Did TARP Reduce Or Increase Systemic Risk? The Effects Of TARP On Financial System Stability
University of South Carolina – Darla Moore School of Business; Wharton Financial Institutions Center; European Banking Center
Raluca A. Roman
Federal Reserve Bank of Kansas City
John Sedunov III
Villanova University – Department of Finance
December 28, 2015
Abstract:
We investigate the effect of bailing out U.S. banks through the Troubled Assets Relief Program (TARP) on systemic risk during the recent financial crisis. We derive and test hypotheses implying the effects could go either way. Our difference-in-difference analysis suggests that TARP statistically and economically significantly reduced systemic risk, primarily through risk reductions of institutions that were safer and/or larger ex ante. The primary conduit for this effect is decreases in leverage. Results are robust to additional tests, including accounting for endogeneity, and are consistent across multiple systemic risk measures. These results contribute to the cost-benefit debate on TARP.
Did TARP Reduce Or Increase Systemic Risk? The Effects Of TARP On Financial System Stability
Did bailing out banks during the recent financial crisis reduce systemic risk, possibly rescuing the financial system from collapse, or did it encourage banks to take on more risk, increasing the possibility of future systemic failure? Reducing systemic risk and preventing financial collapse which could have dragged down the U.S. economy was a major objective of the U.S. bank bailout known as the Troubled Assets Relief Program (TARP). Some argue that it was successful in this regard (e.g., Krugman; 2009; Bernanke, 2009, 2015), while others argue that TARP caused banks to take on more risk due to increased moral hazard incentives and other channels, making the system riskier (e.g., Rogoff, 2010; Barofsky 2011, 2012). In this paper, we formulate and test hypotheses that represent both of these points of view to help determine which one is more consistent with the evidence.
The determination of whether TARP increased or decreased systemic risk is of first order importance in assessing the benefits and costs of the program and the advisability of future bailouts.1 Examining this question is challenging for several reasons. First, to our knowledge, there is no consensus on how to measure of the risk of the financial system, despite a number of studies that compare measures or run “horse races” among them (e.g., Bisias, Flood, Lo, and Valavanis, 2012; Benoit, Colletaz, Hurlin, and Perignon, 2013; Sedunov, 2015; Zhang, Vallascas, Keasey, and Cai, forthcoming). Second, it is hard to observe what the condition of the financial system would have been in the absence of TARP. Third, it is difficult to disentangle the effects of TARP from those of other government programs and market events that occurred around the same time.
However, some progress can be made toward answering this question by investigating whether the individual banks that received the bailouts experienced increases or decreases in their contributions to systemic risk measured several different ways relative to other banks that did not receive TARP funds. We do so controlling for other differences between the banks and the effects of some of the other government programs and market conditions that differed across banks and over time. If it is found that individual TARP banks experienced decreases in their contributions to systemic risk relative to other banks across several different measurement methods, controlling for the effects of other government programs and market conditions, then it is likely that TARP decreased the risk of the system, and vice versa if it is found that these banks experienced consistent increases in their contributions to systemic risk relative to other banks.
Existing research on whether TARP decreased or increased bank risk is limited. Some studies investigate the effects of TARP on the risk of banks’ loans (e.g., Black and Hazelwood, 2013; Li, 2013; Duchin and Sosyura, 2014; and Berger, Makaew, and Roman, 2015) or on whether these banks decreased or increased their lending (e.g., Black and Hazelwood, 2013; Li, 2013; Puddu and Walchli, 2013; Duchin and Sosyura, 2014; and Berger and Roman, 2015). The results of these studies are far from conclusive. Moreover, these studies assess only the loan portfolio risk of TARP banks, rather than their contributions to systemic risk, which also include the rest of their portfolio, their capital, their size, and their connections with the rest of the financial system.
In this paper, we formulate and test hypotheses about the effects of TARP on systemic risk, rather than loan portfolio risk by taking advantage of relatively recently developed measures of individual banks’ contributions to systemic risk. We focus on two measures, the normalized conditional capital shortfall measure of systemic risk, SRISK% and systemic expected shortfall (SES), which are described in more detail in Section 6.2. Our results are also robust to other measures, as discussed below.
Our tests apply difference-in-difference (DID) regression models to the full sample of commercial banks in the U.S. over 2005:Q1-2012:Q4, using SRISK% and SES as the key dependent variables. The key exogenous variables are TARP Bank (a dummy equal to one if a bank was approved for and received TARP support), Post TARP (a dummy equal to one over 2009:Q1-2012:Q4, the period after the TARP program initiation) and a DID term Post TARP x TARP Bank to capture the effect of the TARP treatment. We also control for a large number of bank-related characteristics, including proxies for bank CAMELS examination ratings; three other government intervention programs, the Discount Window, Term Auction Facility (TAF), and stress tests (SCAP/CCAR), and time fixed effects. We also present all analyses in several specifications: models using the full set of controls. models that exclude the proxies for bank CAMELS to mitigate the possibility that TARP affects contribution to systemic risk through affecting the health of the recipient banks through these variables; models which exclude other bank controls to mitigate the possibility that TARP affects contribution to systemic risk through affecting other characteristics of the recipient banks; and models which exclude all controls.
See full PDF below.
