Research Exchange: June 2022

Selected Outside Research

Understanding Bank Deposit Growth during the COVID-19 Pandemic

The onset of the COVID-19 pandemic was followed by rapid and sustained growth in aggregate bank deposits in the U.S. unprecedented in recent decades.  For instance, the year-over-year percentage change in deposits since the first quarter of 2020 has exceeded any of those witnessed over the past 30 years. This note presents evidence that the growth in bank deposits over the fourth quarter of 2019 through the fourth quarter of 2021 can be attributed to four factors: (1) an initial spike in commercial and industrial credit line drawdowns at the onset of the pandemic; (2) asset purchases by the Federal Reserve; (3) large fiscal transfers to households more likely to hold savings in the form of deposits; and (4) a rising personal savings rate. The note concludes by highlighting the macroeconomic factors that will influence future deposit trends; for example, deposit growth may slow if the personal savings rate declines with households spending down their accumulated deposit savings. Read More ⇨

Liquidity in the Mortgage Market: How Does the COVID-19 Crisis Compare with the Global Financial Crisis?

While the mortgage market was subject to extreme liquidity stresses during both the Global Financial Crisis and the COVID‐19 crisis, the market suffered far more significant disruptions during the GFC than compared to the pandemic period. This paper examines the reasons why the mortgage market survived the pandemic period largely intact and assesses its future vulnerabilities. The study finds that the effects of liquidity stresses on the mortgage market in 2020 were mitigated by reforms to the mortgage system since the GFC, by the heavy government presence in the mortgage market, and by strong house prices, but that the market remains vulnerable to liquidity pressures. Read More ⇨

Monetary Policy: U.S. Loan-Level Evidence since the 1990s

The role of nonbanks (including funds, shadow banks and fintech companies) in the transmission of monetary policy is examined in this paper.  The analysis uses loan-level data that extend back to the 1990s, from three U.S. credit markets—corporate loans, consumer auto loans and mortgages—where whether the lender is a deposit-taking institution (bank) or nonbank is observed. The study finds that nonbanks reduce corporate credit supply by less than banks after a monetary contraction, attenuating the effect of the bank lending channel on total credit and implying a shift in credit supply toward ex-ante riskier firms. For auto lending, the analysis finds strong substitution from bank to nonbank credit after a monetary contraction, with the shift from bank to nonbank credit larger for riskier borrowers.  For the mortgage sector, the analysis demonstrates that total mortgage credit supply is less responsive to monetary policy in counties with higher historical dependence on nonbank lenders. Overall, the study suggests that “tighter monetary policy leads to a redistribution of risk from banks to the unregulated nonbank sector as nonbank lending increases relative to banks, particularly for riskier loans.” Read More ⇨

Federal Home Loan Banks and Financial Stability

This paper examines the potential role of Federal Home Loan Banks in amplifying financial stability risks associated with mortgage and housing markets.  The paper argues that “changes in housing finance markets have rendered largely irrelevant their original purpose of increasing the availability of mortgages” and that their activities have “at times both exacerbated risks to financial stability and obstructed the missions of federal financial regulators.”  These undesirable outcomes are attributed to their hybrid nature, such that “the private ownership and control of the FHLBs provide the incentive to take advantage of the considerable public privileges from which they benefit,” including an explicit line of credit from the United States government and an implied public guarantee of their debt.  Finally, the paper offers a framework for regulatory reform to address these undesirable outcomes. Read More ⇨

Forecasting Non-Performing Retail Loans during the COVID-19 Pandemic: the Effect of Forbearance on Model Error

The performance of mortgage and credit card credit loss models during the COVID-19 pandemic is explored in this study, thus shedding light on the role of model risk. The analysis proceeds by first constructing basic probability-of-default, loss-given-default and exposure-at-default models using pre-2019 loan-level data, and then preparing forward-looking four-quarter forecasts through 2020 from the perspective of portfolios observed in 2019Q4 using actual macroeconomic conditions during the pandemic. For both the mortgage and card portfolios, the analysis finds that actual losses were much lower than the models’ projections.  Finally, the study looks at the role of forbearance policies that permitted borrowers to suspend payments without penalty. The policies resulted in non-performance rates that were much lower than would have been the case otherwise. The analysis demonstrates that the forbearance effect was much larger for mortgages than for credit cards. Read More ⇨

What Triggers Mortgage Default? New Evidence from Linked Administrative and Survey Data

This paper examines the factors that trigger default of first-lien, closed-end mortgages during 2016 through 2018 using more detailed data than previously available to study this question. The data are from the American Survey of Mortgage Borrowers, which is designed specifically to gather information on reasons for default and is linked to a comprehensive, administrative database containing other detailed information about the borrowers and loans. The analysis indicates that shocks to borrower income including job loss trigger about two-thirds of defaults, while divorce, health shocks, changes in mortgage payments, other debt payments and other expense shocks each trigger a substantial number of defaults. The analysis also finds that half or more of mortgage defaults in 2016 through 2018 were associated with positive home equity at the time of default. Read More ⇨

Unbanked in America: A Review of the Literature

According to FDIC survey data, 5.4 percent of US households in 2019 were unbanked, meaning that no one in the household had a checking or savings account at a bank or credit union, with the likelihood of being unbanked particularly elevated for low-income and minority households.  This article examines the causes of such financial exclusion and its possible consequences by reviewing the recent literature on account fees and minimum balance requirements and other aspects of access to bank accounts in the United States. The discussion highlights that, while households consider fees and minimum balance requirements when deciding whether to have a bank account, “several studies show that when fees are constrained, banks find low-balance accounts unprofitable, so access to bank accounts can actually decline.”  The discussion emphasizes that the most effective way to reduce financial exclusion remains an open question, particularly in the context of the rapidly evolving payments landscape. Read More ⇨

Analytical Framework for Counterparty Credit Exposure

To calculate regulatory capital requirements for counterparty credit risk under the Basel III framework developed by the Basel Committee on Banking Supervision, banks must calculate exposure at default (EAD) for each netting set they have with their counterparties. Banks that have received a supervisory approval can use the internal model method for calculating EAD, while all other banks must use the standardized approach for counterparty credit risk (SA-CCR). This paper argues that introducing more risk sensitivity into the SA-CCR will allow it to better capture the risks of more complex portfolios and proposes an approach for accomplishing this based on revising some of the modeling assumptions that underlie the SA-CCR. The proposed framework “estimates the expected exposure at multiple future time points based on simple forward projections of internally calculated spot risk factor sensitivities and spot market values,” and “is sufficiently flexible to dial in any desired level of conservatism, which makes it suitable for potential regulatory applications.” Read More ⇨

Does Automated Collateral Evaluation Lower Mortgage Credit Risk Relative to Home Appraisal Valuations?

This paper examines differences in the credit risk of mortgages originated using automated valuation models relative to those originated using traditional appraisals, by modeling conditional default rates in relation to the appraisal channel. The analysis relies on data from Freddie Mac, which distinguishes loans originated through Freddie Mac’s Automated Collateral Evaluation (ACE) program from those originated with appraisals. The analysis demonstrates that ACE loans have about 9.6 percent lower default risk in comparison to otherwise similar loans originated with appraisals, and attributes this superior default performance to the AVM process employing data beyond that found in a traditional appraisal, and the more efficient utilization of available information through statistical modeling. Read More ⇨

Buy Now, Pay Later: Survey Evidence of Consumer Adoption and Attitudes

A new form of deferred payment mechanism, known as Buy Now, Pay Later (BNPL), introduced in 2019 and marketed primarily by fintech companies, allows consumers to make a purchase while splitting the transaction into four payments over six weeks, with no fees or interest for those that pay back on schedule. This paper reports the results from a survey of 2,070 U.S. consumers conducted by the Consumer Finance Institute at the Federal Reserve Bank of Philadelphia to gain insight into BNPL use. The survey includes questions about use of and satisfaction with alternative payment tools (credit cards, debit cards, person-to-person payments and BNPL); reasons for using BNPL; attitudes of BNPL users and nonusers toward the product; and incidence of missed BNPL payments. One key finding is that the use of BNPL among the survey respondents appears to be largely experimental, with most users reporting three or fewer uses over the previous 12 months. Another, somewhat unexpected finding is that lack of credit access is reportedly not a primary reason for using BNPL. A total of 10.7 percent of BNPL users reported difficulties paying, with higher frequency of missing payments among lower-income and younger users. Read More ⇨

Chart of the Month

Featured BPI Research

Potential Costs of Banning Capital Distributions by Banks in Bad Times

This blog post reports survey results and analysis concerning the potential consequences of regulatory bans or restrictions on banks’ paying dividends or repurchasing stock during economic stress. In 2020, in response to the COVID-19 pandemic, EU/UK officials implemented policies that would place bans on capital distributions by large banks regardless of capital levels. The author analyzes the potential economic results of an EU/UK-style ban and presents the results of an informal survey asking U.S. bank investors their reaction if the Fed were to move to an EU/UK-style policy. Bans on capital distributions come with at least two potential serious longer-term costs. The first potential consequence would be investors in bank equity are likely to require a higher expected return to invest in bank stocks. The higher cost of capital will reduce the supply or increase the cost of credit and reduce economic growth. The second potential consequence is that bans or restrictions can create an incentive for banks to reduce lending and increase capital distributions when a potential shock may be on the horizon. The incentives created can raise the likelihood that a feared recession will come to pass. Read More ⇨

DFAST 2022: Volatility, Capital Increases, and the Implications for the U.S. Economy

As a result of this year’s test, capital requirements of many large banks will increase later this year, and in some cases significantly. Some of the increase in capital requirements is driven by the scenario chosen combined with lower allowances for credit losses on banks’ balance sheets at the start of the scenario. The lower allowances require a larger reserve build over the stress horizon, which results in larger declines in banks’ capital ratios and thereby a higher SCB. However, the note shows that another portion of the increase in banks’ capital requirements was caused by noise in the Fed’s projections that arbitrarily inflate expenses (including operational risk losses) based on changes in bank size. The increase in requirements was particularly evident on banks that received a large influx of deposits during the pandemic. Read More ⇨