, but for newly constructed homeowner units it remains considerably higher than old construction.
The homeowner vacancy rate, after increasing from the 2nd and 3rd quarters of 2010 to the 4th quarter of 2010, declined slowly over the year 2011 to reach 2.3 percent last quarter, a number still in considerable excess of the historic average.
The homeowner vacancy rate, one of the more useful gauges of excess supply, differs dramatically across metro areas. At one extreme, Greensboro, NC has an owner vacancy rate of well over 6 percent, whereas El Paso, Texas has a rate of 0 percent. Other metro with excessive high owner vacancy rates include: Dayton, OH (6,2); Las Vegas (5.5); Columbia, SC (5.1); New Orleans (4.6); and Phoenix (3.6). Relatively tight owner markets include: Albany, NY (0.0); Norwalk, CT (0.2); and Tucson, AZ (0.3).
The number of vacant for sale or rent units has increased, on net, by around 3 million units from 2005 to 2011. Of equal concern is that the number of vacant units “held off the market” has increased by about 1.5 million since 2005. In all likelihood, many of these units will re-enter the market once prices stabilize.
The 4th quarter 2011 national homeownership rate fell to 66.0 percent, which is approximately where it was in 1997, effectively eliminating all the gain in the homeownership rate over the last 12 years. Declines in the homeownership rate were the most dramatic for the youngest homeowners, while homeownership rates for those 55 and over were generally stable or even increasing. This should not be surprising given that the largest increase in homeownership rates was among the younger households and that such households have less attachment to the labor market than older households. Interestingly enough, the decline in homeownership was higher among households with incomes above the median than for households with incomes below the median, which held steady.
Homeownership rates declined across the all Census Regions except for the Northeast (which held steady), the steepest decline was in the West, followed by the Midwest. The South witnessed the smallest decline in homeownership since the bursting of the housing bubble.
Homeowner vacancy rates differ dramatically by type of structure, although all structure types exhibit rates considerably above historic trend levels. For 4th quarter 2011, single-family detached homes displayed an owner vacancy rate of 2.0 percent, while owner units in buildings with 10 or more units (generally condos or co-ops) displayed an owner vacancy rate of 8.3 percent. Although single-family detached constitute 95 percent of owner vacancies, condos and co-ops have been impacted disproportionately. Over the last year homeowner vacancy rates have declined slightly for single-family structures but more dramatically for condos or co-ops, albeit from a much higher level.
Owner vacancy rates tend to decrease as the price of the home increases. For homes valued between $100,000 and $150,000 the owner vacancy rate is 2.5 percent, whereas homes valued over $200,000 display vacancy rates of about 1.3 percent. The clear majority, almost 63%, of vacant owner-occupied homes are valued at less than $300,000. Owner vacancy rates are also the highest for the newest homes, with new construction displaying vacancy rates twice the level observed on older homes.
While house prices have fallen considerably since the market’s peak in 2006 — over 23% if one excludes distressed sales, and about 31% including all sales — housing in many parts of the country remains expensive, relative to income. At the risk of oversimplification, in the long run, the size of the housing stock is driven primarily by demographics (number of households, family size, etc), while house prices are driven primarily by incomes. Due to both consumer preferences and underwriting standards, house prices have tended to fluctuate at a level where median prices are approximately 3 times median household incomes. Existing home prices, at the national level, are close to this multiple. In several metro areas, however, prices remain quite high relative to income. For instance, in San Francisco, existing home prices are almost 8 times median metro incomes. Despite sizeable decline, prices in coastal California are still out of reach for many families. Prices in Florida cities are generally above 4 times income, indicating they remain just above long-run fundamentals. In some bubble areas, such as Phoenix and Las Vegas, prices are below 3, indicating that prices are close to fundamentals. Part of these geographic differences is driven by the uneven impact of federal policies.
Household incomes place a general ceiling on long-run housing prices. Production costs set a floor on the price of new homes. As Professors Edward Glaeser and Joseph Gyourko have demonstratedii, housing prices have closely tracked production costs, including a reasonable return for the builder, over time. In fact the trend has generally been for prices to about equal production costs. In older cities, with declining populations, productions costs are often in excess of replacement costs. After 2002, this relationship broken down, as prices soared in relation to costs, which also included the cost of landiii. As prices, in many areas, remain considerably above production costs, there is little reason to believe that new home prices will not decline further.
It is worth noting that existing home sales in 2010 were only 5 percent below their 2007 levels, while new home sales are almost 60 percent below their 2007 level. To a large degree, new and existing homes are substitutes and compete against each other in the market. Perhaps the primary reason that existing sales have recovered faster than new, is that price declines in the existing market have been larger. Again excluding distressed sales, existing home prices have declined 23 percent, whereas new home prices have only declined only about 10 percent. I believe this is clear evidence that the housing market works just like other markets: the way to clear excess supply is to reduce prices.
State of the Mortgage Market
According to the Mortgage Bankers Association’s National Delinquency Survey, the delinquency rate for mortgage loans on one-to-four-unit residential properties decreased to a seasonally adjusted rate of 7.58 percent of all loans outstanding for the end of the 4th quarter 2011, 41 basis points down from 3rd quarter 2011 and down 67 basis points from one year ago.
The percentage of mortgages on which foreclosure proceedings were initiated during the fourth quarter was 0.99 percent, 9 basis points down from 2011 Q3 and down 28 basis points from 2010 Q4. The percentage of loans in the foreclosure process at the end of the 4th quarter was 4.38 percent, down slightly at 5 basis points from 2011 Q3 and 26 basis points lower than 2010 Q4. The serious delinquency rate, the percentage of loans that are 90 days or more past due or in the process of foreclosure, was 7.73 percent, a decrease of 16 basis points from 2011 Q3, and a decrease of 87 basis points from 2010 Q4. The combined percentage of loans in foreclosure or at least one payment past due was 12.53 percent on a non-seasonally adjusted basis, a 10 basis point decrease from 2011 Q3 and 107 basis points lower than 2010 Q4.
Extent of Negative Equity
Despite that the vast majority of underwater borrowers continue to pay their mortgages, concerns about negative equity dominate policy debates surrounding the mortgage market. According to CoreLogic, 11.1 million, or 22.8 percent, of all residential properties with a mortgage (recall that about a third of owners own their homes free and clear) are in a negative equity position. This situation is highly concentrated in terms of geography. The top five states (NV, AZ, FL, MI and GA) display an average negative share of 44.3 percent. The remaining states have a combined average negative share of 15.3 percent. Any taxpayer efforts to reduce negative equity would largely be a transfer from the majority of states to a very small number.
Of those with negative equity, 4.4 million have both first and second mortgages. The average LTV of these borrowers is 138 percent, implying that in the event of a foreclosure, the second lien would likely have little, if any value. Efforts to modify first liens only, or to modify firsts and seconds in proportion, are, in effect, transfer from the first lien holder to the second. We should reject such transfers, as they violate the basic principles of contract and property, and require all seconds to be eliminated before any loss are taken on first liens.
While less than half of those with negative equity have second liens, those that do constitute a far greater share of negative equity borrowers. Those with both first and second liens display a negative equity share of 39 percent, twice that for borrowers with a first lien only. Of the estimated $717 billion in negative equity just over half is from borrowers with both first and second liens. My estimate is that about a fourth of negative equity is in the form of second liens.
For pressing importance for policymakers is the fact that just under 2 million FHA borrowers are underwater. The vast majority of these borrowers took out mortgages since the beginning of the housing bust. Just under a third of all FHA borrowers that took loans out since the housing bust are now underwater. That giving borrowers near-zero equity loans in a deflating housing market would result in widespread negative equity should have been obvious (it was to me), but that is of course “water under the bridge”. The important issue now is mitigating that risk. As FHA’s 203(b) program does have the power of full recourse, I urge FHA to advertise that power and implement programs to exercise it. In addition delinquent FHA borrowers should be reported immediately the to IRS, so that any tax refunds can be used instead to off-set losses to the taxpayer. My estimates are that FHA is likely to require between $10 and $50 billion over the next five to six years in order to honor all claims.
New York Federal Reserve Study
An August 2010 study by economists at the Federal Reserve Bank of New York has generated considerable interest as a road-map for reducing mortgage defaultsiv. Specifically the study has been used to argue for increased principal reduction as a way to reduce defaults. While the study has a number of flaws, for instancing assuming that all re-defaults only occur within 12 months of a modification, the study does take the appropriate approach in examining borrower incentives. The study correctly treats borrowers as choosing to default, rather than modeling default as something that simply “happens” to the borrower. The impact of principal reduction is also relative small, lower the author’s estimated 12 month re-default rate of 56 percent by 4.5 percent to 51.5 percent. So even if we adopted the author’s proposal, over half of modified loans would still re-default.
Not surprisingly proponents of principal reduction are choosing which parts of this study they like and discarding the parts they do not. For instance the study finds that “each additional month that a borrower can expect to live rent-free in the house increases the 12 month re-default rate by 0.6 percentage points.” To put that in perspective, the difference in the overall foreclosure process between judicial states and non-judicial foreclosure states in about 18 months. At 0.6 percentage points a month, if judicial states switched to an administrative process, re-default rates would decline by an estimated 10.8 percentage points or twice the impact one gets from a 10 percent reduction in principal. States with allow recourse have re-default rates that are 1.8 percentage points lower. Interestingly enough the authors find that the lower are area house prices, compared to their 2000 values, the lower are re-default rates. Attempts to keep prices above their pre-bubble rates have, to some extent, increased defaults. The logic is that a borrower’s decision to default is based not solely on current equity but also on the expected path of future home prices. If we can get to the bottom, which I believe we are nearing, then borrowers will have greater incentives to maintain their mortgage.
If you are going to modify…
While I remain quite skeptical of many of the efforts at mortgage modification, as most seem aimed at dragging out the problem and avoiding the inevitable correction of the housing market, if we are going to continue offering modifications to delinquent and/or underwater borrowers, we should include the following provisions:
- All modifications should include and exercise recourse.
- Modifications should be limited to those have been current at some point within the previous year.
- Modifications should be targeted to those who display a “willingness to pay” but lack the ability to do so.
Current modification programs have often been inspired by the creation of the Home Owners Loan Corporation (HOLC) in 1933, which re-financed borrowers into “affordable” long term loans. Apparently the nostalgia for the HOLC has encouraged an ignorance of its actual workings. The HOLC practiced aggressive recourse, for instance. So much so that a third of its total revenues were derived from deficiency judgments. The HOLC also limited assistance to credit-worthy borrowers who demonstrated a willingness to pay. If we wish to mimic the claimed success of the HOLC than we also need to understand how it functioned.v
There are some reports that the recent robo-signing settlement with give banks up to $1.7 billion in credit against the overall settlement if they waive their right to pursue deficiency judgmentsvi. The empirical literature is fairly robust on this point: the existence of deficiency judgments reduces foreclosures. This aspect of the settlement will likely increase foreclosures.
What’s a conservator for?
Criticism has been directed at FHFA for not either allowing or forcing Fannie Mae and Freddie Mac to engage in principal reductions. Much of this criticism has take the form of claims that the GSEs, and hence FHFA, are not “doing enough” to turn around the housing market. Blogger Matt Yglesias suggests that “clearly the purpose of creating the FHFA and taking Fannie and Freddie into conservatorship can’t have been to minimize direct taxpayer financial losses on agency debt.” This claim, and others like it, are mistaken. The Housing and Economic Recovery Act (HERA) of 2008 is quite clear when it comes to the duty and responsibilities of FHFA when acting as a conservator. A simple read of the statute, Section 1145 of HERA, which amends Section 1367 of the 1992 GSE Act, clearly states the purpose, duties, and role of a conservatorship. What does the law say the powers of a conservatorship are? They are to “take such action as may be — (i) necessary to put the regulated entity in a sound and solvent condition; and (ii) appropriate to carry on the business of the regulated entity and preserve and conserve the assets and property of the regulated entity.”
Some proponents of principal reduction have found language elsewhere in HERA which they believe allows for considerations beyond those found in Section 1145. But this argument relies on general introductory sections of the statute, not the powers and duties of FHFA as a conservator. Statutory interpretation requires that more specific sections trump general introductory sections. General sections have “no power to give what the text of the statute takes away” (Demore v. Kim, 538 U.S. 510, 535). Given FHFA’s estimate that a broad based program of principal reduction would cost almost $100 billion, the argument that an unelected, un-appointed, acting agency head should, in the absence of statutory authority, spend $100 billion on taxpayer money is simply inconsistent with our system of government. While agencies such as the Federal Deposit Insurance Corporation felt free to violate the law during the crisis, Acting FHFA Director DeMarco should be commended for his faithfulness to the letter of the law. If $100 billion of taxpayer dollars is to be spent on principal reduction, it is the responsibility of Congress to make that decision. To suggest this action be implemented without Congressional approval would only further erode the already diluted powers of Congress relative to the other branches of government. Members had the opportunity during the passage of HERA to increase the powers and duties of FHFA as conservator. Congress decided not to.
The problem is mortgage availability
The problem facing our housing market is a combination of weak demand and excess supply. All policy proposals should first be evaluated on that basis. One of the constraints on demand is mortgage availability. If one is a prime borrower, who can make a substantial down-payment, then mortgages are both cheap and plentiful. If one is not, then a mortgage is difficult, if not impossible to get.
This decline in mortgage availability derives from a variety of factors, some good, some bad. For instance the most irresponsible lending, with the exception of FHA, is gone (for how long, who knows). That is a good thing. Unfortunately much of the Alt‑A and higher quality subprime lending is also gone. That is not such a good thing. By my estimate about a fifth of the mortgage market has disappeared, holding back housing demand. One of the factors contributing to that disappearance is the combination of Federal Reserve interest rate policy with federal mortgage regulation. For instance under HOEPA, today any mortgage over 5.5 percent is considered “high-cost”. Such mortgages now carry considerable regulatory, reputation and litigation risk. Anyone with just a basic knowledge of financial history knows that 5.5 is, historically speaking, a great rate, not a predatory one. Charts, at the end of this testimony, display the distribution of mortgages rates charged in 2006 and 2011. It should be immediately clear that 2006 largely resembled a normal distribution. 2011, however, has seen the right side of that distribution largely eliminated. Clearly the distribution of mortgage rates in 2011 is near normal nor symmetric. I believe the Federal Reserve’s 2008 HOEPA regulation has contributed to this abnormality. Of course there are other factors, again some good, some bad.
Foreclosure Mitigation and the Labor Market
There is perhaps no more important economic indicator than unemployment. The adverse impacts of long-term unemployment are well known, and need not be repeated here. Although there is considerable, if not complete, agreement among economists as to the adverse consequences of jobless; there is far less agreement as to the causes of the currently high level of unemployment. To simplify, the differing explanations, and resulting policy prescriptions, regarding the current level of unemployment fall into two categories: 1) unemployment as a result of lack of aggregate demand, and 2) unemployment as the result of structural factors, such as skills mismatch or perverse incentives facing the unemployed. As will be discussed below, I believe the current foreclosures mitigation programs have contributed to the elevated unemployment rate by reducing labor mobility. The current foreclosures mitigation programs have also helped keep housing prices above market-clearing levels, delaying a full correction in the housing market.
First we must recognize something unusual is taking place in our labor market. If the cause of unemployment was solely driven by a lack of demand, then the unemployment rate would be considerably lower. Both GDP and consumption, as measured by personal expenditures, have returned to and now exceed their pre-crisis levels. But employment has not. Quite simply, the “collapse” in demand is behind us and has been so for quite some time. What has occurred is that the historical relationship between GDP and employment (which economists call “Okun’s Law) has broken down, questioning the ability of further increases in spending to reduce the unemployment rate. Also indicative of structural changes in the labor market is the breakdown in the “Beveridge curve” — that is the relationship between unemployment and job vacancies. Contrary to popular perception, job postings have been steadily increasing over the last year, but with little impact on the unemployment rate.
Historically many job openings have been filled by workers moving from areas of the country with little job creation to areas with greater job creation. American history has often seen large migrations during times of economic distress. And while these moves have been painful and difficult for the families involved, these same moves have been essential for helping the economy recover. One of the more interesting facets of the recent recession has been a decline in mobility, particular among homeowners, rather than an increase. Between 2008 and 2009, the most recent Census data available, 12.5 percent of households moved, with only 1.6 moving across state lines. Corresponding figures for homeowners is 5.2 percent and 0.8 percent moving across state lines. This is considerably below interstate mobility trends witnessed during the housing boom. For instance from 2004 to 2005, 1.5% of homeowners moved across state lines, almost double the current percentage. Interestingly enough the overall mobility of renters has barely changed from the peak of the housing bubble to today. This trend is a reversal from that witnessed after the previous housing boom of the late 1980s burst. From the peak of the bubble in 1989 to the bottom of the market in 1994, the percentage of homeowners moving across state lines actually increased.
The preceding is not meant to suggest that all of the declines in labor mobility, or increase in unemployment, is due to the foreclosure mitigation programs. Far from it. Given the many factors at work, including the unsustainable rate of homeownership, going into the crisis, it is difficult, if not impossible, to estimate the exact contribution of the varying factors. We should, however, reject policies that encourage homeowners to remain in stagnant or declining labor markets. This is particularly important given the fact that unemployment is the primary driver of mortgage delinquency.
Minimizing Losses to Taxpayers
As the title of today’s hearing implies an important objective of policy should be to protect the taxpayer from further loss. We should never forget that the taxpayer has already poured $180 billion in the rescue of Fannie Mae and Freddie Mac. It is unlikely that much, if any, of this will ever be recovered. In addition the taxpayer potentially faces the cost of rescuing the Federal Housing Administration (FHA). I believe there is a significant likelihood that the taxpayer will have to inject somewhere between $10 to $50 into FHA over the next 5 to 6 years.
The most effective way to protect the taxpayer would be to simply stop. Stop covering the losses of Fannie Mae and Freddie Mae and do not impose policies that would dig the current hole any deeper. We are well past the height of the financial panic. And as the recent mortgage settlement demonstrated, policy-makers appear to have no problem with imposing losses on investors. The same should be applied to Fannie Mae and Freddie Mae. Future losses should be borne by the debt-holders of those companies, not the taxpayer. Accordingly Fannie Mae and Freddie Mac should be moved immediately out of conservatorship and into receivership, where losses can be imposed upon those investors who willingly risked their own money (the same cannot be said for the taxpayer).
As FHFA estimates that a program of principal forgiveness for all underwater GSE mortgages could cost as much as $100 billion, it should be very clear that such would not minimize losses to the taxpayer.
Summary of Policy Proposals
- Repeal/Suspend/Modify Existing HOEPA Regulations.
- Require recourse for all federally related modifications.
- End programs, like “Neighborhood stabilization”, that add to housing supply. If spending, use such to increase demand, not supply.
- Reform FHA to minimize embedded losses.
Conclusion
The U.S. housing market is weak and is expected to remain so for some time. Given the importance of housing in our economy, the pressure for policymakers to act has been understandable. Policy should, however, be based upon fostering an unwinding of previous unbalances in our housing markets, not sustaining said unbalances. We cannot go back to 2006, and nor should we desire to. As the size and composition of the housing stock are ultimately determined by demographics, something which policymakers have little influence over in the short run, the housing stock must be allowed to align itself with those underlying fundamentals. Prices should also be allowed to move towards their long run relationship with household incomes. Getting families into homes they could not afford was a major contributor to the housing bubble. We should not seek to repeat that error. We must also recognize that prolonging the correction of the housing market makes the ultimate adjustment worse, not better. Lastly it should be remembered that one effect of boosting prices above their market-clearing levels is the transfer of wealth from potential buyers (renters) to existing owners. As existing owners are, on average, wealthier than renters, this redistribution is clearly regressive.
Notes
i Denk, Dietz and Crowe. Pent-up Housing Demand: The Household Formations That Didn’t Happen – Yet. National Association of Home Builders. February 2011.
ii Edward Glaeser and Gyourko, Joseph, “The Case Against Housing Price Supports,” Economists’ Voice October 2008.
iii Also see Robert Shiller, “Unlearned lessons from the housing bubble,” Economists’ Voice July 2009.
iv Andrew Haughwout, Ebiere Okah and Joseph Tracy, Second Chances: Subprime Mortgage Modifications and Re-Default. Federal Reserve Bank of New York Staff Reports no. 417. August 2010.
v See C. Lowell Harriss, History and Policies of the Home Owners’ Loan Corporation, National Bureau of Economic Research, 1951. http://www.nber.org/books/harr51‑1
vi Nick Timiraos, Mortgage Deal Built on Tradeoffs, Wall Street Journal, Monday, March 12, 2012, C1.
