About


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 Kristopher Gerardi, Carl Hudson, and analysts, as well as the Boston Fed's Christopher Foote and Paul Willen.

Comment Standards:
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.

June 9, 2016

Construction Lending Update: Have the Banks Finally Opened the Spigots?

When we last blogged about at bank call report data, in June 2014, we found that "aggregate lending remained well below its 2008 peak," but "more than half of banks with a construction lending business line were expanding" their lending. Fast forwarding two years, where does construction lending stand now? We pulled bank call report data through the first quarter of 2016 and found that construction lending has continued to grow, albeit at a measured pace (see table 1).

table-one

Of the insured banks with a construction lending business line, 62.2 percent have stepped up their lending relative to the year-earlier level. Not only are there more banks actively lending, but half of these banks increased their lending by at least 11.9 percent.

Despite this seemingly good news, it appears that most banks remain selective about the loans they make, and a few large banks are largely responsible for the increase in aggregate lending. In the first quarter of 2016, the top 20 construction lenders accounted for more than one-third of all construction lending (that is, 0.4 percent of active construction lenders are responsible for 37 percent of all construction lending). To provide some perspective, the top 20 banks accounted for 32 percent of all construction lending in 2005 and 42 percent in 2010. Slicing the data this way suggests that it is not particularly unusual for the top 20 to play such a large role in construction lending and that smaller lenders have made some progress toward recouping the market share of the top 20, though they aren't as active as they were in 2005.

Shifting attention now to the second and third set of columns in table 1, we'd like to point out that call report data in 2010 started breaking down total construction lending data into "Residential 1–4 family construction loans" and "Other loans, all land development and other land" categories. Note that this "Other" category includes construction loans for nonresidential and multifamily properties. While lending in both categories has increased over the past two years, growth has been much stronger for "Residential 1–4 family construction" relative to "Other construction, all land development and other land." Our interpretation of this divergence remains quite similar to our assessment two years earlier: the slower growth in "Other" is likely the outcome of fairly strong growth in multifamily construction lending weighed down by banks' continued reluctance to lend on land and lot development.

While the data seem to indicate that the construction lending spigots have opened up a little over the past two years, it is less clear who is able to access this credit. Bank call report data is aggregated in a way that prevents us from knowing anything about the borrowers. Anecdotally, using our monthly poll of Southeast homebuilders, we have not picked up much in the way of improved access to construction credit (see table 2). The majority of builders in our monthly poll continued to report that the amount of available credit for construction and development falls short of demand.

table-two

About a year ago, we asked our builder respondents to self-identify as small, medium, or large. By tagging respondents with a size, we've been able to break out the results to see how small-builder responses compared to all responses. Not surprisingly, small builders find credit to be less available than the group as a whole. Moreover, there has only been a slight change in the responses over the past year (three out of four small builders still find credit to be insufficient compared to four out of five one year ago). While a few smaller builders may have had better luck in securing construction and development lending over the past year, we haven't been able to detect much in the way of broad improvement in access to credit for construction and development.

We also looked to the April 2016 Senior Loan Officer Opinion Survey (SLOOS), published by the Federal Reserve Board, for insights into construction lending. The results seem to paint a construction lending picture that is similar to but not completely aligned with the one we outlined above. In short, the SLOOS reports that a "significant net fraction of banks reported tightening standards for construction and land development loans" while a "moderate net fraction of banks reported stronger demand for construction and land development loans." It is not clear that the call report data and the SLOOS are telling the same story on construction lending behavior, but perhaps this difference is simply an early signal of what we can expect from the second quarter call report.

Photo of Jessica Dill By Jessica Dill, economic policy analyst in the Research Department and

Photo of Carl Hudson Carl Hudson, director of the Center for Real Estate Analytics

July 1, 2015

Are Millennials Responsible for the Decline in First-Time Home Purchases? Part 2

Recall that, in our last post, we investigated the claim that millennials were to blame for the decline in first-time home purchases. Our data analysis confirmed that home purchases by first-time buyers have indeed plummeted since the crisis. We did not, however, find evidence that millennials were driving this decline. We found that, if anything, first-time homebuyers have become younger since the crisis, not older. By contrast, location appeared to be a much stronger predictor of declines in first-time buying than age.

Notwithstanding, many commentators still believe that millennials are behind sluggish sales. In this post, we take a closer look at the timing of first-time home purchases and the credit trends of first-time homebuyers with an eye towards the changing composition of homebuyers. We use the same credit bureau data set that we used in the previous post (take a look for a description of the data and our definition of first-time homebuyer). Using this data, we dig a bit deeper into two theories that are often cited for why millennial homebuyers are not buying as many homes as in the past. We first analyze whether millennials delayed the purchase of their first home in response to the crisis. Then we investigate what role, if any, credit tightening has played.

In short, we can't confirm any delay in the timing of home purchases. What we do find is that the distribution of first-time home purchases changed after the crisis. First-time home purchases by younger buyers peak earlier and persist at an elevated level over a longer period of time than before. We also find, contrary to the popular theory that credit became too tight for millennials to buy homes, that mortgage credit actually became tighter for older first-time buyers than for younger first-time buyers. Taken together, we think these data observations help to explain why the median age of the first-time buyer shifted downwards (instead of upwards) after the housing downturn.

Timing

To examine how the housing downturn affected the timing of purchases by young first-time homebuyers, we separated this group out by birth year and examined the number of home purchases from 2000 to 2014. We looked at millennial homebuyers born in 1983 and 1985 and compared them to Gen X homebuyers born in 1975 and 1977.

Chart 1 shows the number of first-time home purchases for each year, with each line representing a different birth year. The time series for the older birth years peaks between the ages of 27 and 29 while the time series for the younger birth years peaks between the ages of 24 and 25. For Gen Xers who came of age before the crash, their peak appears to be the culmination of a steep increase in purchases and an almost equally steep decline resulting in a curve that looks roughly like an inverted V. For the millennial birth years, who came of age after the real estate crash, the peak in first-time purchases occurs earlier and the decline of the curve is much more gradual. The change in the distribution of purchases after the crash suggests that the younger first-time home buyers are still purchasing homes at relatively high rates, but purchases are spread out over a wider time period.

Chart-01-first-time-home-purchases-by-birth-year

The distribution of millennial first-time homebuyers has clearly shifted. Not only has the distribution of first-time homebuyers become younger over time (refer to previous post) but first-time home buying among the most recent birth years is peaking at an earlier age. Why might this be? We think a closer look at credit trends can shed some light on this question.

Credit scores

In Chart 2, we examine the number of first-time home purchases and median credit scores of first-time home purchasers by age bracket. The two age brackets are adults under 35 years old and adults between the ages of 35 and 48. By grouping first-time home purchases into age brackets, we are able to examine whether credit is tighter for younger borrowers than for older borrowers using the median Equifax risk score as a proxy for credit tightness.1

Chart-02-credit-scores-of-first-time-home-buyers-by-age-bracket

From 2001 to 2014, median FICO scores increased by 5.0 percent for the younger group and 5.1 percent for the older group. In general, the median credit scores of both groups appear to behave similarly, except during the years when subprime lending prevailed. The median credit scores for both younger and older buyers shifted down between 2003 and 2006, signaling that there were more purchases by higher-risk buyers. With that said, the decline in the median credit score was more pronounced for older buyers (down 4.1 percent, from 689 in 2003 to 661 in 2006) than for younger buyers (down just 1.6 percent, from 693 in 2003 to 682 in 2006). Since the crisis, the gap between the median credit scores of younger and older buyers has closed (in other words, credit has become tighter for older buyers), which may explain why first-time home purchases have fallen faster for older buyers than it has for younger buyers. Indeed, between 2001 and 2014, first-time home purchases fell by 36 percent for younger homebuyers and by 54 percent for older buyers.

The table and Charts 3 and 4, below, delve deeper into the credit trends of younger and older first-time homebuyers, showing home purchases by credit bracket, year, and age. We determined credit brackets by taking the quartiles of every individual with a credit record in the time period. As the charts show, purchases by those with the lowest credit scores, marked in blue, have plummeted steeply. Credit scores in the middle, marked in green and orange, fell sharply, too, but particularly among older buyers. Older first-time homebuyers with moderate credit scores were much less likely to buy homes after than before the crisis, falling by 50 to 60 percent. Purchases by those with stellar credit, marked in purple, were barely affected by the crisis. Perhaps a more interesting observation is that young homebuyers in the highest credit bracket were the one subgroup to increase their purchases during and after the recession. First-time purchases by this group of young buyers actually rose by 25 percent.

First-time-home-purchases-by-age-credit-bracket

Charts-3-and-4

We believe this collection of charts demonstrates in more detail that credit became tighter for older homebuyers during the crisis—and also that an uptick in home purchases by the most creditworthy millennials has buoyed purchases for that age bracket.

The Federal Reserve Bank of New York Consumer Credit Panel/Equifax data is an unusual data set in that it allows us to compare first-time homebuyers without first conditioning by age. By comparing older and younger first-time homebuyers, we have been able to examine the claim that millennial homebuyers are behind the stagnation in home sales. In addition to our earlier findings that first-time buyers have become younger, not older, since the crisis, we find that the distribution of first-time home purchases has changed since the housing downturn. Specifically, first-time purchases by younger buyers tend to peak at an earlier age and persist at an elevated level over a longer period of time. This is in contrast to the trend before the downturn, when first-time purchases by younger buyers peaked at an older age and dropped off precipitously after peaking. Moreover, the data reveal that, while younger and older first-time buyers have similar credit trends when tracked as the median credit score, credit may have loosened more for older first-time buyers than younger first-time homebuyers during the run-up and as a consequence tightened more for older buyers than younger buyers during the recovery, resulting in a lower number of first-time purchases by this older group. Despite the fact that many believe tight mortgage credit, student loan debt burdens, and stagnating wages have made it more difficult for millennials to buy homes, it appears that credit tightness has actually weighed more heavily on older first-time homebuyers.

Photo of Carl Hudson By Elora Raymond, graduate research assistant, Center for Real Estate Analytics in the Atlanta Fed's research department, and doctoral student, School of City and Regional Planning at the Georgia Institute of Technology, and

Photo of Jessica DillJessica Dill, economic policy analysis specialist in the Atlanta Fed's research department

_______________________________________

1 Examining credit scores over time can be misleading. Credit scores measure a person's ranking of creditworthiness at a given time. A credit score does not give an absolute measure of someone's default probability, just where they are relative to others. So, someone with a 700 credit score in one time period may have a different default probability than someone in another time period, though their rank relative to others remains the same. This becomes relevant if the creditworthiness of the American public as a whole shifts dramatically over time

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: Measures of the Relationship between within-County Income Growth and Credit Growth for U.S. ZIP Codes

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."

Chart 2: Serious Delinquency Rates by Loan Vintage

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.

1 Mian and Sufi's contribution to the data-quality debate can be found here.

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.


March 25, 2015

Where Is the Credit Availability Pendulum Now? (Part 2 of 2)

In our previous post, we considered survey-based and index measures of mortgage credit availability. We concluded that availability has slowly but steadily been improving since early 2013. In this post, we focus on borrower characteristics for originated mortgages. We think of availability of credit as the willingness of lenders to lend while borrower characteristics shape the quantity of purchasers that are qualified to buy. By turning to mortgage origination data, we can look at the “credit box” and track changes (that is, expansions and contractions) in the credit box over time. We do this acknowledging that this approach fails to capture variation in loans that have been declined and allows us only to observe variation in loans that have been originated.

Looking at trends in credit characteristics of purchase mortgages originations, we find data that support the idea that the credit box, which tightened during the Great Recession, has not gotten looser. For one, the distribution of FICO scores on conventional mortgages shifted during the housing downturn to a distribution dominated by borrowers with higher credit scores—those above 680—and has yet to show much movement in the other direction.

Distribution-of=credit-scores

Yet credit scores represent just one dimension of a multidimensional credit box. To paint a fuller picture, consider loan-to-value (LTV) ratios before and after the housing downturn. The table shows summary statistics of this data. Comparing the distributions of LTV ratios of mortgages originated in 2006 and 2014, it seems somewhat counterintuitive that the share of conventional mortgages with high LTVs was greater in 2014 (35.3 percent) than in 2006 (21.2 percent).

150325-table

Layering the distribution of FICO scores on the distribution of LTVs helps to explain away some of this peculiarity. A sizable share of the 2006 loans that were originated with an LTV greater than 80 percent fell on the lower end of the credit score spectrum. In contrast, most of the loans originated in 2014 with LTVs greater than 80 percent fell on the higher end of the credit score distribution.

One thing that is not in our data set is the extent to which these mortgages had piggyback mortgages. Data provided by Inside Mortgage Finance indicates that second-mortgage originations decreased from $430 billion in 2006 to $59 billion in 2013 (the most recent year for which data are available). That is, seconds shrank from 14 percent of total originations to just 3 percent of total originations. So it is possible that the share of conventional mortgages with an LTV greater than 80 percent is understated—especially in 2006.

So what are the takeaways? Clearly, there has been a shift in conventional mortgage originations towards borrowers with better credit records. Also, we have to be careful in interpreting the trend in LTV ratios when information on second and third liens is not available. Finally, while survey-based and index measures of availability of credit may be improving, evidence from borrower characteristics of originated mortgages tells a less compelling story and suggests the pendulum still has some distance to go before we can consider it in the loosening range.

Photo of Carl Hudson By Carl Hudson, director for the Center for Real Estate Analytics in the Atlanta Fed´s research department, and

 

Photo of Jessica DillJessica Dill, senior economic research analyst in the Atlanta Fed's research department

June 9, 2016

Construction Lending Update: Have the Banks Finally Opened the Spigots?

When we last blogged about at bank call report data, in June 2014, we found that "aggregate lending remained well below its 2008 peak," but "more than half of banks with a construction lending business line were expanding" their lending. Fast forwarding two years, where does construction lending stand now? We pulled bank call report data through the first quarter of 2016 and found that construction lending has continued to grow, albeit at a measured pace (see table 1).

table-one

Of the insured banks with a construction lending business line, 62.2 percent have stepped up their lending relative to the year-earlier level. Not only are there more banks actively lending, but half of these banks increased their lending by at least 11.9 percent.

Despite this seemingly good news, it appears that most banks remain selective about the loans they make, and a few large banks are largely responsible for the increase in aggregate lending. In the first quarter of 2016, the top 20 construction lenders accounted for more than one-third of all construction lending (that is, 0.4 percent of active construction lenders are responsible for 37 percent of all construction lending). To provide some perspective, the top 20 banks accounted for 32 percent of all construction lending in 2005 and 42 percent in 2010. Slicing the data this way suggests that it is not particularly unusual for the top 20 to play such a large role in construction lending and that smaller lenders have made some progress toward recouping the market share of the top 20, though they aren't as active as they were in 2005.

Shifting attention now to the second and third set of columns in table 1, we'd like to point out that call report data in 2010 started breaking down total construction lending data into "Residential 1–4 family construction loans" and "Other loans, all land development and other land" categories. Note that this "Other" category includes construction loans for nonresidential and multifamily properties. While lending in both categories has increased over the past two years, growth has been much stronger for "Residential 1–4 family construction" relative to "Other construction, all land development and other land." Our interpretation of this divergence remains quite similar to our assessment two years earlier: the slower growth in "Other" is likely the outcome of fairly strong growth in multifamily construction lending weighed down by banks' continued reluctance to lend on land and lot development.

While the data seem to indicate that the construction lending spigots have opened up a little over the past two years, it is less clear who is able to access this credit. Bank call report data is aggregated in a way that prevents us from knowing anything about the borrowers. Anecdotally, using our monthly poll of Southeast homebuilders, we have not picked up much in the way of improved access to construction credit (see table 2). The majority of builders in our monthly poll continued to report that the amount of available credit for construction and development falls short of demand.

table-two

About a year ago, we asked our builder respondents to self-identify as small, medium, or large. By tagging respondents with a size, we've been able to break out the results to see how small-builder responses compared to all responses. Not surprisingly, small builders find credit to be less available than the group as a whole. Moreover, there has only been a slight change in the responses over the past year (three out of four small builders still find credit to be insufficient compared to four out of five one year ago). While a few smaller builders may have had better luck in securing construction and development lending over the past year, we haven't been able to detect much in the way of broad improvement in access to credit for construction and development.

We also looked to the April 2016 Senior Loan Officer Opinion Survey (SLOOS), published by the Federal Reserve Board, for insights into construction lending. The results seem to paint a construction lending picture that is similar to but not completely aligned with the one we outlined above. In short, the SLOOS reports that a "significant net fraction of banks reported tightening standards for construction and land development loans" while a "moderate net fraction of banks reported stronger demand for construction and land development loans." It is not clear that the call report data and the SLOOS are telling the same story on construction lending behavior, but perhaps this difference is simply an early signal of what we can expect from the second quarter call report.

Photo of Jessica Dill By Jessica Dill, economic policy analyst in the Research Department and

Photo of Carl Hudson Carl Hudson, director of the Center for Real Estate Analytics

July 1, 2015

Are Millennials Responsible for the Decline in First-Time Home Purchases? Part 2

Recall that, in our last post, we investigated the claim that millennials were to blame for the decline in first-time home purchases. Our data analysis confirmed that home purchases by first-time buyers have indeed plummeted since the crisis. We did not, however, find evidence that millennials were driving this decline. We found that, if anything, first-time homebuyers have become younger since the crisis, not older. By contrast, location appeared to be a much stronger predictor of declines in first-time buying than age.

Notwithstanding, many commentators still believe that millennials are behind sluggish sales. In this post, we take a closer look at the timing of first-time home purchases and the credit trends of first-time homebuyers with an eye towards the changing composition of homebuyers. We use the same credit bureau data set that we used in the previous post (take a look for a description of the data and our definition of first-time homebuyer). Using this data, we dig a bit deeper into two theories that are often cited for why millennial homebuyers are not buying as many homes as in the past. We first analyze whether millennials delayed the purchase of their first home in response to the crisis. Then we investigate what role, if any, credit tightening has played.

In short, we can't confirm any delay in the timing of home purchases. What we do find is that the distribution of first-time home purchases changed after the crisis. First-time home purchases by younger buyers peak earlier and persist at an elevated level over a longer period of time than before. We also find, contrary to the popular theory that credit became too tight for millennials to buy homes, that mortgage credit actually became tighter for older first-time buyers than for younger first-time buyers. Taken together, we think these data observations help to explain why the median age of the first-time buyer shifted downwards (instead of upwards) after the housing downturn.

Timing

To examine how the housing downturn affected the timing of purchases by young first-time homebuyers, we separated this group out by birth year and examined the number of home purchases from 2000 to 2014. We looked at millennial homebuyers born in 1983 and 1985 and compared them to Gen X homebuyers born in 1975 and 1977.

Chart 1 shows the number of first-time home purchases for each year, with each line representing a different birth year. The time series for the older birth years peaks between the ages of 27 and 29 while the time series for the younger birth years peaks between the ages of 24 and 25. For Gen Xers who came of age before the crash, their peak appears to be the culmination of a steep increase in purchases and an almost equally steep decline resulting in a curve that looks roughly like an inverted V. For the millennial birth years, who came of age after the real estate crash, the peak in first-time purchases occurs earlier and the decline of the curve is much more gradual. The change in the distribution of purchases after the crash suggests that the younger first-time home buyers are still purchasing homes at relatively high rates, but purchases are spread out over a wider time period.

Chart-01-first-time-home-purchases-by-birth-year

The distribution of millennial first-time homebuyers has clearly shifted. Not only has the distribution of first-time homebuyers become younger over time (refer to previous post) but first-time home buying among the most recent birth years is peaking at an earlier age. Why might this be? We think a closer look at credit trends can shed some light on this question.

Credit scores

In Chart 2, we examine the number of first-time home purchases and median credit scores of first-time home purchasers by age bracket. The two age brackets are adults under 35 years old and adults between the ages of 35 and 48. By grouping first-time home purchases into age brackets, we are able to examine whether credit is tighter for younger borrowers than for older borrowers using the median Equifax risk score as a proxy for credit tightness.1

Chart-02-credit-scores-of-first-time-home-buyers-by-age-bracket

From 2001 to 2014, median FICO scores increased by 5.0 percent for the younger group and 5.1 percent for the older group. In general, the median credit scores of both groups appear to behave similarly, except during the years when subprime lending prevailed. The median credit scores for both younger and older buyers shifted down between 2003 and 2006, signaling that there were more purchases by higher-risk buyers. With that said, the decline in the median credit score was more pronounced for older buyers (down 4.1 percent, from 689 in 2003 to 661 in 2006) than for younger buyers (down just 1.6 percent, from 693 in 2003 to 682 in 2006). Since the crisis, the gap between the median credit scores of younger and older buyers has closed (in other words, credit has become tighter for older buyers), which may explain why first-time home purchases have fallen faster for older buyers than it has for younger buyers. Indeed, between 2001 and 2014, first-time home purchases fell by 36 percent for younger homebuyers and by 54 percent for older buyers.

The table and Charts 3 and 4, below, delve deeper into the credit trends of younger and older first-time homebuyers, showing home purchases by credit bracket, year, and age. We determined credit brackets by taking the quartiles of every individual with a credit record in the time period. As the charts show, purchases by those with the lowest credit scores, marked in blue, have plummeted steeply. Credit scores in the middle, marked in green and orange, fell sharply, too, but particularly among older buyers. Older first-time homebuyers with moderate credit scores were much less likely to buy homes after than before the crisis, falling by 50 to 60 percent. Purchases by those with stellar credit, marked in purple, were barely affected by the crisis. Perhaps a more interesting observation is that young homebuyers in the highest credit bracket were the one subgroup to increase their purchases during and after the recession. First-time purchases by this group of young buyers actually rose by 25 percent.

First-time-home-purchases-by-age-credit-bracket

Charts-3-and-4

We believe this collection of charts demonstrates in more detail that credit became tighter for older homebuyers during the crisis—and also that an uptick in home purchases by the most creditworthy millennials has buoyed purchases for that age bracket.

The Federal Reserve Bank of New York Consumer Credit Panel/Equifax data is an unusual data set in that it allows us to compare first-time homebuyers without first conditioning by age. By comparing older and younger first-time homebuyers, we have been able to examine the claim that millennial homebuyers are behind the stagnation in home sales. In addition to our earlier findings that first-time buyers have become younger, not older, since the crisis, we find that the distribution of first-time home purchases has changed since the housing downturn. Specifically, first-time purchases by younger buyers tend to peak at an earlier age and persist at an elevated level over a longer period of time. This is in contrast to the trend before the downturn, when first-time purchases by younger buyers peaked at an older age and dropped off precipitously after peaking. Moreover, the data reveal that, while younger and older first-time buyers have similar credit trends when tracked as the median credit score, credit may have loosened more for older first-time buyers than younger first-time homebuyers during the run-up and as a consequence tightened more for older buyers than younger buyers during the recovery, resulting in a lower number of first-time purchases by this older group. Despite the fact that many believe tight mortgage credit, student loan debt burdens, and stagnating wages have made it more difficult for millennials to buy homes, it appears that credit tightness has actually weighed more heavily on older first-time homebuyers.

Photo of Carl Hudson By Elora Raymond, graduate research assistant, Center for Real Estate Analytics in the Atlanta Fed's research department, and doctoral student, School of City and Regional Planning at the Georgia Institute of Technology, and

Photo of Jessica DillJessica Dill, economic policy analysis specialist in the Atlanta Fed's research department

_______________________________________

1 Examining credit scores over time can be misleading. Credit scores measure a person's ranking of creditworthiness at a given time. A credit score does not give an absolute measure of someone's default probability, just where they are relative to others. So, someone with a 700 credit score in one time period may have a different default probability than someone in another time period, though their rank relative to others remains the same. This becomes relevant if the creditworthiness of the American public as a whole shifts dramatically over time

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: Measures of the Relationship between within-County Income Growth and Credit Growth for U.S. ZIP Codes

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."

Chart 2: Serious Delinquency Rates by Loan Vintage

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.

1 Mian and Sufi's contribution to the data-quality debate can be found here.

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.


March 25, 2015

Where Is the Credit Availability Pendulum Now? (Part 2 of 2)

In our previous post, we considered survey-based and index measures of mortgage credit availability. We concluded that availability has slowly but steadily been improving since early 2013. In this post, we focus on borrower characteristics for originated mortgages. We think of availability of credit as the willingness of lenders to lend while borrower characteristics shape the quantity of purchasers that are qualified to buy. By turning to mortgage origination data, we can look at the “credit box” and track changes (that is, expansions and contractions) in the credit box over time. We do this acknowledging that this approach fails to capture variation in loans that have been declined and allows us only to observe variation in loans that have been originated.

Looking at trends in credit characteristics of purchase mortgages originations, we find data that support the idea that the credit box, which tightened during the Great Recession, has not gotten looser. For one, the distribution of FICO scores on conventional mortgages shifted during the housing downturn to a distribution dominated by borrowers with higher credit scores—those above 680—and has yet to show much movement in the other direction.

Distribution-of=credit-scores

Yet credit scores represent just one dimension of a multidimensional credit box. To paint a fuller picture, consider loan-to-value (LTV) ratios before and after the housing downturn. The table shows summary statistics of this data. Comparing the distributions of LTV ratios of mortgages originated in 2006 and 2014, it seems somewhat counterintuitive that the share of conventional mortgages with high LTVs was greater in 2014 (35.3 percent) than in 2006 (21.2 percent).

150325-table

Layering the distribution of FICO scores on the distribution of LTVs helps to explain away some of this peculiarity. A sizable share of the 2006 loans that were originated with an LTV greater than 80 percent fell on the lower end of the credit score spectrum. In contrast, most of the loans originated in 2014 with LTVs greater than 80 percent fell on the higher end of the credit score distribution.

One thing that is not in our data set is the extent to which these mortgages had piggyback mortgages. Data provided by Inside Mortgage Finance indicates that second-mortgage originations decreased from $430 billion in 2006 to $59 billion in 2013 (the most recent year for which data are available). That is, seconds shrank from 14 percent of total originations to just 3 percent of total originations. So it is possible that the share of conventional mortgages with an LTV greater than 80 percent is understated—especially in 2006.

So what are the takeaways? Clearly, there has been a shift in conventional mortgage originations towards borrowers with better credit records. Also, we have to be careful in interpreting the trend in LTV ratios when information on second and third liens is not available. Finally, while survey-based and index measures of availability of credit may be improving, evidence from borrower characteristics of originated mortgages tells a less compelling story and suggests the pendulum still has some distance to go before we can consider it in the loosening range.

Photo of Carl Hudson By Carl Hudson, director for the Center for Real Estate Analytics in the Atlanta Fed´s research department, and

 

Photo of Jessica DillJessica Dill, senior economic research analyst in the Atlanta Fed's research department