Increasing mortgage delinquencies and the resulting financial turmoil have put the US financial system on trial. In particular, lenders have been accused of having engaged in a reckless expansion of credit to borrowers that had no capacity to repay their debt.
Financial innovation, deregulation, and the widespread failure of the supervisory and regulatory frameworks have often been blamed as the culprits. In particular, several studies have identified the rapid expansion of securitization as an important factor behind the excessive decline in lending standards. Less attention has been paid to how changes in market structure have affected lender behavior or to the role played by regulation. This paper attempts to fill that gap.
The structure of the US mortgage market has changed dramatically over the past decade. In particular, the entry of large, nation-wide active players has altered the competitive landscape of many local markets. The paper examines how these changes affected banks’ lending behavior. The paper also investigates whether capital requirements played their traditional role in limiting risk-taking. More specifically, the main questions we seek to answer are: How did local market conditions and changes in local market structure affect banks in their decisions on whether or not to grant a loan? And in particular, how did the entry of national players affect local lenders? On regulation, did better-capitalized banks apply more stringent standards as theory predicts bank capital requirements perform their expected disciplining role in curbing excessive risk taking?
We attempt to answer these questions by merging “demand-side” information from a detailed dataset on mortgage applications (HDMA) with “supply-side data on banks’ balance sheets and income statements (Call Report files). We also employ demographic and “macro” information to control for the local economic conditions. Indeed, we conjecture that banks make lending decisions based not only on the characteristics of borrowers but also based on local economic factors and strategic interaction with their competitors. For instance, loan applications from potential borrowers with comparable characteristics could be denied in a poor location with little growth prospects, but be approved in a booming area with strong growth prospects and intense competition. Our sample spans from 2000 to 2007 (corresponding to the period of particularly fast market expansion that preceded the crisis). Our panel has three dimensions: time (yearly data), markets (Metropolitan Statistical Areas), and lender.
We focus on an admittedly limited aspect of lending standards, the denial rate of new mortgage applications. The rationale for this choice is that, controlling for other factors such as borrower characteristics, underlying market structure, and local economic conditions, an increase (decrease) over time in the probability of a lender approving (rejecting) a loan application is a sign of loosening lending standards. This, however, is also a choice of necessity. The HDMA dataset, while very comprehensive (it essentially covers the universe of mortgage applications in the U.S.), lacks some important information for a more precise evaluation of lending standards, such as interest rates, FICO scores, and loan-to-value ratios. On the plus side, by relying on this dataset, we are able to exploit information on loans that were rejected in addition to mortgages that were actually originated.
A caveat: the results in this version of the paper are to be considered incomplete and preliminary and should not be cited without permission from the authors. That said, the evidence so far suggests that competitive dynamics during the boom period led to a decline in denial rates that can be related to the recent increase in delinquencies. This is consistent with recent theories of competition under adverse selection (see next section for a discussion) linking credit expansion and lending standards to bank strategic behavior, and show that competitive pressures from peer banks played an important role. In particular, we find that denial rates tend to decline when the demand for loans is particularly strong. In previous work, we had found robust evidence of a causal relationship between the growth rate of loan applications and denial rates within a MSA. In this paper, we are able to distinguish between the effects on a bank’s denial rates of applications for its own loans and for those of its competitors. We find that strong application (and loan origination) growth at competing banks leads lenders to decrease denial rates. On the contrary, a strong demand for its own loan leads a bank to become more choosy and reject a larger fraction of applications.
We also find that bank-level variables “dominate” geographical variables variables. Most MSA-level variables are consistently significant, but there effect may be economically relatively small. For example, a lender’s nationwide denial rate appears to explain a large portion of the denial rate variability. This may explain why it is hard to find a significant effect for changes in local competitive conditions. Consistent with popular stories blaming the drop in lending standards on competitive pressures due to the entry of nationwide competitors into local markets, we find a negative coefficient on various variables proxing for entry activity. These coefficients, however, are never significant once controlling for other factors.
Finally, capital appears to have played its expected disciplining role in the sense that less capitalized lenders behaved more aggressively in their lending decisions. This finding offer evidence of moral hazard behavior on the part of banks with relatively low capital.
The preliminary nature of the analysis calls for caution when it comes to policy implications. So far, the results in the paper support the view that competitive dynamics during upswings lead banks to take more risk. In that context, the findings in the paper are supportive of proposals for reform aimed at reducing the procyclicality of the current regulatory framework. Adding a macroprudential dimension to banking regulation could be a step in this direction. For example capital requirements could be linked to aggregate credit growth, reducing the effects of the cycle on risk taking documented in this paper.
The rest of the paper is organized as follows. Section 2 summarizes the related literature on bank lending and credit booms. Sections 3 and 4 describe the data in detail. Section 5 lays out the empirical strategy. Section 6 presents the empirical results. Section 7 concludes.
