Search

This paper studies the spillover effect of home mortgage delinquencies using a discrete-choice spatial network model. In our empirical study, a main challenge in estimating this model is that mortgage repayment decisions can only be observed for a sample of all the borrowers in the study region. We show that the nested pseudolikelihood (NPL) algorithm can be readily modified to accommodate this missing data issue. Monte Carlo simulations indicate that the proposed estimator works well in finite samples and ignoring this issue leads to a downward bias in the estimated spillover effect. We estimate the model using data on single-family residential mortgage delinquencies in Clark County of Nevada in 2010, and find strong evidence of spillover effects. We also conduct some counterfactual experiments to illustrate the policy relevance of the spillover effect.

In the first portion of this paper, we utilize millions of loan-level servicing records for mortgages originated between 2004 and 2016 to study the performance of predictive models of mortgage default. We find that the logistic regression model -- the traditional workhorse for consumer credit modeling -- as well as machine learning methods can be very inaccurate when used to predict loan performance in out-of-time samples. Importantly, we find that this model failure was not unique to the early-2000s housing boom. We use the Panel Study of Income Dynamics in the second part of our paper to provide evidence that this model failure can be attributed to intertemporal heterogeneity in the relationship between variables that are frequently used to predict mortgage performance and the realized post-origination path of variables that have been shown to trigger mortgage default. Our findings imply that model instability is a significant source of risk for lenders, such as financial technology firms ("Fintechs"), that rely heavily on predictive statistical models and machine learning algorithms for underwriting and account management.

We study the distribution of savings from mortgage refinancing across income groups during the COVID-19 pandemic. Between February and June 2020, the difference in savings from refinancing between high- and low-income borrowers was ten times higher than before the pandemic. This was the result of two factors: individuals in the top quintile of the income distribution increased their refinancing activity more than comparable borrowers in the bottom quintile and, conditional on refinancing, they captured slightly larger improvements in interest rates. Exploiting idiosyncratic variation in COVID-19 case rates within zip codes over time, we find that changes in local economic conditions explain up to 74 percent of the increase in refinancing inequality tied to the pandemic. Using data on refinancing applications and funding rates we find that, conditional on applying, the funding rates and processing times for low-income borrowers were not differentially affected by the pandemic. Instead, low-income borrowers were underrepresented in the pool of applications. We estimate a difference of $5 billion in savings from refinancing between the top quintile of the income distribution and the rest of the market. This discrepancy has implications for the transmission of monetary policy and the evolution of wealth inequality.

This paper investigates spillover effects of mortgage defaults in the neighborhood on a homeowner's default decision. Following the interactions-based model of discrete choices in Lee et al. (2014), we explicitly model a homeowner's default decision as a function of predetermined risk factors as well as rational expectations on her neighbors' default decisions and find strong empirical evidence of spillover effects — in forms of time-lagged “contagion effects” and contemporaneous “multiplier effects”. Furthermore, the estimated model can be used to identify the “power neighbor” through whom a foreclosure prevention policy can generate the largest impact on a neighborhood. Compared to other homeowners, the “power neighbor” on average has less neighbors that defaulted in the past, a less risky loan, a smaller payment size, a higher credit score, and a more central location in the neighborhood.

In this study we use a large database of real estate transactions to assess the magnitude of measurement error associated with using popular house price indices (HPIs) to value individual properties. In the 4 large U.S. counties that we analyze, we find that the bias associated with using these HPIs to value individual homes increased from near zero in 2005 to between 26% and 113% in 2010. In the second part of the analysis, we use data from Florida to demonstrate that forecast combination methods can be used to improve the accuracy of property-level valuations, in some cases reducing the estimated bias by more than a factor of 3. We find that even the simplest forecast combination method – a simple average – has the potential to significantly improve value estimates.

This paper identifies instruments for house prices and foreclosure rates and estimates a Dynamic Spatial Simultaneous Equation System (DSSES) to investigate the dynamics of them across space and time. Shocks to the foreclosure rate in one state not only affect house prices in that state but also the foreclosure rates and house prices in nearby states. When it comes to the housing market, what happens in Vegas doesn’t always stay in Vegas. A one standard deviation foreclosure shock leads to a 2 percent decline in real house prices over the long run.

Although a broad-based increase in house prices has been observed over the past year, not everyone is convinced the rise of house prices will persist and lead to a steady recovery of the economy. The main reason for this skepticism is uncertainty about the "shadow inventory:" foreclosed homes held by investors or as REOs, which have not yet hit the market but likely will as market prices rise. The volume of shadow inventory itself in local markets is largely unknown, as is its impact on the housing market. This study quantifies the size of the shadow inventory and investigates the spatial impact of the of the outflow of shadow inventory. The scope of our study is a set of housing markets that vary in both their historic housing price volatility as well as institutional factors -- such as foreclosure law statutes -- that may influence the relationship between the shadow inventory and house price dynamics. To address the endogeneity that characterizes the spatial interaction of house prices and the outflow of the shadow inventory, we utilize a simultaneous equation system of spatial autoregressions (SESSAR). The model is estimated using measures of the shadow inventory derived from DataQuick's national transaction history database and county-level house price indices provided by Lender Processing Services. Lastly, because our estimate -- as well as all other existing estimates -- of the shadow inventory relies upon string matching algorithms to identify entry into and exit out of REO status, we validate the accuracy of our measures of REOs using loss mitigation data from the OCC Mortgage Metrics database.

In this study we use property-level transaction data to construct several different out-of-sample forecasts for the sales of individual homes in 10 counties in Florida. We utilize a number of common forecast combination methods and analyze the properties of their forecast errors. The results from this analysis suggest that relatively simple forecast combination schemes have the potential to reduce the error in property value estimates, possibly by a wide margin. Lastly, we find that the forecast error associated with constructing valuations using popular house price indices has been very large in our sample.

This paper examines the importance of social interactions on a household's location decision. The theory argues that individuals' utility will be greater when socially interacting with similar others. The hypothesis that a household desires to find a good community match is tested through the application of a discrete residential location choice model. In addition, this paper also tests Tiebout's hypothesis that households search for a community where their benefits from local public goods will exceed their local tax costs. The findings tend to support both hypotheses, indicating that a household prefers neighbors with a similar socio-economic background and somewhat larger houses.

We propose a model selection methodology for credit default modeling in the presence of a large number of variables and candidate models. Accurate credit default models are critical to financial institutions for making effective underwriting and pricing decisions in terms of profit maximization and loss mitigation. Credit default modeling routinely involves large data sets and considers an extremely large set of candidate models. This leads to deriving statistical inference under a multiple hypothesis-testing scheme. An unguarded use of single-inference procedures or the recently popular data snooping techniques such as variable reduction via decision tree analysis and stepwise procedure leave a modeler at risk of making numerous false statistical discoveries, that is pure chance makes the likelihood of a type I error extremely high in data rich environments. To mitigate these concerns we control for the false discovery rate in our model selection procedure and make inference when p-values are dependent. A Monte Carlo study shows that in large data sets with high co-linearity between observations, a naive data snooping approach leads to multiple false discoveries, and a reduction in prediction accuracy. An empirical application of this proposed methodology uses the Office of the Comptroller of the Currency Consumer Credit Database, which is a large random sample of individual and tradeline data from one of the three national credit bureaus between 1999 and 2009.