ByYang Liu , Rajdeep Sengupta
The origins of the recent financial crisis have often been traced to the excesses in the U.S. mortgage market. Most accounts of the crisis tend to focus on a significant decline in underwriting standards for mortgages since 2000. After the crisis, the pendulum appears to have swung in the other direction. Anecdotal evidence suggests that borrowers are finding it difficult to obtain housing loans. Some observers have remarked that this difficulty may be one of the causes of the slump in the U.S. market for housing.
Using a data set of loan applications and originations, we analyzed these trends for the Federal Reserve's Eighth District, based in St. Louis. Our data came from the Home Mortgage Disclosure Act (HMDA) files for 2004, 2009 and 2010. The HMDA data for 2004 were used as an indicator of the precrisis mortgage market conditions, whereas HMDA data for 2009-2010 were used to indicate postcrisis mortgage conditions. We restricted our observations to first-lien, one-to four-family home mortgage loans.
As expected, the data show that the financial crisis adversely affected the demand for mortgage loans in the District. Figure 1 displays a panel of scatter plots showing pre- and postcrisis mortgage applications in each county of the District. The horizontal axis of each plot measures the level of 2004 mortgage loan applications, while the vertical axis measures the annual average of 2009-2010 mortgage loan applications. Each dot in the chart represents one of the 339 counties. The plot also shows the 45-degree line where the level of 2004 applications equals the annual average of 2009-2010 applications. Simply put, a dot below the 45 degree line indicates that postcrisis applications for that county were fewer than precrisis applications; a dot above indicates the opposite.
For the District, there were 290,091 fewer mortgage applications annually during 2009-2010 than in 2004 (a reduction of 33.3 percent). Figure 1A shows that 327 out of the 339 counties in the District were located below the 45-degree line—a widespread drop in mortgage applications across the District. The drop was greater for new purchases (Figure 1B) when compared with refinances (Figure 1C). Annual applications for purchases fell by 47.3 percent (139,707 applications) after the crisis; 319 counties experienced a decline in purchase applications. In contrast, applications for refinances fell by 25.3 percent (138,634 applications); 307 counties experienced a decline in refinance applications. Clearly, the drop in numbers was roughly the same for both purchases and refinances, but purchases constituted a smaller proportion of applications near the peak of the boom in 2004.
Interestingly, HMDA data also allowed us to sort the applications by the agency that supervises each lending institution to which the application is made. Since different agencies supervise different types of lending institutions, we could use this variable to examine the differences in pre- and postcrisis applications by lending institutions. We sorted loan data by three different types of financial institutions: banks and thrifts, credit unions and "HUD-supervised mortgagees." This last category denotes loans made by institutions that are not supervised by any of the major agencies.
Banks and thrifts in the District experienced a moderate decrease in annual mortgage applications of 14 percent (or 71,738 applications) after the crisis (Figure 1D). Consumers filed fewer mortgage application loans to banks and thrifts in 252 counties. HUD-supervised mortgagees suffered the largest loss in mortgage loan applications on an annual basis (Figure 1F). They received 229,219 fewer loan applications, or a decline of 65.5 percent. In all but one of the District's counties, consumers filed fewer mortgage loan applications to HUD-supervised mortgagees. It is important to point out that the reduction of 229,219 applications in this sector accounted for 79 percent of the annual loan application decline in the District.
In contrast, credit unions enjoyed a surprising boom in home mortgage applications (Figure 1E). On an annual basis, mortgage applications rose by 10,813—an increase of 122 percent. Of the 275 counties in the District that recorded loan applications filed with credit unions, 222 counties recorded an increase in applications. Furthermore, annual applications increased by more than 100 percent in 123 District counties.
Although further research is needed to account for this rapid and anomalous increase, some anecdotal evidence may explain this rapid growth in the popularity of credit unions. First, there has been record growth in the membership of credit unions—much of this has been attributed to consumer disillusionment with big banks. Moreover, a large share of the growth in mortgage business is concentrated among the largest credit unions—which typically have lower limits on membership. Second, at least two of these large credit unions have reportedly been offering members mortgages without requiring any down payment or mortgage insurance.
To find out how loan-approval patterns in 2009-2010 differed from those in 2004, we examined the mortgage loan origination rate during the two periods. Figure 2 displays a panel of scatter plots showing pre- and postcrisis mortgage origination vis-à-vis applications for each county of the District. The horizontal axis of each plot shows the number of applications in the county, while the vertical axis measures the number of originations. The dots in red show the 2004 levels for each county, while the blue dots show the annual average for 2009-2010 in the same counties.
We plotted the corresponding "line of fit" for each period. A higher line indicates a higher origination rate for a given level of applications. At first glance, therefore, it is surprising that the postcrisis line of fit in almost all plots of Figure 2 appears higher than the precrisis trend lines. A possible explanation of this feature of the data is that although there are fewer applications postcrisis, their quality is significantly better. This may be partly due to the fact that real-estate salesmen are only willing to do business with preapproved buyers.
Figure 2 reveals two important patterns. First, the differences in origination rates for refinances (Figure 2C) appear to be greater than those for purchases (Figure 2B). Refinancing after a sharp decline in home prices can be tricky because existing homeowners would likely have to cover for the shortfall in home equity if they wanted to take advantage of lower mortgage rates. While this reduces the set of applicants, it can also ensure an improvement in the applicant pool, thereby resulting in higher origination rates.
Second, among all lending institutions, only credit unions' loan origination rates show a marginal decline (Figure 2E), primarily due to smaller origination growth relative to a larger increase in applications. In light of the anecdotal evidence given above, a possible explanation is that a significant increase in annual mortgage applications made credit unions more selective.
Prevost, Lisa. "The Credit Union Alternative," The New York Times, Dec. 13, 2012. See www.nytimes.com/2012/12/16/realestate/mortgages-credit-unions-grow-in-popularity.html?_r=0.
Morrison, David. "Two Credit Unions Offering No Money Down Mortgages," Credit Union Times, March 14, 2013. See http://www.cutimes.com/2013/03/14/two-credit-unions-offering-no-money-down-mortgages.