The Persistent Consequences of Tight Credit: Evidence from Minimum Credit Score Mortgage Lending Rules

The Persistent Consequences of Tight Credit: Evidence from Minimum Credit Score Mortgage Lending Rules∗ Steven Laufer†and Andrew Paciorek‡ Board of Go...
Author: Junior Cameron
9 downloads 3 Views 391KB Size
The Persistent Consequences of Tight Credit: Evidence from Minimum Credit Score Mortgage Lending Rules∗ Steven Laufer†and Andrew Paciorek‡ Board of Governors of the Federal Reserve System June 29, 2016

PRELIMINARY AND INCOMPLETE. PLEASE DO NOT CITE WITHOUT PERMISSION. Abstract Since the housing bust and financial crisis, mortgage lenders have introduced progressively higher minimum thresholds for acceptable credit scores. Using loan-level data, we document the the introduction of these thresholds, as well as their effects on the distribution of newly originated mortgages. Combining the timing and nonlinearity of these supply-side changes allows us to credibly identify their short- and medium-run effects on various individual outcomes. Using a large panel of consumer credit data, we show that the credit score thresholds have very large negative effects on borrowing in the short run, and that these effects attenuate over time but remain sizable up to four years later. Under a counterfactual in which the thresholds were not imposed, we calculate that there would have been about 5 percent more first mortgages originated from 2012 to 2015, with much larger relative effects among individuals close to the thresholds.



All errors are our own. The views we express herein are not necessarily those of the Board of Governors or its staff. † E-mail: [email protected] ‡ E-mail: [email protected]

1

Introduction

Since the housing bust and subsequent financial crisis, US mortgage lenders have significantly tightened their lending standards. These tight lending conditions have likely contributed to the large decline in the homeownership rate as well as the weak recovery in residential construction. In addition, tight mortgage credit could pose a problem for housing affordability, as the historically low interest rates over the past few years means that mortgage-financed owner occupied housing is less expensive than rental housing in most areas of the country. While the evidence that mortgage credit conditions have tightened is fairly strong, it is difficult to quantify the magnitude of the tightening or to disentangle the effects of tight mortgage supply from low mortgage demand. Factors that prevent households from qualifying for a mortgage—such as low credit scores, high debt balances, and a lack of liquid assets—are also likely to reduce demand for owner-occupied housing. For example, the decline in mortgage originations to less credit-worthy borrowers over the past few years likely reflects more stringent lender standards, but it also likely reflects relatively weak labor market conditions among such borrowers, as well as reluctance of more financially vulnerable households to assume housing market risk following a period of extreme volatility. In this paper, we address this identification challenge by focusing on a particular aspect of tight mortgage credit, lenders’ requirements that borrowers must satisfy a minimum credit score requirement in order to qualify for a loan. In some cases, these credit score thresholds may be imposed to allow the lenders to securitize the mortgages through government programs that specify minimum credit scores. In other cases, they may simply reflect a ruleof-thumb about which mortgages are too risky to underwrite. Importantly for our work, lenders’ use of these minimum credit scores has varied over time in response to concerns that are likely unrelated to changes in demand from potential borrowers.1 We take advantage of these minimum credit score thresholds in order to construct a measure of mortgage credit availability that is plausibly orthogonal to changes in borrower demand. Our credit availability measure captures the difference in the ability of borrowers just above the credit-score threshold to obtain mortgages compared to borrowers whose credit scores fall just below the threshold. Specifically, our measure assigns a low degree of credit availability to individuals if 1) their credit scores are likely to fall below the threshold value and 2) the credit score threshold appears to be an important restriction determining access to mortgage credit at that point in time. The nonlinear relationship between our measure 1

Figure 1A, which we discuss below, shows the density of mortgages by credit score across four years. There are readily apparent thresholds at 620 in 2010 (the blue line) and 640 in 2012 (the green line).

1

and borrowers’ credit scores allows us to separately identify its effect while still controlling for variation in mortgage demand that may also be correlated with borrowers’ credit scores. Furthermore, because the importance of these credit score thresholds changes over time, our measure exhibits time-series variation that should be independent of other changes across time that may be related to borrowed demand. Our results confirm that our credit availability measure strongly predicts whether households take out new first mortgages even after controlling for time effects and borrowers’ credit score. We estimate that between 2008 and 2011, falling below the relevant credit score thresholds reduced an individual’s probability of obtaining a mortgage by 2.5 percentage points, compared to an average probability of taking out a mortgage of under 1 percent. The effects on mortgage originations are still very large over longer horizons, up to 16 quarters out, although they shrink in magnitude relative to the baseline probabilities. Even so, these results suggest that credit availability (or the lack thereof) has persistent consequences for individual borrowing behavior. We also examine the implications of mortgage credit availability for other outcomes that we can observe in our credit data. We find that credit availability modestly increases the probability of a major derogatory event—like foreclosure or bankruptcy—for individuals who did not previously have a mortgage. In contrast, availability lowers the probability that an individual who already had a mortgage balance experiences a derogatory event, suggesting that the ability to refinance is an important financial cushion. In addition, we find effects of credit availability on moving and migration behavior. Finally, credit availability seems to lead to more borrowing through auto loans, indicating that auto purchase and home purchase (or refinancing) are complementary. Our paper is related to a growing literature that tries to isolate the effects of mortgage credit availability during the recent housing cycle. Most similar to our paper is Anenberg et al. (2016), who characterize mortgage credit availability as the largest mortgage that a borrower can obtain given his credit score, income and ability to make a down payment. The authors show that tighter credit conditions depress both house prices and new residential construction. Gropp et al. (2014) attempt to separate out the supply and demand-based explanations for the slow consumption growth and reduction in household borrowing following the financial crisis. The authors conclude that that the reluctance of lenders to provide credit was the most important factor behind the reduction in aggregate borrowing. Gete and Reher (2016) identify local variations in mortgage credit tightness based on the share of mortgage lending by the largest banks in different areas prior the crisis. They argue that these banks

2

tightened credit more in response to new financial regulations and use the variation in their lending share to show that tight credit helps explains higher residential rents. Our paper presents yet another way of identifying the effects of mortgage credit availability by focusing explicitly on minimum credit scores as a clearly identified restriction on mortgage credit. We find that this form of credit restriction is important for understanding the low borrowing rates among less credit-worthy households since the crisis. Our paper also falls within a larger literature that has tried to identify the effects of mortgage credit availability on individual households attainment of homeownership. Early work in this literature includes Barakova et al. (2003) and Rosenthal (2002) who constructed measures of mortgage credit access from responses to the Federal Reserve’s Survey of Consumer Finances (SCF). More recently, Barakova et al. (2014) constructed a measure of mortgage credit access from the National Longitudinal Survey of Youth and Acolin et al. (2016) use more recent waves of the SCF. Among the few papers that have explicitly considered the effect of credit score, Chomsisengphet and Elul (2006) use credit scores merged with mortgage data to shed light on the effect of personal bankruptcy exemptions on secured lending. Our paper makes two advances in measuring households’ access to mortgage credit. First, we link credit scores available for a wide cross-section of individuals to the actual FICO credit scores used in mortgage underwriting decisions. Second, we take advantage of lenders’ use of particular minimum credit scores in particular time periods to translate these scores into a concrete measure of mortgage credit access. One advantage of our approach is that although our measure of credit access is based on individuals’ credit scores, we can assess its impact while controlling for credit score. This is important because scores are likely correlated with borrowers’ decisions through channels other than just affecting their ability to take out a mortgage. However, like other studies based on consumer credit data, we are unable to see household income or assets and therefore unable to account for the impact of those factors on households’ ability to borrow. In using credit score thresholds, our study is also related to work by Keys et al. (2009, 2010, 2012) who argue that the greater ability to securitize mortgages originated to borrowers with credit scores above 620 caused a moral hazard problem that led to lax screening. Bubb and Kaufman (2014) instead argue that the use of 620 as a threshold arose as a lender response to a fixed cost of screening potential borrowers. Our analysis is less concerned with the origin of lenders’ decision to apply minimum credit scores and more concerned with the effect of these rules on households’ ability to obtain mortgage credit. However, during the more recent period we study, lenders’ reliance on minimum credit scores quite clearly does

3

not reflect their difficulty in securitizing these loans. As we describe below, most securitized loans issued since the financial crisis to borrowers were placed into pools guaranteed by the FHA, whose explicit credit score minimums were notably lower than the thresholds we study. The rest of the paper proceeds as follows. Section 2 describes lenders’ use of minimum credit scores, how we observe the effects of these rules in the data, and the construction of our credit availability measure. We present our empirical results in section 3 and estimate the cumulative effects of the credit restrictions in section 4. Finally, section 5 concludes the paper.2

2

Data Sources and the Credit Availability Measure

2.1

A Recent History of Minimum Credit Score Lending Rules

Since the financial crisis, lenders have progressively tightened their standards for minimum credit scores. In figure 1, we plot the density and cumulative distribution of credit scores for mortgages originated in 2005, 2008, 2010, and 2012.3 Looking at the graphs, we can see that at certain key scores, there are drops in the number of loans originated to borrowers with credit scores just below those thresholds. As a result, the distribution of loans by FICO score has changed markedly since the crisis. In 2010 (the blue line), there were very few loans made to borrowers with credit scores below 620. By 2012 (the green line), the most significant threshold score was 640. The discontinuities in the distribution of credit scores are largely explained by lenders’ changing policies on issuing mortgages guaranteed by the Federal Housing Administration (FHA), which have dominated the market for low-score mortgages since the financial crisis. As can be seen in figure 2A, prior to the crisis, neither the FHA nor lenders imposed any particular minimum credit scores (the blue line), but the FHA’s market share was very low because of competition from conventional sub-prime lenders. By 2008 (figure 2B), most of those lenders had disappeared from the market, leaving the FHA as a “guarantor of last resort” for loans to borrowers with low scores. Since 2010, the FHA’s official guidelines have allowed it to guarantee mortgages to borrowers with credit scores as low as 500 as long as the down payment was at least 10 percent. Borrowers with credit scores above 580 can qualify for mortgages with down payments as 2

Future drafts of the paper will include an additional section with robustness checks, some of which are shown in tables 10 and 11. 3 The data, which come from Black Knight, are described more fully in section 2.2.

4

low as 3.5 percent. However, mortgage lenders have imposed additional overlays on their origination and purchase of FHA mortgages. In early 2009, immediately after the financial crisis, many lenders reportedly began requiring minimum scores of 620 on all the FHA mortgages they originated. Relative to 2008, figure 2C shows a dramatic reduction in the fraction of FHA mortgages to borrowers with scores below 620 in 2010. In response to the new FHA rules issued in October 2010, Wells Fargo and Bank of America, as well as other large banks, stopped buying FHA loans to borrowers with credit scores below 640, though both continued to originate loans down to 620 through their retail channels. Lenders reported that these changes were in response to high costs associated with servicing delinquent loans. Lenders also reportedly feared other sanctions from the FHA if their overall default rate rose too high including the possible loss of eligibility as FHA lenders. Figure 2D shows that, by 2012, very few FHA mortgages—or mortgages of any other type— were made to borrowers with scores below 640, a situation that has remained essentially unchanged since then.

2.2

Measuring Credit Availability

These sorts of discontinuities in the distribution of mortgages at particular credit scores indicate that lenders are using these scores in their underwriting decisions and are exhibiting some reluctance to lend to borrowers with credit scores that fall below this value. Intuitively, if borrowers with credit scores just above the threshold have a similar demand for mortgages compared to borrowers just below the threshold, then the difference in the number of mortgages originated to these two groups must reflect pure differences in the supply of mortgage credit. We can use these differences to identify the effects of credit supply on borrowers. From the distribution of newly originated mortgages, there appear to be many scores that exhibit discontinuities in the number of mortgages originated. However, in the period since the financial crisis, the two most prominent discontinuities occur at 620 and 640 and we use these thresholds in our analysis. Our credit availability measure is constructed to capture the difference in the ability of borrowers just above those thresholds to obtain mortgages compared to borrowers just below them. In practice, computing this measure requires two steps. First, we need to estimate the impact of falling above or below the threshold at each point in time. Second, we need to determine how likely it is that each individual would actually fall above the threshold if she applied for a mortgage. In this section, we describe the implementation of these two steps.

5

2.2.1

Credit Score Thresholds in Originated Mortgages

In order to identify the use of the thresholds, we look at the distribution of credit scores on loans originated each quarter, as captured in a data set of mortgages provided by Black Knight Financial Services (formerly known as “LPS” and “McDash”). For each mortgage, Black Knight reports detailed information including the origination date, the loan-to-value ratio, the debt-to-income ratio, and the borrower’s credit score. Importantly for our purposes, the credit score reported in the data is the FICO score used in the lender’s mortgage underwriting decision, an issue we return to below. As discussed above, figure 1 plots the density and cumulative distribution of FICO scores for mortgages in the Black Knight data originated in 2005, 2008, 2010, and 2012. We quantify the size of the the 620 and 640 thresholds by calculating the ratio of the number of mortgages originated within five points below the threshold compared to the number of mortgages originated to borrowers with credit scores within five points above the threshold. Even prior to the crisis (2005; the black line in figure 1), lenders appear to have used 620 as a relevant threshold in their lending decisions.4 Only 70 percent as many mortgages were originated to borrowers just below the thresholds compared to those just above. In contrast, the ratio around 640 was about 90 percent, suggesting that 640 was not a particularly important score in underwriting decisions during that time period. These ratios were similar in 2008 (the red line). By 2010 (the blue line), however, the ratio at 620 had plummeted to just 20 percent, suggesting a dramatic tightening of mortgage credit for borrowers with credit scores under 620. By 2012 (the green line), the ratio at 640 had also fallen sharply, to about 45 percent.5 These ratios have changed relatively little since 2012. 2.2.2

Using Credit Scores in the Consumer Credit Panel

The second, less obvious, step in computing our mortgage credit availability measure is identifying whether each individual in the population has a credit score that falls above or below the relevant threshold. In principle, all we would need to do this is to observe the 4 As described in the introduction, Keys et al. (2010) argue that loans with credit scores above 620 were easier to securitize, while Bubb and Kaufman (2014) dispute this conclusion. 5 As the number of mortgages to borrowers with credit scores between 620 and 640 fell between 2010 and 2012, the ratio at 620 actually rose back to 40 percent, a mechanical response to the decrease in loans to borrowers with scores just above 620, the denominator. A combined measure of the two discontinuities, which calculates the ratio of mortgages just above 640 to the number of mortgages just below 620, shows an overall tightening during this period.

6

individual’s FICO score at a given point in time. In practice, there are two complications. First, a FICO score is the output of a proprietary scoring model, which has changed over time, applied to data reported by any one of the three credit bureaus. As a result, there is no single “FICO score” for an individual at any given point in time. Moreover, scores change almost continuously as new information is reported to the credit bureaus. The scores reported in the Black Knight data, which we used to construct figure 1, are the results of the particular scoring model and credit bureau data captured at the time of underwriting. For both these reasons, even if we observed some FICO score from around the same time that a mortgage was originated, it would not necessarily match exactly to the score reported in the Black Knight data. The empirical relevance of the observed 620 and 640 thresholds in a different data set is thus something that we need to test for, not something that we can assume. The second complication is that we do not observe any FICO scores in our main data for this project, which is the Equifax consumer credit panel (CCP) from the Federal Reserve Bank of New York. Instead, the CCP contains a “riskscore”, which is a similar credit score intended to capture the probability that individual will default on any loan. In order to relate the riskscore in the CCP to a FICO score, we use a linked monthly panel data set that contains both types of credit scores. Using the joint distribution of the CCP riskscores and FICO scores, we predict the probability that an individual with a given riskscore in the CCP would have a FICO score (using the particular model and credit bureau data in the linked data set) that exceeded the a given threshold value.6

2.3

Our Credit Availability Measure

Combining these two steps produces our mortgage credit availability measure, which captures the ability of each borrower to obtain a mortgage as a result of her credit score and lenders’ use of credit score thresholds in making underwriting decisions at that point in time. Roughly speaking, a borrower has access relative to a particular threshold if either she has a score above the threshold or if the threshold is not relevant because the ratio of loans just below to loans just above is close to one. Conversely, she lacks access if he has a score below the threshold and the threshold matters. Differences in risk scores across individuals produce cross-sectional variation in our credit availability measure, while changes in the importance 6

The linked data only contain information on mortgage borrowers, which is why we cannot use them for our main estimates. To characterize the relationship between the probability that a FICO score exceeds a threshold and the CCP riskscore, we estimate logit models using data six months prior to origination and allow the relationship between the two scores to vary across years.

7

of the credit-score thresholds across time produce time series variation. We compute our measure separately for both the 620 and 640 thresholds and combine these into our preferred measure of credit availability. Formally, for an individual with a credit score in the consumer credit panel given by “CCP score,” we express the probability that he would have a FICO score over 620 as P r(F ICO >= 620|CCP scorei )y(t) , where t and y(t) are the the quarter and year of observation, respectively. As described above, we measure the importance of having a score above 620 in quarter t by looking at the ratio of the number of loans with FICO scores just below 620 ((Loan Count|F ICO ≥ 615, F ICO < 620)t ) to the number of loans with FICO score just below 620 ((Loan Count|F ICO ≥ 620, F ICO < 625)t ), using 5-point windows. Our measure of mortgage credit availability from the 620 threshold is the product of these two factors: credit avail620 i,t =1 − (1 − P r(F ICO >= 620|CCP scorei )y(t) ) × (1 − (Loan Count|F ICO ≥ 615, F ICO < 620)t / (Loan Count|F ICO ≥ 620, F ICO < 625)t ) Similarly, we produce a measure that captures mortgage credit availability due to the threshold at 640: credit avail640 i,t =1 − (1 − P r(F ICO >= 640|CCP scorei )y(t) ) × (1 − (Loan Count|F ICO ≥ 635, F ICO < 640)t / (Loan Count|F ICO ≥ 640, F ICO < 645)t ) For the composite measure, both the 620 and 640 measures need to be greater than zero, since the relevant constraint is the most binding one. Our final measure of credit availability is then 640 credit availi,t = credit avail620 i,t × credit availi,t . Although it is easy to think of credit availi,t in a binary context—one either has access to credit or one does not—in our data it is a continuous variable with outcomes ranging from 0 to 1, both because the link between CCP riskscore and FICO threshold is probabilistic and because our quantification of the importance of the threshold is never actually 0 or 1. Figure 3 shows the evolution of the credit availability measure. The left panel shows the time series of average credit availability for individuals with Equifax riskscores between

8

530 and 730, our estimation sample. The timing of the sharp drops in the series correspond to the narrative provided above and the thresholds we identified in the Black Knight data. The three shaded regions denote periods between 2008 and 2011 in which availability was roughly stable. Taking a different slice through the data, the right panel compares average credit availability, by 10-point Equifax riskscore bin, across those three stable periods of credit availability around the changes in the threshold. Because credit scores are highly correlated with characteristics like age and income, the imposition of the thresholds has had disproportionate effects on particular segments of the population. Figure 4 shows the relationship between credit availability and age. The left panel plots the distribution of credit scores by age, while the right panel compares average credit availability by age across the three stable periods of availability shown in figure 3.7 These figures make clear that the introduction of the credit score thresholds since 2008 has had much larger implications for the young: As shown by the blue line in the right panel, just two-thirds of 30 year-olds had access to mortgage credit in 2011 by this measure, compared with 90 percent of 65 year-olds.

2.4

Estimation Sample

The Equifax/FRBNY CCP consists of a 5 percent random sample of people 18-years and older who have a credit file. For computational tractability, we use a random sample containing 5 percent of the individuals in the panel, or a 0.25 percent sample of the population. We restrict our estimation sample to the years 2008-2011, a period when we can clearly identify changes in credit availability, as discussed above. Ending our sample in 2011 has the further advantage that we are able to observe everyone in our sample through 2015, a full four years after the end of the estimation period, allowing us to estimate longer-term effects of our credit availability measure. We restrict our analysis to borrowers within a relatively narrow range of risk-scores around the thresholds at 620 and 640 that we identified above. This restriction has two motivations. First, borrowers with credit scores far from the threshold values are less likely to be affected by lender’s use of these thresholds in making lending decisions. Results suggesting that such borrowers are significantly affected by these mortgage thresholds are likely to be spurious. Second, the relationship between credit score and mortgage demand is likely non-linear. However, within a narrow band of scores, a linear function of credit score 7 The geographic variation in credit availability over this period is also quite interesting, and we plan to include more results along these lines in future drafts.

9

should be a reasonable control for other effects of credit score on outcomes of interest. Our baseline specification uses a sample of borrowers with scores between 530 and 730, but our robustness checks (not included in this draft) indicate that the main results are not sensitive to the size of the window.

3

Results

Having constructed a measure of mortgage credit availability for each member of the consumer credit panel, we next explore the relationship between this measure and various outcomes. Depending on the outcome, the models can be linear regressions, logit models in the case of probabilities, or negative binomial models in the case of count variables. For each outcome variable, we consider horizons of 4, 8, 12, and 16 quarters to assess both the short-term and longer-term effects of restrictions to mortgage credit. In each model, we include dummy variables for the quarter of observation and also an interaction of this quarter dummy with riskscore. Each regression also includes the first quarterly lag of credit availability for the individual, and lagged credit score interacted with the quarter dummies, which we discuss below. In the table for each specification, we report results using the entire sample and also separately for those who already have a mortgage and those who do not. In determining whether someone has a mortgage, we use total outstanding balance on all mortgages appearing on her credit report and say an individual has a mortgage if the total is greater than zero. Finally, it is worth noting that the coefficients on our mortgage credit availability measure capture the differences between a borrower with a credit availability of one, meaning she is unaffected by minimum credit scores, and a hypothetical borrower with credit availability of zero, meaning she falls for certain below the credit-score threshold and we observe no mortgages to borrowers with credit scores just below this threshold. In practice, we always estimate some positive probability of even a low-riskscore individual being above a FICO threshold, and we always see some mortgages issued below the FICO thresholds in the Black Knight data. As a result, our credit availability measure is never less than about 0.2. As shown in figure 3B, for borrowers with scores toward the bottom of our range, credit availability fell from about 0.7 to 0.2 between 2008 and 2011.8 As we noted in the introduction, we identify the effects of credit availability on an outcome 8

We also note that we are cautious about using our measure to compare people with very high credit scores to those with very low credit scores, as our identification comes largely from the curvature in our measure around the credit-score thresholds at 620 and 640.

10

using the timing and nonlinearity of our availability measure. Formally, we require that the credit availability measure be uncorrelated with any other factors affecting that outcome, conditional on the controls, especially the quarter fixed effects and quarter fixed effects interacted with linear riskscore. Thus our identification is secure against any confounding factors that vary only in the time series dimension, as well as any that are correlated with credit score in a linear fashion, even if that linear relationship with credit score shifts over time. In particular, our view is that credit demand could be correlated over time with the level and slope of many of our outcomes but that it is unlikely to have a nonlinear effect with those outcomes that happens to shift at the particular times our credit availability measure does.

3.1

Interpreting Lagged Credit Scores

As mentioned above, our baseline specification includes lagged credit score interacted with the quarter dummies. Our choice to include the individual’s lagged credit score has several possible interpretations and we are agnostic as to which one is most appropriate. First, the lagged credit score may simply be viewed as a control, in that we may be concerned that the non-linear dependence of our credit availability measure on credit score means that our credit availability measure is simply picking some other non-linear response of a dependent variable to credit score, which may be unrelated to mortgage credit. Because lagged credit score exhibits the same non-linear dependence on credit score, including it in the regression should control for any such effects. Second, we may be directly interested in the effect of a person’s previous access to mortgage credit on her current decisions. For example, if prior credit availability was low, that may signal pent up demand for mortgage credit. Alternatively, if previous credit availability was high and the individual did not subsequently take out a mortgage, that may indicate someone with lower than expected demand for mortgage credit. Finally, including lagged mortgage credit may be interpreted as an attempt to identify the effects of changes in an individual’s access to credit. For example, if borrowers took out mortgages during quarters when we measure an increase in their access to mortgage credit, that could be an additional result on the impact of mortgage credit availability.

11

3.2

Mortgage outcomes

Our first set of models is intended in part to confirm that our measure of mortgage credit availability actually captures borrowers’ ability to obtain a mortgage. In these models, the dependent variable is whether the person takes out one or more new mortgages within the specified horizon and we use a logit specification. We use the CCP’s trade line data on individual mortgages to determine the date on which the mortgage was opened.9 In addition to considering longer horizons, this first set of regressions also includes a specification where the outcome variable is whether there in a new mortgage in the following quarter. We can get a good sense of the data by examining plots of the relationship between credit score and the probability of taking out a mortgage. Figure 5 shows the one-quarter probability of mortgage attainment by credit score, across the three stable periods of availability in our data. Just as credit availability, as shown in figure 3B, declined most sharply for those at the bottom of the credit distribution between the 2008 (the black lines) and 2009:Q2-2010:Q2 periods (the red lines), so does the probability of taking out a mortgage. Once the 640 threshold came into being, the 2012 line (the blue line) shows evidence of a further decline in mortgage originations in the middle of our sample, again similar to the pattern in figure 3B. Importantly, neither credit availability nor mortgage origination probability shows any significant gaps between the three lines for the higher riskscores in our sample. More formally, our first main result is shown in the first column of Panel A of table 1. Even after controlling for time effects and credit score, we estimate that the average marginal effect from our credit availability measure of the probability of taking out a new first mortgage in the following quarter is 2.5 percentage points and is highly statistically significant. This estimate is also very large and economically significant, compared to the average probability in our sample of taking out a new mortgage (“Dep. Var. Mean”), which is just 0.9 percent. This result confirms both the importance of these credit score thresholds in determining who receives mortgages and also the ability of our credit availability measure to capture these effects. We can also look at longer horizons, again starting graphically. Figure 6 shows the cumulative probabilities of taking out a mortgage, with each panel plotting a different horizon. The figure indicates that the effects shrink, at least in relative terms, as the horizon length9

This is a subtle but important step. Many of the aggregate variables in the CCP only update with a lag as the information is reported to Equifax. For example, a change in an individual’s reported mortgage balance may occur only one or two quarters after they actually take out a mortgage. By using the dates from the trade lines, we are able to precisely measure the timing of the mortgage origination.

12

ens: While the gaps in panel A (1 to 4 quarters) are still quite large, the gaps in panel D (1 to 16 quarters) are much smaller, at least relative to the range on the y-axis. In the corresponding columns 2 through 5 of panel A of table 1, the coefficient on our credit availability measure remains statistically significant and increases in magnitude in percentage point terms. However, the coefficient rises much more slowly than the mean of the dependent variable, suggesting that our measure of mortgage credit access becomes less important over time compared to other factors that determine whether people take out new mortgages. Considering whether people take out any new mortgages over the following four quarters, the average marginal effect of our measure is 6.0 percentage points, while overall, 3.5 percent of people in the sample take out a mortgage within this period. At a 16-quarter horizon, the average marginal effect of our measure rises to 10.0 percentage points, but this estimate is smaller relative to the average probability of taking out a mortgage over this horizon, which is 13 percent. While attenuated relative to the short run, these relative effects are still very large, suggesting that the effects of credit availability are quite persistent, a result difficult to discern in the graphical analysis discussed above. In panels B and C, we repeat our analysis on the sub-samples of people who already have positive mortgage balances and those who have no previous balance. Because our sample is relatively concentrated towards the bottom of the credit score distribution, the sub-sample of people with no mortgage balance makes up about 85 percent of our estimation sample and the average marginal effects for this group are similar to the effects for the sample as a whole, though they are slightly smaller at shorter horizons. For people who already have a mortgage, the probability of taking out an additional mortgage is considerably higher—likely because many of these individuals are refinancing an existing mortgage during this period of falling interest rates—and the estimated marginal effects from our credit availability measure are a bit larger. However, relative to this group’s average baseline probability of taking out a mortgage, the coefficients are actually smaller, suggesting that credit availability plays a smaller role in borrowing decisions for those who already have a mortgage. In table 2, the dependent variable is the total number of new mortgages taken out over the various horizons and we estimate a negative binomial model.10 The results here are quite similar to those presented above. Indeed, at short horizons they are nearly identical to those in table 1, because few people take out more than one new first mortgage and thus the probability of taking out a mortgage is nearly the same as the average number of new mortgages. At longer horizons the average marginal effects are somewhat larger 10

We use these results to calculate our counterfactual in Section 4.

13

in magnitude than in the previous table. Since most of these differences must come from mortgages taken out subsequent to the initial period of credit availability, they suggest that getting a mortgage in one period can improve the chances of taking out another one subsequently, perhaps through refinancing. In tables 3 and 4, we consider alternative measures of new mortgage borrowing, including the change in the number of mortgages and the change in the total mortgage balance on an individual’s credit record. Qualitatively, the results from these exercises are similar to the results in tables 1 and 2.

3.3

Additional Outcomes

Aside from the direct question of whether restrictions on mortgage credit are preventing households from obtaining mortgages and how these effects attenuate over time, we are also interested in understanding the broader relevance of credit supply. We are limited in this exploration to the set of outcomes observable in the consumer credit panel. 3.3.1

Derogatory Credit Events

We next consider whether access to mortgage credit can allow households to avoid negative credit events. In table 5, we show results of a logit estimation where the dependent variable is whether individuals have had a “major derogatory event” (such as a foreclosure, bankruptcy or charge-off) recorded on their mortgage credit history in the past 24 months. For the larger sample in panel A, we find no statistically or economically significant effects. However, these results seem to mask more interesting patterns in the sub-samples. In panel B, among people with no prior mortgage balance, we see that credit availability does seem to cause derogatory events, although mostly with a lag. This pattern makes sense given that these individuals need to take out mortgages in order to subsequently experience a mortgage-related derogatory event. The effects are much smaller than the effects on mortgage attainment in table 1, suggesting that only a small fraction of the mortgages taken out by those with credit availability experience distress within a few years. In contrast, for those who already have a mortgage balance (panel C), we find that continued access to mortgage credit lowers their probability of having a derogatory event within four quarters by 4 percentage points, a very large effect relative to the dependent variable mean. The effect is similar in percentage point terms at longer horizons. We conjecture that these homeowners are able to avoid such an event either by borrowing against their houses through a cash-out refinance

14

or second mortgage, or else by lowering their existing mortgage payments by refinancing at a lower interest rate. 3.3.2

Migration

Because we observe the residence of an individual in the CCP down to the Census block, we can also examine the effects of credit availability on moving and migration decisions. The address data in the CCP tend to be unstable because they reflect the most recent address reported to Equifax, which can fluctuate back and forth if that person is receiving bills at more than one address. To try to isolate actual moves, we limit the sample to those individuals who we can observe in a location for at least four quarters before we measure their credit availability and who appear to remain in a location for four quarters after the end of the horizon we use. As a consequence of this approach, the samples are smaller and we can only show effects out through 12 quarters, because we cannot establish address stability for 2015 observations, which are 16 quarters out from 2012, the end of our estimation sample. Tables 6 and 7 show the effects on the individual’s probability of moving across blocks and states, respectively. In panel A of table 6, we see small positive effects at short horizons and small negative effects at longer horizons. As before, however, these estimates mask heterogeneity between those who do and do not already have a mortgage balance. For those who do not (panel B), we see somewhat larger positive effects at shorter horizons and smaller negative effects at longer horizons. The positive effect in particular is sensible, since nonhomeowners who have credit available to them usually have to move to buy a home, which we observed them doing in table 1. The negative longer-term effect may occur because this is a highly mobile group—27 percent move at least once over 12 quarters, even imposing our stability requirement—and so the relative probability of moving among those with credit availability is lower at this horizon. For those who do have a mortgage balance (panel C), the effects are negative and grow over time, suggesting that having the option to refinance leads some of these homeowners to remain in their homes for longer. In table 7, we see uniformly negative but mostly relatively small and insignificant effects of credit availability on cross-state migration. The effect us statistically distinguishable from zero in column 3 of panel C, however, and at -1.0 percentage point is quite large relative to the mean probability of moving across states, which is only 2.3 percent. Again, access to mortgage credit may be allowing some homeowners to remain in their homes by refinancing, preventing them from moving to a different state. While presumably beneficial to these individuals, via revealed preference, this effect could interfere with the re-balancing of the 15

labor market away from economically depressed areas. 3.3.3

Auto Loans

Finally, we explore whether we can observe interactions between mortgaging borrowing and other kinds of consumer credit. In particular, we consider whether our measure of mortgage credit availability has implications for consumers’ use of auto loans. Results from this exercise are shown in table 8, where the dependent variable is the change in the number of auto loans on the individuals credit record, and table 9, where we use the change in the total auto loan balance. In theory, the effect could go either way. On one hand, individuals who cannot buy a house because they are denied mortgage credit could instead substitute into cars. On the other hand, home purchases could be positively correlated with auto lending because of complementarities between homeownership and driving behavior, among new purchasers (panel B), or the ability to refinance, among prior borrowers (panel C). The results here, which are uniformly positive and statistically significant, strongly suggest that the complementarity and refinancing channels dominate.

4

Counterfactual

To gauge the magnitude of our estimates, we perform a simple counterfactual experiment using some of the results from table 2, which shows the effects of our credit availability measure on the total number of first mortgages taken out by an individual. Specifically, we predict the number of first mortgages that would have been originated if the credit availability measure had remained at its level in the first quarter of 2008, holding all else constant. First, using the specification in column 2, panel A, we predict the number of mortgages that would have been originated one to four quarters ahead. Using data from the fourth quarter of 2011, when the full effect of the 640 threshold had kicked in, we find that the imposition of the thresholds lowered mortgage originations in our estimation sample—people with riskscores between 530 and 730—by 700,000 in 2012. We can divide that figure by the 3.1 million mortgages that were actually originated in our sample to conclude that mortgage originations would have been more than 20 percent higher without the thresholds. For a broader comparison, we note that first mortgage originations in 2012 to people of all credit scores totaled about 15 million. If we assume, somewhat conservatively, that the thresholds had no effect one people with scores outside of the 530 to 730 range, then total mortgage 16

originations would have been about 4.5 percent higher. We can take a longer view by doing essentially the same exercise with the specification in column 5, panel A, and predicting the number of first mortgages that would have been originated one to 16 quarters ahead. Again using data from the fourth quarter of 2011, we find that the imposition of the thresholds lowered originations in our sample by 2.2 million between 2012 and 2015. Dividing that figure by the 12.4 million mortgages actually taken out by individuals in our sample yields an effect of about 18 percent. Comparing the 2.2 million figure to 45 million, the total number of first mortgages originated to all individuals from 2012 to 2015, implies that there would have been about 5 percent more mortgages originated, similar to the effect we calculate for 2012 alone.

5

Conclusion

The question of how tight mortgage credit should be is an important one that forces policy makers to balance concerns in both directions. On the one hand, tight mortgage credit prevents marginal borrowers from realizing the benefits of homeownership and, on a larger scale, reduces economic activity in the housing sector. On the other hand, the recent financial crisis demonstrates the risk of mortgage credit that is too loose: Banks’ losses on defaulting mortgages can cause instability in the financial sector, borrowers may take out loans they are unable to repay, and an excess supply of credit can potentially contribute to a bubble in housing prices. Our paper aims to shed additional light on one of these issues, the effect of mortgage credit on individual borrowers. We find that lenders’ restrictions on mortgage credit through the use of minimum credit score thresholds have very large negative effects on borrowing. Furthermore, although these effects attenuate somewhat over time, we find that they persist for several years, suggesting that the impact of these policies on the welfare of constrained households could be quite large. Further research is necessary to study these effects together with the other consequences of tight mortgage credit and thereby give policy makers a better understanding of the total effects of policies that may reduce or expand borrowers’ access to mortgages.

17

References Acolin, A, J Bricker, PS Calem, and SM Wachter, “Borrowing Constraints and Homeownership,” American Economic Review: Papers and Proceedings, 2016, 5. Anenberg, Elliot, Aurel Hizmo, Edward Kung, and Raven Molloy, “The Effect of Mortgage Credit Availability on House Prices and Construction: Evidence from a Frontier Estimation Approach,” May 2016. Working Paper. Barakova, Irina, Paul S Calem, and Susan M Wachter, “Borrowing Constraints During the Housing Bubble,” Journal of Housing Economics, 2014, 24, 4–20. , Raphael W Bostic, Paul S Calem, and Susan M Wachter, “Does Credit Quality Matter for Homeownership,” Journal of Housing Economics, 2003, 12 (4), 318–336. Bubb, Ryan and Alex Kaufman, “Securitization and Moral Hazard: Evidence from Credit Score Cutoff Rules,” Journal of Monetary Economics, 2014, 63, 1–18. Chomsisengphet, Souphala and Ronel Elul, “Bankruptcy Exemptions, Credit History, and the Mortgage Market,” Journal of Urban Economics, 2006, 59, 171–188. Gete, Pedro and Michael Reher, “Systemic Banks, Mortgage Supply and Housing Rents,” 2016. Working Paper. Gropp, Reint, John Krainer, and Elizabeth Laderman, “Did Consumers Want Less Debt? Consumer Credit Demand versus Supply in the Wake of the 2008-2009 Financial Crisis,” January 2014. SAFE Working Paper No. 42. Keys, Benjamin J., Tanmoy Mukherjee, Amit Seru, and Vikrant Vig, “Financial Regulation and Securitization: Evidence from Subprime Loans,” Journal of Monetary Economics, 2009, 56 (5), 700–720. , , , and , “Did Securitization Lead to Lax Screening? Evidence from Subprime Loans,” Quarterly Journal of Economics, 2010, 125 (1), 307–362. , , , and , “Lender Screening and the Role of Securitization: Evidence from Prime and Subprime Mortgage Markets,” Review of Financial Studies, 2012, 25 (7), 2071–2108. Rosenthal, Stuart S., “Eliminating Credit Barriers: How Far Can We Go,” Low-Income Homewonership, 2002, pp. 111–145. 18

640 ●

● ●

4



















640

● ● ●

● ●





40











620

2005 2008 2010 2012



● ●









● ●











2



19

20

1



10

● ●

● ●

550

● ●

● ●











● ●

● ●



600









0

0

● ●







● ●









● ●









● ●



● ●



● ●



● ●

30





Percent



3

Percent





60





620

2005 2008 2010 2012

50

5



650 FICO score bin

Panel A. Densities

700

● ● ●

● ● ●

● ● ●

● ● ●

550

● ● ●



● ●

● ●

● ●

● ●

600

● ●

















650

700

FICO score bin

Panel B. Cumulative Distributions

Fig. 1.—Mortgages by FICO Score. This figure plots the densities and cumulative distributions of newly originated first mortgages by 10-point FICO score bin in the Black Knight data set, across four years.

3.5

640

● ●

640

● ●



Percent

1.5

● ●







1.5

2.0

● ●

2.0





Percent



620

Conventional (GSE) Convent. (Non−GSE) FHA VA Other

2.5



620

Conventional (GSE) Convent. (Non−GSE) FHA VA Other

2.5

3.0



3.0

3.5 ●



● ● ●

● ●





1.0

1.0





0.0

● ●







● ●



0.5









● ● ●

























550





























600

● ●













650















0.0

0.5





● ●



● ●

● ●

700

● ●

● ●

● ●

● ●

● ●

550



3.5 ● ●

● ●











650

















700

640





Percent





● ● ●

2.0

2.5



620

Conventional (GSE) Convent. (Non−GSE) FHA VA Other

1.5

2.0

2.5

3.0

640

1.5

Percent

● ●



Panel B. 2008

3.0

3.5

20



620

Conventional (GSE) Convent. (Non−GSE) FHA VA Other

● ●



FICO score bin

Panel A. 2005



● ●

600

FICO score bin









● ●







● ●



1.0

1.0



● ● ● ●

● ●

● ●

● ●

550

● ●

● ●









● ●

600



● ●

● ●

● ●

● ●

● ●

● ●

● ●

● ●

● ●



● ●



● ●

0.0



0.0



0.5

0.5



● ●

650 FICO score bin

Panel C. 2010

700

● ●

● ●

● ●

550

● ●

● ●

● ●

● ●

● ● ●

600



● ●

● ●



● ●

● ●

● ●













650

● ●

● ●



● ●

● ●

700

FICO score bin

Panel D. 2012

Fig. 2.—Mortgage Densities, across Types. This figure plots the densities of newly originated first mortgages by 10-point FICO score bin in the Black Knight data set, across types of loans. Each panel shows the data from a separate year.



● ● ●

● ●



1.0

● ●

● ●





● ●



● ● ● ● ●

● ●







● ● ●





0.80











0.75

● ● ● ●

● ●

● ●

● ●













● ●

● ●









● ●











0.65



0.60



Aggregate Using 620 Threshold Using 640 Threshold

0.2

21











2008



0.4



0.70



2006

● ●



● ●



● ●







● ●







0.85



● ●















0.8

0.90





● ●





Credit Availability Measure





2008 2009:Q3−2010:Q2 2011

0.6



Credit Availability Measure

0.95

● ● ● ●

2010 Time

Panel A. Time Series

2012

2014

● ●







● ● ●





550

600

650

700

EFX Riskscore Bin

Panel B. Cross-Section Across Periods

Fig. 3.—Credit Availability. This figure shows the evolution of the credit availability measure in two different ways. The left panel plots the time series of average credit availability for all individuals with Equifax riskscores between 530 and 730. The three shaded regions denote periods between 2008 and 2011 in which availability was roughly stable. The right panel compares average credit availability, by 10-point Equifax riskscore bin, across the three stable periods of credit availability.

0.95

800

● ●

2008 2009:Q3−2010:Q2 2011

● ● ●



0.85



40

50

60

70

Age Bin

Panel A. Credit Score by Age

80



0.80



● ●



0.70



● ●



0.65

550

30





600

22 20









Median 25th / 75th Percentile 10th / 90th Percentile





0.75

Credit Availability Measure

700 650

EFX Riskscore

750

0.90









30

40

50

60

70

80

Age Bin

Panel B. Credit Availability by Age

Fig. 4.—Credit Availability by Age. This figure shows the relationship between credit availability and age. The left panel plots the distribution of credit scores by 5-point age bin. The right panel compares average credit availability, by 5-point age bin, across the three stable periods of credit availability.

2.0 ● ●

2008 2009:Q3−2010:Q2 2011

● ●



● ●



1.5



● ●





● ●

1.0

Percent













● ●

● ●

0.5

● ●

● ●





● ●

● ●











0.0



● ●



550

600

650

700

EFX Riskscore Bin Fig. 5.—One-Quarter Mortgage Origination Probability. This figure compares the probability of taking out at least one mortgage in the next quarter, by 10-point Equifax riskscore bin, across the three stable periods of credit availability.

23



2008 2009:Q3−2010:Q2 2011



● ●





15

8 ●





2008 2009:Q3−2010:Q2 2011

● ●





● ●



6







● ●

● ●



10

● ●

Percent

● ●

4

Percent













● ●

● ●

● ●



● ●







5



● ● ●

2















● ●

● ●

● ●

● ●

● ●



● ●







● ●



● ●





0

● ●





● ● ● ●

550

600

650

700

550

600

EFX Riskscore Bin

Panel A. 1-4 Quarters



700

Panel B. 1-8 Quarters

2008 2009:Q3−2010:Q2 2011

● ●



25

20



650

EFX Riskscore Bin



● ●

2008 2009:Q3−2010:Q2 2011

● ●

● ● ●

● ●









● ●

15



20





● ●



● ● ●



● ●

10

● ●





15



Percent

Percent



● ●



















10





● ●











5

● ● ●

● ● ● ●

5

● ● ● ●





























550

600

650

700





550

EFX Riskscore Bin





600

650

700

EFX Riskscore Bin

Panel C. 1-12 Quarters

Panel D. 1-16 Quarters

Fig. 6.—Longer-Horizon Mortgage Origination Probabilities. This figure compares the probability of taking out at least one mortgage, by 10-point Equifax riskscore bin, across the three stable periods of credit availability. Each panel shows a different horizon.

24

TABLE 1 Effects on Probability of Taking Out a First Mortgage

Horizon in Quarters:

(1) 1

(2) 1-4

(3) 1-8

(4) 1-12

(5) 1-16

Panel A: Entire Sample Credit Availability

0.025 (0.002)

0.060 (0.003)

0.084 (0.004)

0.093 (0.005)

0.100 (0.006)

Lagged Availability

-0.008 (0.002)

-0.012 (0.003)

-0.010 (0.005)

-0.005 (0.006)

0.000 (0.007)

0.009 3,254,299

0.035 3,202,697

0.068 3,147,877

0.100 3,098,179

0.130 3,048,165

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

0.021 (0.001)

0.051 (0.003)

0.080 (0.004)

0.093 (0.005)

0.110 (0.006)

Lagged Availability

-0.007 (0.002)

-0.009 (0.003)

-0.011 (0.004)

-0.006 (0.006)

0.000 (0.006)

0.007 2,774,306

0.027 2,727,902

0.053 2,678,177

0.078 2,632,851

0.100 2,586,787

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

0.050 (0.005)

0.120 (0.009)

0.130 (0.012)

0.140 (0.014)

0.130 (0.015)

Lagged Availability

-0.009 (0.005)

-0.005 (0.010)

0.026 (0.013)

0.045 (0.014)

0.047 (0.015)

Dep. Var. Mean Observations

0.020 479,993

0.079 474,795

0.150 469,700

0.220 465,328

0.280 461,378

Note.—Logit estimates of effect of credit availability on the cumulative probability of taking out a mortgage, at various horizons. Average marginal effects, with standard errors clustered at quarterriskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

25

TABLE 2 Effects on Total Number of New First Mortgages

Horizon in Quarters:

(1) 1

(2) 1-4

(3) 1-8

(4) 1-12

(5) 1-16

Panel A: Entire Sample Credit Availability

0.025 (0.002)

0.064 (0.003)

0.100 (0.005)

0.120 (0.006)

0.150 (0.008)

Lagged Availability

-0.008 (0.002)

-0.010 (0.003)

-0.007 (0.005)

0.000 (0.007)

0.007 (0.009)

0.009 3,254,299

0.036 3,202,697

0.074 3,147,877

0.110 3,098,179

0.160 3,048,165

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

0.021 (0.002)

0.055 (0.003)

0.092 (0.005)

0.120 (0.006)

0.140 (0.008)

Lagged Availability

-0.007 (0.002)

-0.009 (0.003)

-0.010 (0.005)

-0.006 (0.007)

-0.001 (0.008)

0.007 2,774,306

0.028 2,727,902

0.057 2,678,177

0.088 2,632,851

0.120 2,586,787

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

0.050 (0.005)

0.130 (0.010)

0.180 (0.015)

0.220 (0.019)

0.250 (0.022)

Lagged Availability

-0.007 (0.006)

0.004 (0.011)

0.046 (0.016)

0.079 (0.019)

0.096 (0.022)

Dep. Var. Mean Observations

0.021 479,993

0.084 474,795

0.170 469,700

0.260 465,328

0.340 461,378

Note.—Negative binomial estimates of effect of credit availability on the number of new mortgages taken out, over various horizons. Average marginal effects, with standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

26

TABLE 3 Effects on Change in Number of First Mortgages

Horizon in Quarters:

(1) 4

(2) 8

(3) 12

(4) 16

Panel A: Entire Sample Credit Availability

0.070 (0.005)

0.093 (0.006)

0.110 (0.007)

0.110 (0.008)

Lagged Availability

-0.024 (0.005)

-0.017 (0.007)

-0.005 (0.008)

0.006 (0.009)

0.000 3,072,513

-0.001 2,967,912

0.000 2,889,856

0.003 2,827,019

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

0.049 (0.004)

0.065 (0.005)

0.076 (0.006)

0.080 (0.007)

Lagged Availability

-0.029 (0.004)

-0.029 (0.006)

-0.026 (0.007)

-0.019 (0.007)

0.018 2,597,016

0.031 2,496,269

0.045 2,422,101

0.060 2,362,781

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

0.180 (0.013)

0.230 (0.017)

0.230 (0.020)

0.210 (0.021)

Lagged Availability

-0.001 (0.013)

0.014 (0.018)

0.032 (0.020)

0.031 (0.022)

Dep. Var. Mean Observations

-0.094 475,497

-0.170 471,643

-0.230 467,755

-0.290 464,238

Note.—Linear regression estimates of effect of credit availability on the change in the number of mortgages on an individual’s credit record, over various horizons. Standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

27

TABLE 4 Effects on Change in First Mortgage Balance

Horizon in Quarters:

(1) 4

(2) 8

(3) 12

(4) 16

Panel A: Entire Sample Credit Availability

10732 (794)

12775 (1046)

12792 (1247)

14308 (1367)

Lagged Availability

-6216 (859)

-5796 (1168)

-2970 (1398)

-3137 (1534)

-270 3,212,605

-773 3,174,450

-865 3,143,800

-536 3,115,545

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

5237 (502)

4860 (638)

4057 (773)

3332 (898)

Lagged Availability

-6427 (551)

-6707 (695)

-6030 (855)

-5814 (983)

3067 2,736,148

5217 2,700,688

7259 2,672,031

9434 2,645,157

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

42803 (3275)

54428 (4144)

53860 (4879)

59292 (5185)

Lagged Availability

-1209 (3464)

-202 (4516)

8095 (5211)

3222 (5621)

Dep. Var. Mean Observations

-19435 476,457

-34921 473,762

-46874 471,769

-56606 470,388

Note.—Linear regression estimates of effect of credit availability on the change in an individual’s mortgage balance, over various horizons. Standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

28

TABLE 5 Effects on Probability of Major Derogatory Mortgage Event

Horizon in Quarters:

(1) 1-4

(2) 1-8

(3) 1-12

(4) 1-16

Panel A: Entire Sample Credit Availability

-0.003 (0.003)

-0.003 (0.004)

-0.001 (0.004)

0.003 (0.004)

Lagged Availability

-0.010 (0.003)

-0.015 (0.004)

-0.017 (0.005)

-0.020 (0.005)

0.020 3,132,942

0.030 3,010,127

0.038 2,903,118

0.044 2,805,795

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

0.005 (0.003)

0.005 (0.003)

0.008 (0.004)

0.011 (0.004)

Lagged Availability

-0.013 (0.003)

-0.017 (0.004)

-0.019 (0.004)

-0.023 (0.004)

0.014 2,658,909

0.020 2,541,865

0.025 2,440,381

0.029 2,348,412

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

-0.043 (0.008)

-0.051 (0.009)

-0.052 (0.010)

-0.048 (0.011)

Lagged Availability

-0.011 (0.008)

-0.020 (0.010)

-0.017 (0.011)

-0.015 (0.012)

Dep. Var. Mean Observations

0.053 474,033

0.083 468,262

0.100 462,737

0.120 457,383

Note.—Logit estimates of effect of credit availability on the cumulative probability of having a “major derogatory” mortgage event (foreclousure, bankruptcy, charge off, etc.) on an individual’s credit record within the last 24 months, at various horizons. Average marginal effects, with standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

29

TABLE 6 Effects on Moving to Different Census Block

Horizon in Quarters:

(1) 4

(2) 8

(3) 12

Panel A: Entire Sample Credit Availability

0.011 (0.005)

-0.006 (0.007)

-0.023 (0.008)

Lagged Availability

0.001 (0.006)

0.016 (0.007)

0.019 (0.008)

0.100 2,188,096

0.180 2,181,382

0.250 2,167,929

Dep. Var. Mean Observations

Panel B: No Init. Mort. Bal. Credit Availability

0.016 (0.006)

0.001 (0.008)

-0.014 (0.009)

Lagged Availability

-0.005 (0.007)

0.009 (0.009)

0.010 (0.010)

0.110 1,824,185

0.200 1,817,178

0.270 1,804,156

Dep. Var. Mean Observations

Panel C: Pos. Init. Mort. Bal. Credit Availability

-0.015 (0.008)

-0.036 (0.011)

-0.059 (0.013)

Lagged Availability

0.015 (0.008)

0.024 (0.012)

0.037 (0.014)

Dep. Var. Mean Observations

0.054 363,911

0.100 364,204

0.150 363,773

Note.—Logit estimates of effect of credit availability on the probability of moving to a different state, at various horizons. Average marginal effects, with standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

30

TABLE 7 Effects on Moving to Different State

Horizon in Quarters:

(1) 4

(2) 8

(3) 12

Panel A: Entire Sample Credit Availability

-0.001 (0.002)

-0.003 (0.003)

-0.005 (0.003)

Lagged Availability

0.000 (0.002)

0.001 (0.003)

0.005 (0.003)

0.017 2,878,352

0.031 2,846,654

0.044 2,817,600

Dep. Var. Mean Observations

Panel B: No Init. Mort. Bal. Credit Availability

-0.001 (0.002)

-0.004 (0.003)

-0.005 (0.004)

Lagged Availability

0.000 (0.002)

0.002 (0.003)

0.006 (0.004)

0.019 2,428,423

0.034 2,399,032

0.048 2,371,383

Dep. Var. Mean Observations

Panel C: Pos. Init. Mort. Bal. Credit Availability

-0.004 (0.003)

-0.007 (0.004)

-0.010 (0.005)

Lagged Availability

-0.002 (0.003)

-0.003 (0.004)

0.000 (0.005)

Dep. Var. Mean Observations

0.008 449,929

0.016 447,622

0.023 446,217

Note.—Logit estimates of effect of credit availability on the probability of moving to a different state, at various horizons. Average marginal effects, with standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

31

TABLE 8 Effects on Change in Number of Auto Loans

Horizon in Quarters:

(1) 4

(2) 8

(3) 12

(4) 16

Panel A: Entire Sample Credit Availability

0.044 (0.006)

0.070 (0.009)

0.100 (0.010)

0.120 (0.012)

Lagged Availability

0.019 (0.007)

0.048 (0.010)

0.065 (0.011)

0.064 (0.013)

0.000 3,072,513

0.002 2,967,912

0.014 2,889,856

0.035 2,827,019

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

0.040 (0.007)

0.064 (0.009)

0.095 (0.011)

0.110 (0.013)

Lagged Availability

0.016 (0.007)

0.044 (0.010)

0.057 (0.012)

0.059 (0.014)

0.003 2,597,016

0.009 2,496,269

0.025 2,422,101

0.050 2,362,781

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

0.078 (0.016)

0.120 (0.022)

0.160 (0.026)

0.180 (0.030)

Lagged Availability

0.045 (0.017)

0.077 (0.023)

0.100 (0.027)

0.074 (0.030)

Dep. Var. Mean Observations

-0.016 475,497

-0.033 471,643

-0.042 467,755

-0.040 464,238

Note.—Linear regression estimates of effect of credit availability on the change in the number of auto loans on an individual’s credit record, over various horizons. Standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

32

TABLE 9 Effects on Change in Auto Loan Balance

Horizon in Quarters:

(1) 4

(2) 8

(3) 12

(4) 16

Panel A: Entire Sample Credit Availability

378 (81)

474 (113)

611 (131)

542 (144)

Lagged Availability

45 (86)

56 (121)

68 (138)

74 (152)

-16 3,212,605

21 3,174,450

139 3,143,800

322 3,115,545

Dep. Var. Mean Observations

Panel B: No Initial Mortgage Balance Credit Availability

270 (77)

368 (110)

470 (127)

397 (142)

Lagged Availability

3 (82)

-29 (118)

-18 (135)

42 (151)

18 2,736,148

90 2,700,688

232 2,672,031

435 2,645,157

Dep. Var. Mean Observations

Panel C: Positive Initial Mortgage Balance Credit Availability

1271 (264)

1436 (364)

1630 (423)

1343 (467)

Lagged Availability

491 (275)

708 (377)

678 (429)

313 (472)

-209 476,457

-373 473,762

-391 471,769

-311 470,388

Dep. Var. Mean Observations

Note.—Linear regression estimates of effect of credit availability on the change in an individual’s auto loan balance, over various horizons. Standard errors clustered at quarter-riskscore level in parentheses. Models are estimated separately on the whole sample (panel A) and on samples split by whether the individual has a positive mortgage balance at t=0 (panels B and C). All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term.

33

34 0.009 530-730 3,254,299

-0.007 (0.002)

0.025 (0.002)

(2) Age Controls

0.009 530-730 3,254,299

-0.007 (0.002)

0.015 (0.003)

(3) Score Splines

0.010 530-730 2,924,284

-0.006 (0.002)

0.027 (0.002)

(4) More Lags

0.015 580-680 7,346,669

-0.006 (0.002)

0.049 (0.002)

(5) Wide Range

0.008 500-850 1,510,524

-0.008 (0.005)

0.037 (0.004)

(6) Narrow Range

Note.—Robustness checks for the logit estimate of the effect of credit availability on the probability of taking out a mortgage one quarter ahead. Average marginal effects, with standard errors clustered at quarter-riskscore level in parentheses. All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term. Column 1 (“Baseline”) is the estimate from Panel A of column 1 of table 1. Column 2 (“Age Controls”) includes linear age term interacted with quarter. Column 3 (“Score Splines”) includes flexible spline controls for riskscore (not interacted with quarter). Column 4 (“More Lags”) includes the second through fourth lags of credit availability, as well as the second through fourth lags of the riskscore interacted with quarter. Column 5 (“Wide Range”) includes all observations with current riskscores between 500 and 850. Column 6 (“Narrow Range”) includes only observations with current and lagged riskscores between 580 and 680.

0.009 530-730 3,254,299

-0.008 (0.002)

1Q Lagged Availability

Dep. Var. Mean Score Range Observations

0.025 (0.002)

(1) Baseline

Credit Availability

Specification:

TABLE 10 Robustness Checks: Effects on 1 Quarter Probability of Taking Out a Mortgage

35 0.130 530-730 3,048,165

0.007 (0.006)

0.100 (0.006)

(2) Age Controls

0.130 530-730 3,048,165

0.005 (0.006)

0.011 (0.010)

(3) Score Splines

0.140 530-730 2,747,246

0.016 (0.007)

0.110 (0.006)

(4) More Lags

0.200 580-680 6,955,343

0.120 (0.006)

0.250 (0.005)

(5) Wide Range

0.120 500-850 1,412,273

0.022 (0.022)

0.150 (0.017)

(6) Narrow Range

Note.—Robustness checks for the logit estimate of the effect of credit availability on the cumulative probability of taking out a mortgage one to 16 quarters ahead. Average marginal effects, with standard errors clustered at quarter-riskscore level in parentheses. All models include quarter fixed effects, quarter fixed effects interacted with linear riskscore term, and quarter fixed effects interacted with lagged linear riskscore term. Column 1 (“Baseline”) is the estimate from Panel A of column 5 of table 1. Column 2 (“Age Controls”) includes linear age term interacted with quarter. Column 3 (“Score Splines”) includes flexible spline controls for riskscore (not interacted with quarter). Column 4 (“More Lags”) includes the second through fourth lags of credit availability, as well as the second through fourth lags of the riskscore interacted with quarter. Column 5 (“Wide Range”) includes all observations with current riskscores between 500 and 850. Column 6 (“Narrow Range”) includes only observations with current and lagged riskscores between 580 and 680.

0.130 530-730 3,048,165

0.000 (0.007)

1Q Lagged Availability

Dep. Var. Mean Score Range Observations

0.100 (0.006)

(1) Baseline

Credit Availability

Specification:

TABLE 11 Robustness Checks: Effects on 1-16 Quarter Probability of Taking Out a Mortgage

Suggest Documents