The Dodd-Frank Act, enacted in 2010 in the wake of the Great Recession, introduced mandatory stress-testing for the largest U.S. banks. Dodd-Frank Act Stress Testing (DFAST) was intended to ensure that banks have sufficient capitalization to absorb the losses they may experience in an economic downturn and, more importantly, continue providing credit to the economy. The stress-testing exercise, which is conducted by the Federal Reserve, uses hypothetical macroeconomic scenarios to predict a bank’s portfolio return under stress and its implied equity values. Thus, the exercise indicates whether a bank, given its current equity position and portfolio allocation, could withstand a severe economic downturn and maintain the credit supply to the economy.
In a new study, we examine the impact DFAST has had on bank behavior. Specifically, we study the effects of stress-testing on banks’ portfolio allocation, including the effects on financial stability risk, credit supply, and borrowers’ investment. Our empirical analysis builds in large part on detailed loan-level data (FR Y-14) that the Federal Reserve collects from banks to perform the DFAST exercise. Our findings shed light on the broader macroeconomic consequences of bank behavior in response to the stress-testing requirements of the Dodd-Frank regulatory reform.
We show that, since the inception of DFAST, the largest U.S. banks have become more similar in terms of their overall asset allocation (i.e., portfolio shares held in cash, securities, commercial real estate loans, commercial and industrial loans, etc.). Moreover, using detailed loan-level data we find that their commercial and industrial loan portfolios have also become more similar in terms of borrower rating, industry, and geographic region. In particular, the portfolio similarity of banks subject to stress-testing has increased more than 5 percent between 2011 and 2016, on average, while the portfolio similarity of large banks not subject to stress-testing largely remained unchanged.
Consistent with a stylized model of portfolio choice under stress testing, we find that banks choose to hold relatively fewer assets that perform poorly under the stress-test scenario, while they increase their relative holding of assets that perform well in the stress test and look safer in this respect. This portfolio rebalancing leads to an increase in portfolio similarity. However, from a macroprudential point of view, we present evidence that it is indeed the case that BHC portfolios have reacted to stress-testing in ways that may inadvertently result in a build-up in systematic risk factors for the banking sector as a whole that is not captured by the single severely-adverse scenario imposed on all banks undergoing DFAST. The reason is that, as we show, bank’s portfolio rebalancing happens in response to the DFAST adverse scenarios.
Loan-level data for banks subject to DFAST show that banks with poor stress-test results tend to have dissimilar portfolios before the test. However, after receiving the poor stress-test results, these banks tend to adjust their portfolios so that they more closely resemble those of banks with good DFAST results. In our analysis, to isolate the stress-test-driven supply shifts, we perform for a number of tests to rule out that demand factors are behind the observed increase in bank portfolio similarity.
The established supply shifts in response to stress testing have consequences for borrowers. Banks with poor stress-test results reduce loans that perform poorly under the stress test, but those banks do not increase the supply of loans that perform well. This portfolio rebalancing thus can lead to an overall reduction of credit supply to the real economy relative to banks that do not experience large stress-test losses.
Borrowers with higher reliance on banks that perform poorly in stress-testing are not able to substitute for the loss in credit with borrowing from other large banks. A firm that borrows only from banks that perform poorly in the stress test faces a 14 percentage point greater decline in investment growth compared with a comparable firm that does not borrow from poorly performing banks.
While more diversified and well-capitalized banks in general result in lower bank-level risk, stress-testing has led to banks’ rebalancing their portfolios in ways that may inadvertently result in a build-up in systematic risk factors for the banking sector as a whole.
Furthermore, while our loan-level analysis confirms the desirable policy outcome of a credit supply reduction by banks in loans that add most to stress-test losses, the analysis also shows that this credit supply reduction can have real effects in the economy. Firms that borrow from banks that fare worse in the stress-tests experience a substantial decline in investment growth with adverse economic consequences.
 To isolate the effects of stress testing on bank behavior, our sample period ends in 2016 when the new Basel III capital standards were phased in.
This post comes to us from Falk Bräuning and José Fillat, senior economists at the Federal Reserve Bank of Boston. It is based on their recent article, “The Impact of Regulatory Stress Tests on Bank Lending and its Macroeconomic Consequences,” available here. The views expressed in the post do not necessarily reflect those of the Federal Reserve Bank of Boston, the Federal Open Market Committee, or the Federal Reserve System.