In the old English crime novels, the butler was always the prime suspect. In the present crisis, the “butler” is deregulation, assisted by other sleazy characters such as “cheap money”, “Asian savings”, “excessive liberalism and laissez-faire” and, according to presidential candidate John McCain, “wild greed” manifested in rampant speculative activities that created the house price bubble.
How blame is allocated is an important task because it will determine the characteristics of the long term “solutions”. The list of suspects aligned above is comprised of “vague” and “intangible” characters (although many have tried to go behind “cheap money” to point the finger at Greenspan), and so cannot provide a solid foundation for the design of good rules and institutions to improve the system.
Recently (WSJ Oct 3 Opinion page) Russell Roberts of George Mason University took on the role of defense attorney for the collection of suspects listed above. By way of an “opening statement” and introduction to Roberts’s arguments, I propose that credit is a “derivative”, in this case flowing from the increase in the demand for houses (and other consumer durables). If so, an evaluation of the causes of the crisis, that is unlikely to be associated with “wild greed” or “excessive liberalism and laissez-faire”, should direct attention to its origin: the residential real estate market.
Furthermore, given that the residential market has a high “social” content, we cannot refrain from considering the role played by the government and if we are able to put government at the base of the problem, the cause of the crisis cannot be excessive liberalism, quite the contrary!
Roberts´s main points (The Facts) are:
· Beginning in 1992, Congress pushed Fannie and Freddie (F&F) to increase their purchases of mortgages going to low and moderate income borrowers.
· For 1996, HUD gave F&F an explicit target – 42% of their mortgage financing had to go to borrowers with income below the median in their area. The target increased to 50% in 2000 and 52% in 2005
· For 1996, HUD required that 12% of all mortgage purchases by F&F be “special affordable” loans, typically to borrowers with income less than 60% of their area´s median income. That number was increased to 20% in 2000 and 22% in 2005.
· Between 2000 and 2005, F&F met those goals every year, funding hundreds of billions of dollars worth of loans, many of them subprime and adjustable-rate loans, and made to borrowers who bought houses with less than 10% down.
· F&F also purchased hundreds of billions of subprime securities for their own portfolios to make money and to help satisfy HUD affordable housing goals, making them important contributors to the demand for subprime securities.
· The Community Reinvestment Act (CRA) did the same thing with traditional banks. CRA was “strengthened” in 1995, causing an increase of 80% in the number of bank loans going to low-and moderate-income families.
· F&F were part of the CRA story, too. In 1997, Bear Sterns (!) did the first securitization of CRA loans, a $384 million offering guaranteed by Freddie Mac. Over the next 10 months, Bear Sterns issued $1.9 billion of CRA mortgages backed by F&F. Between 2000 and 2002 Fannie Mae securitized $394 billion in CRA loans.
· While F&F and the CRA were pushing up the demand for low-priced property, the Taxpayer Relief Act of 1997 increased the demand for higher valued property by expanding the availability and size of the capital-gains exclusion to $500,000 from $125,000. It also made it easier to exclude capital gains from rental property, further pushing up the demand for housing.
· Between 1997 and 2005 the average price of a house in the US more than doubled. It wasn´t simply a speculative bubble. Much of the rise in housing prices was the result of public policies that increased the demand for housing.
The consequence of all this: “On top of putting the entire financial system at risk, the hidden cost has been hundreds of billions of dollars funneled into the housing market instead of more productive assets”.
Now, for the Evidence:
Exhibit 1 shows what happened to new home sales over time.
The change in the pattern of the series after 1992 is evident to the naked eye. Until then the pattern was clearly cyclical. From 1992 to the end of 2005 the upward trend is clear. Possible reasons: greater macroeconomic stability beginning in the mid 80´s and consolidated in the early 90´s; better income prospects; expanded financing alternatives, falling interest rates, among others. All these reasons are relevant and comprise the fundamentals of the housing market. But after reading the “facts of the case” above, I´m inclined to think that the “things” concealed in the “among others” category is key to understanding the trend.
Consistent with the facts, the trend begins at the same time that Congress/HUD established “house ownership targets”. Now, let us examine the evidence from house prices.
Exhibit 2A and 2B show two different indices of house prices. Exhibit 2A illustrates house prices calculated by OFHEO for the US while Exhibit 2B shows the index of house prices compiled by S&P Case-Shiller as a composite of 10 metropolitan areas.
The S&P C-S index has become the house price “standard” in the last couple of years, coinciding with the start of the house price fall in the index, something that only now is becoming evident in the OFHEO Index.
The methodological differences between the two indices helps to explain the major differences in house price behavior observed. The S&P C-S index is weighed by value – more expensive houses have greater weight in the index, while the OFHEO index has unit weight. While S&P C-S gives greater weight to Census regions that contain more expensive houses (notably California and the North East), the OFHEO Index distributes weights in accordance with the number of residential units. In this case the weight attributed to the South is higher given the larger share of the population (and houses) in the region.
Figure 3 shows that the OFHEO Index and the S&P C-S Index are mutually consistent when we take account of the methodological differences in their construction. The OFHEO Index for the Pacific region strongly resembles the composite-10 S&P C-S of figure 2B while the Index for the “South” (here excluding the South Atlantic) region (with greater weight due to more house units) is a close image of the National OFHEO Index in 2A.
These two regions are important in compiling evidence on the presence of the “hand of the government” in the crisis. As the Facts above showed, the home ownership incentives for the lower income population were the driving force in the rise in demand for and prices of houses. The “program” succeeded. Between 1994 and 2006, overall homeownership climbed from 64% to 70%. Among Hispanics it went up by 20.2%, for Asians the increase was 17.2% and for African Americans 14%. Among non Hispanic whites the increase was a “paltry” 8.2%.
Now, 44% of the Hispanic population and 49% of the Asian population reside in the Pacific region (notably California) while about 30% of the Hispanic population and about 45% of the African American population reside in the “South”, so we would expect that prices in these two regions would come under pressure. But as Exhibit 3 shows the behavior of prices in the two regions is distinctively different.
A few years ago Ed Glaeser argued that zoning laws were important in constraining house supply and that these laws were more restrictive in the West than in the “South”. This helps to explain a significant part of the difference in house price behavior. Apparently in the Pacific, “affordable” housing was not so affordable after all! So now, prices drop more steeply.
In the S&P C-S Index, California weighs a lot. Exhibit 4 compares the index of prices in San Diego and LA with Dallas, the only “Southern” representative in the S&P C-S index. House supply has made all the difference in the world!
Exhibit 5 compares prices in different pairs of metropolitan areas according to the S&P C-S Index.
The more recent price drop in Detroit, higher than in other regions, is also tied to local conditions associated with higher unemployment due to the auto industry. For the past two years Michigan´s unemployment rate has hovered 1.5 to 2 percentage points above Ohio´s, for example. The relative stronger rise in house prices in Boston is associated, like in California, to local supply constraints.
Exhibit 5 also lends support to the view that the problems associated with house price falls cannot be generalized. In places like Charlotte (NC) and Denver (CO) as well as Dallas (TX), prices are better characterized as stable rather than falling.
In other places, for example Phoenix (AZ) and Las Vegas (NV) – Exhibit 6 – house prices don´t just rise; they are better characterized as “jumping late in the day”, with a steep rise towards the end of 2003 and a steep fall beginning in 2006. It is unlikely that much of this price action can be associated with the Congress/HUD “program”, being indicative o speculative activity.
The search for a “political free lunch” – off budget home ownership subsidy – and the consequent house price rise together with the belief that house prices don´t fall spawned a fantastic increase in financial leverage so that, when house prices fell significantly in some important regions, bank losses and the need for deleveraging set in. Since the trigger for the process is located in the mortgage market, mortgage default (price declines) will have to stop before we can say the system has stabilized. Fortunately, the problem is not as generalized as commonly thought.