Financial Regulation Policy Uncertainty and Bank Profits

Transcription

1 Financial Regulation Policy Uncertainty and Bank Profits Robert N. Killins Seneca College of Applied Arts and Technology School of Accounting and Financial Service 1750 Finch Avenue E. Toronto, Ontario, M2J 2X5, Canada Tel: and fax: Peter V. Egly Department of Economics and Finance University of Texas - Rio Grande Valley One West University Blvd Brownsville, TX 78526, USA peter.egly@utrgv.edu Tel: and fax: David W. Johnk* School of Professional Studies Department of Business Rogers State University 1701 W. Will Rogers Blvd Claremore, OK , USA djohnk@rsu.edu Tel: and fax: Abstract: Using quarterly call report data from 2000 through 2016, we explore the relationship between bank profits (i.e. return on assets-roa) and a news-based index of financial regulation policy uncertainty developed by Baker et al. (2016) for more than 4,760 U.S. commercial banks. In the full sample, financial regulation uncertainty has a positive impact on ROA which varies with bank size; the effect is negative only for small banks. Splitting the sample around the 2009Q1 change in presidential partisan cycle, financial regulation policy uncertainty has a negative (positive) impact on ROA during the republican (democratic) presidency. Keywords: Bank profitability, financial regulation policy uncertainty, panel data JEL Classification Numbers: E44, G21. *Corresponding Author: Tel ; fax ; address: djohnk@rsu.edu 1

2 1. Introduction Over the last few decades, the financial service industry (including commercial banks) has undergone major changes in function and form. Both internal and external factors have affected the industry s structure and performance. A major trend affecting financial firm performance is the shift from deregulation to re-regulation. The rising competition and proliferation of financial services spurred deregulation. In the U.S., deregulation began with the lifting of government-imposed interest rate ceilings on savings deposits (in favor of free-market interest rates) to provide the public a reasonable return on their savings. Subsequent regulation allowed banks to create securities and insurance subsidiaries and engage in other complementary activities. The speculation in real estate and mortgage markets, uncontrolled innovation in financial derivative contracts, inadequate regulator insight on financial institution sector transformation, and increasing commercial and investment bank failures (which culminated with the financial crisis) led to reregulation. The swings in the financial regulatory environment create important challenges to financial services firms when planning for future business growth and scope of activities, investment and lending standards, deposit base, and capital requirements. There are, of course, multiple factors that influence bank profit behavior such as business cycles, macro-economic indicators, institutional guidelines as to lending and competition in the industry, and bank-specific characteristics (e.g. Albertazzi and Gambacorta (2009), Athanasoglou et al. (2008), Aliaga-Diaz and Olivero (2011) and Bolt et al. (2012)). Financial service institutions are among the most heavily regulated of all industries due to their key roles in attracting and protecting public s savings, providing credit to borrowers, and creating money as the principal medium of exchange in our economy. Therefore, it is conceivable that general 2

3 economic policy and financial regulation uncertainty influence financial service firms current and future operating environments. Examining the impact of economic forces on bank profitability is particularly interesting given the significant transformation in financial intermediation, the gradual shift in sources of bank revenue (i.e. interest income vs. fee income), and swings in the financial regulatory environment over the time frame of our study which covers periods of economic expansion and contraction and changes in in presidential partisan cycle. The empirical literature on the impact of policy uncertainty on the economy has explored different paths focusing on monetary policy effects (e.g. Friedman, M. (1968) and Aastveit et al. (2017)), fiscal policy decisions (e.g. Hassett & Metcalf (1999) and Fernandez-Villaverde (2015)), regulatory policy action (e.g. Higgs (1997)), and private investment decisions (e.g. Bernanke (1983), Rodrik (1991) and Gilchrist et al. (2014)). The broad findings of the reviewed literature are that higher levels of policy uncertainty lead to negative responses in the overall economic system. Pastor and Veronesi (2012) examine the theoretical links between economic fluctuations, policy uncertainty, and stock market volatility and suggest a negative relationship between stock prices and policy uncertainty. Another stream of research investigates the effect of changes in financial regulation on security prices (e.g. Schafer et al. (2016), security prices and bank CDS spreads (e.g. Hachenberg et al. (2017), and on mortgage lending (Gissler et al. (2016)). Other studies suggest regulatory uncertainty can have different impacts on investment expenditures (e.g. Julio and Yook (2012)) and equity returns (Santa-Clara & Valkanov (2003) and Belo et al (2013)) based on the political party in power. The studies on key financial regulation changes mostly use event study methodology. Although there are advantages to using a discrete event, it may not capture the expectations or 3

4 anticipation of a change. The referenced regulation studies that control for political party in power exclude (or don not identify) financial services firms in their sample. We apply panel data methods to a very large number of U.S. commercial banks following a four-size classification scheme based on asset size that yields 4,336 small banks, 389 medium banks, 26 large banks, and 12 moneycenter banks to examine sensitivity of bank profitability to fluctuations in financial regulation policy uncertainty (FRPU) during presidential partisan cycles. Controlling for bank balance composition (namely deposits-to-loans) and macro forces, we find that over the full sample period, with the exception of small banks, financial regulation uncertainty has a positive impact on Return on Assets (a measure of bank profitability) which varies with bank size. Splitting the sample around the 2009Q1 change in presidential partisan cycle, the effect of FRPU on return on assets (ROA) is negative during republican presidency and turns positive during democratic presidency. The paper is structured as follows: Section 2 describes the data used and introduces our empirical model. Section 3 discusses the results and Section 4 summarizes our conclusion and offers implications. 2. Data and Methodology The sample draws from the population of commercial banks that are insured through the Federal Deposit Insurance Corporation (FDIC) over the time period 2000Q1 to 2016Q4. 1 To minimize the impact of outliers, we follow Loutskina (2011) and eliminate all bank-quarter data with asset growth over the preceding quarter in excess of 50%, total loan growth exceeding 100%, and total loans-to-assets ratio of less than 10%. This sample selection process left us with a total of 4,763 banks and 291,847 bank quarter observations. Following Verma and Jackson 1 The bank data used for this study are available through Federal Deposit Insurance Corporation statistics on depository institutions (SDI) database at the following website: (last accessed on 11/15/17) 4

5 (2008), we further divide our bank sample into four groups based on average total asset size as of beginning of period as follows: small banks (average total assets < U.S. $1 billion), medium banks (average total assets U.S. $1 billion and < U.S. $20 billion), large banks (average total assets U.S. $20 billion and < U.S. $90 billion) and money- center banks (average total assets U.S. $90 billion). The bank information extracted for this study includes: 1) return on assets (ROA) computed as net income after taxes and extraordinary items (annualized) divided by average total assets, 2) total assets, 3) total loans, and 4) FDIC insured domestic deposits less than $100, To control for the macroeconomic environment on bank profitability, we include the yield curve, real GDP growth rate, and inflation rate. The selection of our bank-level and external variables is supported by related papers including Bolt et al. (2012), who analyze the asymmetric effects of the business cycle on bank profitability for 17 OECD countries; and Egly et al. (2018), who reexamine the relationship between net interest margins and the yield curve for more than 5,500 U.S. commercial banks. A news text-based index of FRPU developed by Baker et al. (2016) is used to study the impact on bank profitability 3. We use panel estimation techniques to model the impact of FRPU on bank profits, proxied by (ROA), similar to the empirical specification presented by Bolt et al. (2012) and modified by Egly et al. (2018) that incorporate bank-specific determinants (deposits-to-loans ratio) and macro-variables. This modeling approach recognizes that net interest income is a 2 The deposit threshold for FDIC reporting purposes increased to $250,000 effective September 2009.Accordingly, SDI reporting of insured deposits was also revised (SDI variable code name depsmamt replaced with iddepsam ). 3 Policy uncertainty indices for the U.S. are in monthly frequency and rely on 10 leading newspapers including, among others: USA Today, Chicago Tribune, New York Times, Boston Globe, and the Wall Street Journal. Indices are primarily text based and articles must include key terms pertaining to uncertainty, the economy, and policy. Policy categories indices (including the financial regulation index) use expanded search criteria and include the presence of one or more category-relevant terms. Refer to Baker et al. (2016) for details. 5

6 substantial component of bank profit that is impacted by the full history of outstanding loan balances and deposits and their respective interest rates. A priori, an inverse relationship between deposits-to-loans ratios and ROA is consistent with the view that purchases of large volumes of liquid, readily-marketable securities tend to lower the average yield from a bank s earning assets and reduce its profitability and ROA. Since banks loans and asset security holdings tend to have longer maturities than the liability sources that support these assets, it is reasonable to anticipate a positive relationship between an upward-sloping yield curve and net interest margins, and between an upward-sloping yield curve and ROA. Beforehand, we expect a direct relationship between real GDP and ROA in line with the banking literature. The direction of the relationship between inflation and ROA arguably depends on bank management s ability to fully anticipate (and plan for) inflation expectations suggesting that banks can adjust their pricing in order to increase revenues to offset rising costs. We posit that increased FRPU will curb banks medium-to-long-term investment and loan generation activities based on their idiosyncratic risk aversion preferences, and increase implementation and monitoring costs which dampen profitability and ROA. The model is expressed as follows: y i,t = 4 τ=1 β 1,t τ bk i,t τ + 4 τ=1 β 2,t τ macro t τ + 4 τ=1 β 3,t τ fin reg t τ + 4 τ=1 β 4,t τ pol party t τ + α i + ε i,t (1) where the prefix represents the change in the profitability variable y consisting of the ROA; bk denotes our deposits-to-loans variable expressed in ratio form; macro consists of a vector of macro-economic variables, which includes: 1) real GDP growth rate, 2) inflation rate, and 3) yield curve that enters the model in first differences. 4 The fin reg represents our FRPU index 4 Dickey-Fuller (DF) and Augmented Dickey-Fuller (ADF) unit root tests are applied to all macro variables. The null of these tests is that the series contain unit roots (i.e. non-stationary series). The RGDP growth rate and inflation rate variable series were stationary in their original form; it is noted that the yield curve (yc) variable is non- 6

7 variable; pol party is a dummy variable that enters the model as an independent regressor and takes on a value of zero from 2000Q1 through 2008Q4 (the Republican Presidential cycle) and one from 2009Q1 through 2016Q4 which corresponds to the Democratic Presidential cycle. Finally, α i represents the bank-specific fixed effects while ε i,t captures the remaining disturbance term. 4. Results Table 1 reports the panel regression results applied to the small, medium, large, and money center bank samples and for all banks combined for the full sample period. We focus our discussion on the impact of FRPU on bank profitability measured through ROA since it is the key variable of interest in our study. All other variables incorporated in our profitability model control for bank-specific and external factors influencing bank profit behavior as supported by our review of the literature. For all banks combined, FRPU has a positive statistically significant, albeit small, net impact on bank profitability of across quarters based on negative (positive) significant first (second) lag coefficients. Therefore, the change in bank ROA grows by 0.071% given a 1% increase in the FRPU index. Evaluating this coefficient at the mean of (translates to 1.2% in percent terms) of the FRPU index, the overall impact on the change in ROA is or 0.085%. 5 The positive effect appears at odds with our hypothesis and supports the view that strong capital buffers and favorable market based data (e.g. rising bank stock prices, decreasing interest costs on senior debt instruments, and declining cost of stationary in levels yet it becomes stationary in first differences. Unit root tests are not reported in this study but are available upon request. 5 Due to space limitations, variable descriptive (and contemporaneous bivariate correlation) information is not reported in table form for each bank size class but is available from the authors upon request. To improve visual impact of regression coefficients, we chose to scale the original financial regulation uncertainty index by 10,000. Changing the units of measurement of this independent variable does not alter the interpretation of the model or goodness-of- fit. 7

8 equity capital), usually seen in periods of strong bank profitability, signal bank financial strength that temper financial regulation policy uncertainty. The impact of financial regulation uncertainty on ROA (whether positive or negative) may depend on expectations regarding the effect of proposed regulation on bank operations, competitors, and performance. The overall impact of FRPU on ROA is positive for the medium, large, and money center banks (with large and money center banks reflecting coefficients of greater magnitude) that turns negative at (first lag is negative and second lag is positive) in the small bank sample. In the small bank sample case, it suggests that a 1% increase in the financial regulation policy index negatively impacts the change in ROA by % which evaluated at the index s mean of implies an overall negative impact of or % on the change in ROA. The results support our view that banks may respond differently to FRPU given their size. For example, smaller banks may not have influential political interest groups than provide them with valuable lobbying power to diminish the unfavorable effects of financial regulation changes. Depending on the severity of the regulatory changes, smaller banks may not be able to efficiently absorb implementation and monitoring costs, thus resulting in bank industry consolidations and shifts in deposit and loan market share towards larger banks. Conversely, large banks are more likely able to benefit from information and political influence thus resulting in minimal (or even positive) impacts of regulatory changes on bank performance. Finally, the political party dummy variable is statistically significant across all samples (with a negative impact only in the large bank sample), suggesting that the political party cycle influences the profitability (ROA) of U.S. banks complementing the findings of Santa-Clara and Valkanov (2003) who document the "presidential puzzle" on equity excess returns 6. The overall positive coefficients range from (across 6 Santa-Clara and Valkanov (2003) find that excess returns in the stock market are higher during Democratic versus Republican presidencies (i.e. 9% for their value-weighted portfolio and 16% for their equally-weighted portfolio). 8

9 quarters) for the small bank sample to (across quarters) for the money center bank sample. The coefficients which measure the average difference in ROA between the republican and democratic presidential cycle, suggest that ROA is roughly higher by 1 to 19 basis points during the democratic cycle. The two subsamples provide an interesting complement to the main hypothesis of our study. According to Table 2 for our republican presidential cycle, panel regression results show that the FRPU variable has an overall negative effect of across quarters on the change in ROA for all banks combined. Assessing this coefficient at the index s mean of 0.011, the overall impact is or -3.72% which contrasts with our positive, and small, effect recorded in the full sample period of or 0.085%. Further, the magnitude of the coefficients across quarters for small and medium size banks ( and , respectively) suggest that smaller banks are more unfavorably impacted by FRPU compared to their larger counterparts. While our results imply that larger banks are less impacted by financial regulation policy uncertainty, Hachenberg et al. (2017) find that Globally Systemically Important Banks (G-SIB) reacted more positively to the 2016 U.S. republican presidential election than non-g-sibs evidenced by larger gains in bank stock prices and fairly stable CDS spreads. They surmise that the market s reaction appear to anticipate a lowering of financial sector regulation primarily with respect to the Dodd- Frank Act. The effect of our FRPU variable on the change in ROA reported in Table 3 for our democratic presidential cycle subsample, resembles what is found in our full sample period with the exception of the small bank sample in which a negative (positive) effect is found in the full They find that the difference in returns is not explained by business-cycle variables or concentrated around election dates. Their results show that the riskiness of the stock market remains consistent across presidencies. They conclude that the difference in returns through the political cycle is a puzzle. 9

10 (democratic) sample. The overall impact is or 0.25% for all banks combined (sum of across quarters evaluated at the index s mean of 0.014). The positive impact reflected in Table 3 across bank sizes (only the large bank is insignificant) coincides with a time period in which the U.S. and the world are coming out of the great recession and where bank cost of funds have been at historically low levels thereby boosting profits and ROA. To complement the regression results, Figure 1 depicts ROA patterns for each bank size class (small, medium, large, and money center) contrasted against the FRPU index. In Figure 1, ROA across banks reflect their greatest decline during the financial crisis with notable upside shown in the post-crisis period. While ROA has not fully reverted to the pre-crisis period, the small and large banks (medium and money center banks) reflect the stronger (weaker) recovery. As expected, there is an inverse relationship between ROA and our FRPU index that appears stronger for the large and money center bank samples compared to the small and medium bank samples over the entire sample period. We conducted several robustness checks. First, we run our bank profitability model replacing ROA with ROE as our dependent variable. Second, we replace our main explanatory variable of interest (FRPU index) with a principal components variable that summarizes economic, monetary, and FRPU indices developed by Baker et al. (2016). Third, we replace our FRPU index with a Chicago Board Options Exchange volatility (VIX) index which captures the stock market s expectation of volatility implied by the S&P 500 index options. Our central findings remain qualitatively unchanged. 7 7 Due to space constraints, regression result details for the robustness checks are not reported in this study but are available upon request. 10

11 5. Conclusion Our results suggest that financial regulation policy uncertainty impacts bank profits proxied through ROA. The impact varies depending on bank size and presidential political cycle. Our findings lead to several important implications for financial services regulators, investors, and executives. First, financial services firms profit performance is sensitive to uncertainty surrounding the policies and laws that they must operate within. Financial regulators and government officials must convey clear and timely communication regarding proposed changes in legislation to financial services firms in a way similar to the forward guidance delivered to Federal Reserve stakeholders surrounding monetary policy changes. Arguably, this may be more important when presidential office power is held by the Republican party (given weaker ROAs seen during the cycle). Moreover, during these political periods, it is essential to ensure that communication channels are open especially to small- and medium-sized banks for proper strategic planning given their greater sensitivity to regulatory uncertainty. Second, election cycles matter not only for the general economic environment, as documented by Santa-Clara and Valkanov (2003), but also for the banking sector as suggested by this study. Our findings suggest that ceterius paribus, investors should examine strategies that rebalance their financial services portfolio to hold large (small-to-medium) financial services firms during Republican (Democratic) held offices periods. This strategy implies that bank profitability is an important driver of stock prices and is influenced by political cycles. Finally, financial executives need to be aware of the presidential cycle and potentially adjust their balance sheets accordingly to maintain the required liquidity, capitalization, and profitability that shareholders require. A possible extension of this research may include compiling 11

12 firm-specific political donations and tracking/measuring lobbying power to further understand how financial services firm s political connections and lobbying donations influence profitability, regulation uncertainty, and overall performance. Furthermore, regulators must be cognizant (and respond to) the evolution in regulatory complexity that fosters the too big to fail issue, since it protects the largest banks that have access to powerful lobbyists and lawyers from competition from smaller banks. 12

13 References Aastveit, K. A., Natvik, G. J., & Sola, S. (2017). Economic uncertainty and the influence of monetary policy. Journal of International Money and Finance, 76, Albertazzi, U., & Gambacorta, L. (2009). Bank Profitability and the Business Cycle. Journal of Financial Stability, 5, Aliaga-Diaz, R., & Olivero, M. (2011). The Cyclicality of Price- Cost Margins in Banking: An Empirical Analysis of its Determinants. Economic Inquiry, 49, Athanasoglou, P., Brissimis, S., & Delis, M. (2008). Bank-Specific, Industry-Specific and Macroeconomic Determinants of Bank Profitability. Journal of International Financial Markets, Institutions & Money, 18, Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring Economic Policy Uncertainty. The Quarterly Journal of Economics, 131(4), Belo, F., Gala, V. D., & Li, J. (2013). Government Spending, Political Cycles, and the Cross Section of Stock Returns. Journal of Financial Economics, 107(2), Bernanke, B. S. (1983). Irreversibility, Uncertainty, and Cyclical Investment. The Quarterly Journal of Economics, 98(1), Bolt,W., De Haan, L., Hoeberichts, M., Van Oordt, M., and Swank, J., (2012). Bank Profitability During Recesssion. Journal of Banking and Finance, 36, Egly, P. V., Johnk, D. W., & Mollick, A. V. (2018). Bank Net Interest Margins, the Yield Curve, and the Financial Crisis. Review of Financial Economics, 36 (1), Fernández-Villaverde, J., Guerrón-Quintana, P., Kuester, K., & Rubio-Ramírez, J. (2015). Fiscal Volatility Shocks and Economic Activity. The American Economic Review, 105(11),

14 Friedman, M. (1968). The Role of Monetary Policy. The American Economic Review, 58(1), Gilchrist, S., Sim, J. W., & Zakrajšek, E. (2014). Uncertainty, financial frictions, and investment dynamics. National Bureau of Economic Research. (No. w20038). Gissler, S., Oldfather, J., & Ruffino, D. (2016). Lending on Hold: Regulatory Uncertainty and Bank Lending Standards. Journal of Monetary Economics, 81, Hachenberg, B., Kiesel, F., Kolaric, S., & Schiereck, D. (2017). The Impact of Expected Regulatory Changes: The Case of Banks Following the 2016 US Election. Finance Research Letters, 22, Hassett, K. A., & Metcalf, G. E. (1999). Investment with Uncertain Tax Policy: Does Random Tax Policy Discourage Investment. The Economic Journal, 109(457), Higgs, R. (1997). Regime Uncertainty: Why the Great Depression Lasted so Long and why Prosperity Resumed After the War. The Independent Review, 1(4), Julio, B., & Yook, Y. (2012). Political Uncertainty and Corporate Investment Cycles. The Journal of Finance, 67(1), Loutskina, E. (2011). The Role of Securitization in Bank Liquidity and Funding Management. Journal of Financial Economics, 100, Pastor, L., & Veronesi, P. (2012). Uncertainty about Government Policy and Stock Prices. The Journal of Finance, 67(4), Rodrik, D. (1991). Policy Uncertainty and Private Investment in Developing Countries. Journal of Development Economics, 36(2), Santa-Clara, P., & Valkanov, R. (2003). The Presidential Puzzle: Political Cycles and the Stock Market. The Journal of Finance, 58(5),

15 Schäfer, A., Schnabel, I., & Weder di Mauro, B. (2016). Financial Sector Reform After the Subprime Crisis: Has Anything Happened? Review of Finance, 20(1), Verma, P., & Jackson, D. (2008). Interest Rate and Bank Stock Return Asymmetry: Evidence from U.S. Banks. Journal of Economics and Finance, 32,

16 2000Q1 2000Q3 2001Q1 2001Q3 2002Q1 2002Q3 2003Q1 2003Q3 2004Q1 2004Q3 2005Q1 2005Q3 2006Q1 2006Q3 2007Q1 2007Q3 2008Q1 2008Q3 2009Q1 2009Q3 2010Q1 2010Q3 2011Q1 2011Q3 2012Q1 2012Q3 2013Q1 2013Q3 2014Q1 2014Q3 2015Q1 2015Q3 2016Q1 2016Q3 Average ROA by Bank Size Financial Regulation Policy Uncertainty Figure 1 Return on Assets by Bank Size with Financial Regulation Policy Uncertainty US Recession Democratic from here through 2016Q4 Large Medium Money Center Small Financial Regulation Policy Uncertainty Year/Quarter 0 Notes: ROA is separated by bank size and averaged by quarter. US recessionary periods are quarterly frequency (USRECQ) per NBER downloaded from (data retrieved on 12/19/2014). Financial Regulation Policy Uncertainty data downloaded from (data retrieved on 10/04/17). 16

17 Table 1 Panel Models by Bank Size. Dependent Variable: Return on Assets Small Banks Medium Banks Large Banks Money Center Banks All Banks Real GDP growth t *** / (0.301) 3.728*** / (1.165) 8.684* / (4.476) / (3.161) 1.481*** / (0.297) Real GDP growth t ** / (0.266) / (1.076) / (4.122) / (2.909) 0.728*** / (0.257) Real GDP growth t *** / (0.200) 4.057*** / (1.055) / (4.042) 6.534** / (2.836) 3.091*** / (0.195) Real GDP growth t *** / (0.289) / (1.211) / (4.697) / (3.251) *** / (0.280) Yield curve t *** / (0.002) 0.023* / (0.013) / (0.052) / (0.036) 0.016*** / (0.002) Yield curve t *** / (0.004) *** / (0.014) / (0.053) / (0.037) *** / (0.003) Yield curve t / (0.004) / (0.015) / (0.058) / (0.040) / (0.004) FRYield curve t *** / (0.003) / (0.014) / (0.055) / (0.038) 0.024*** / (0.003) Inflation t *** / (1.442) / (3.768) / (9.415) / (11.587) *** / (1.486) Inflation t ** / (0.971) / (3.888) / (11.692) / (11.880) 1.591* / (0.938) Inflation t *** / (0.920) / (4.009) / (12.224) *** / (12.179) *** / (0.925) Inflation t *** / (0.513) ** / (4.169) / (11.045) / (12.690) 3.126*** / (0.511) Deposits t-1/loans t-1*inflation t / (1.484) / (5.168) / (10.958) / (25.421) / (1.540) Deposits t-2/loans t-2*inflation t / (0.958) / (5.227) / (16.579) / (25.413) / (0.938) Deposits t-3/loans t-3*inflation t * / (0.872) / (5.245) / (17.184) *** / (25.439) / (0.913) Deposits t-4/loans t-4*inflation t * / (0.220) ** / (5.247) / (13.273) / (25.798) 0.418* / (0.231) Deposits t-1/loans t / (0.016) 0.108** / (0.048) / (0.021) 1.100** / (0.456) 0.027*** / (0.010) Deposits t-2/loans t *** / (0.017) / (0.060) / (0.026) ** / (0.621) 0.030* / (0.016) Deposits t-3/loans t *** / (0.002) *** / (0.067) / (0.025) 1.077* / (0.615) ** / (0.002) Deposits t-4/loans t ** / (0.001) / (0.067) / (0.017) / (0.456) * / (0.001) Financial Regulation Uncertainty t *** / (0.173) ** / (0.864) / (3.326) ** / (2.357) *** / (0.170) Financial Regulation Uncertainty t *** / (0.298) 4.393*** / (1.031) / (3.979) 6.956** / (2.796) 1.500*** / (0.291) Financial Regulation Uncertainty t / (0.276) / (1.043) 7.078* / (4.038) / (2.828) / (0.271) Financial Regulation Uncertainty t / (0.231) / (0.974) / (3.778) 4.701* / (2.636) / (0.228) Political Party t *** / (0.017) / (0.071) / (0.267) / (0.194) *** / (0.017) Political Party t *** / (0.026) 0.623*** / (0.075) / (0.284) 0.677*** / (0.209) 0.360*** / (0.025) Political Party t *** / (0.025) *** / (0.061) * / (0.229) *** / (0.174) *** / (0.025) Political Party t *** / (0.011) 0.113** / (0.044) / (0.161) / (0.127) 0.072*** / (0.011) Constant *** / (0.007) ** / (0.027) / (0.086) * / (0.079) *** / (0.009) Bank specific effects - F statistic Probability > F Durbin Watson d statistic Probability > Std Normal N (0,1) Hausman Test-Chi square Prob > Chi-square Breusch Pagan Test Statistic Prob > Chi-square Pooled OLS/Fixed/Random Effects a fixed random random random fixed R 2 within; between; overall ; ; ; ; ; ; ; ; ; ; Table 1 notes: Standard errors are reported in parenthesis. The symbols *, **, *** refer to the levels of significance of 10%, 5% and 1%, respectively. There are 4,336 small banks, 389 medium banks, 26 large banks, and 12 money center banks as of the beginning of the sample period. There are 265,641, 23,860, 1,590, and 756 observations, respectively. The All Banks column contains 4,763 banks and 219,847 observations. 17

18 Table 2 Panel Models by Bank Size. Dependent Variable: Return on Assets Time Period: 2000Q1 through 2008Q4 Small Banks Medium Banks Large Banks Money Center Banks All Banks Real GDP growth t *** / (0.394) 5.831*** / (1.863) / (9.213) / (6.582) 2.221*** / (0.392) Real GDP growth t *** / (0.390) 5.465*** / (1.766) / (8.715) / (6.196) 4.088*** / (0.410) Real GDP growth t *** / (0.308) / (1.819) / (9.215) / (6.399) 0.800*** / (0.297) Real GDP growth t *** / (0.333) / (1.569) / (7.902) / (5.495) *** / (0.338) Yield curve t *** / (0.003) 0.034** / (0.016) / (0.083) / (0.057) 0.031*** / (0.003) Yield curve t *** / (0.004) ** / (0.019) / (0.096) / (0.067) *** / (0.004) Yield curve t / (0.006) / (0.017) / (0.087) / (0.061) / (0.006) Yield curve t ** / (0.003) / (0.017) / (0.085) / (0.059) / (0.003) Inflation t *** / (1.789) *** / (4.485) / (14.360) / (19.142) *** / (1.852) Inflation t ** / (1.314) ** / (4.753) / (19.993) / (19.422) *** / (1.279) Inflation t *** / (1.563) *** / (5.948) / (25.233) / (27.922) *** / (1.519) Inflation t *** / (0.680) *** / (5.950) / (19.220) / (28.482) *** / (0.735) Deposits t-1/loans t-1*inflation t / (1.926) / (6.392) / (16.414) / (45.062) / (1.994) Deposits t-2/loans t-2*inflation t ** / (1.305) / (6.466) / (31.813) ** / (43.840) / (1.292) Deposits t-3/loans t-3*inflation t * / (1.492) / (7.811) / (42.038) / (64.130) / (1.457) Deposits t-4/loans t-4*inflation t / (0.226) / (7.675) / (21.715) / (64.765) / (0.233) Deposits t-1/loans t / (0.037) 0.157* / (0.085) / (0.038) 2.349*** / (0.759) 0.033*** / (0.006) Deposits t-2/loans t *** / (0.038) ** / (0.118) / (0.054) / (1.073) / (0.018) Deposits t-3/loans t * / (0.003) / (0.117) / (0.041) / (1.062) ** / (0.002) Deposits t-4/loans t ** / (0.001) / (0.087) / (0.030) / (0.801) ** / (0.001) Financial Regulation Uncertainty t ** / (0.284) / (1.431) / (7.134) ** / (5.055) ** / (0.277) Financial Regulation Uncertainty t *** / (0.343) 5.159*** / (1.647) / (8.211) * / (5.775) 1.791*** / (0.335) Financial Regulation Uncertainty t ** / (0.415) * / (1.607) / (7.990) / (5.637) *** / (0.402) Financial Regulation Uncertainty t *** / (0.397) *** / (1.662) / (8.310) / (5.835) *** / (0.407) Constant / (0.013) 0.090* / (0.046) / (0.185) / (0.202) / (0.017) Bank specific effects - F statistic Probability > F Durbin Watson d statistic Probability > Std Normal N (0,1) Hausman Test-Chi square Prob > Chi-square Breusch Pagan Test Statistic Prob > Chi-square Pooled OLS/Fixed/Random Effects a fixed random random random fixed R 2 within; between; overall ; ; ; ; ; ; ; ; ; ; Table 2 notes: Standard errors are reported in parenthesis. The symbols *, **, *** refer to the levels of significance of 10%, 5% and 1%, respectively. There are 4,336 small banks, 389 medium banks, 26 large banks, and 12 money center banks as of the beginning of the sample period. There are 141,450, 12,792, 815, and 396 observations, respectively. The All Banks column contains 4,763 banks and 155,453 observations 18

19 Table 3 Panel Models by Bank Size. Dependent Variable: Return on Assets Time Period: 2009Q1 through 2016Q4 Small Banks Medium Banks Large Banks Money Center Banks All Banks Real GDP growth t *** / (0.556) 8.083*** / (2.313) / (6.045) 7.250* / (3.804) 4.270*** / (0.544) Real GDP growth t *** / (0.525) ** / (2.151) / (5.514) / (3.490) *** / (0.513) Real GDP growth t *** / (0.444) 6.203*** / (1.713) / (4.432) / (2.804) 4.263*** / (0.434) Real GDP growth t *** / (0.463) *** / (2.055) * / (5.188) *** / (3.469) *** / (0.455) Yield curve t *** / (0.006) / (0.030) / (0.077) / (0.050) 0.022*** / (0.006) Yield curve t *** / (0.008) *** / (0.030) / (0.077) / (0.049) *** / (0.008) Yield curve t *** / (0.006) 0.055* / (0.028) / (0.074) / (0.047) 0.031*** / (0.006) Yield curve t *** / (0.007) *** / (0.027) / (0.071) ** / (0.045) *** / (0.007) Inflation t *** / (2.449) *** / (9.311) / (18.345) * / (16.228) 9.919*** / (2.501) Inflation t *** / (1.491) / (8.999) / (16.743) / (15.703) 7.360*** / (1.440) Inflation t *** / (1.592) *** / (6.807) / (13.668) *** / (11.554) *** / (1.665) Inflation t *** / (1.727) *** / (6.498) ** / (12.951) / (10.993) *** / (1.691) Deposits t-1/loans t-1*inflation t *** / (2.388) *** / (10.616) / (19.116) / (28.953) *** / (2.494) Deposits t-2/loans t-2*inflation t *** / (1.337) / (10.608) / (18.723) / (28.735) *** / (1.321) Deposits t-3/loans t-3*inflation t *** / (1.476) *** / (8.324) / (16.624) *** / (21.676) *** / (1.551) Deposits t-4/loans t-4*inflation t *** / (1.699) / (8.572) / (17.149) / (22.001) *** / (1.682) Deposits t-1/loans t / (0.017) 0.189*** / (0.063) / (0.213) / (0.433) / (0.020) Deposits t-2/loans t *** / (0.058) 0.238*** / (0.076) / (0.323) * / (0.571) 0.241*** / (0.049) Deposits t-3/loans t ** / (0.064) *** / (0.088) / (0.319) 1.606*** / (0.523) *** / (0.065) Deposits t-4/loans t / (0.036) / (0.104) / (0.215) / (0.382) / (0.037) Financial Regulation Uncertainty t *** / (0.435) ** / (1.524) 8.857** / (3.888) / (2.598) *** / (0.427) Financial Regulation Uncertainty t *** / (0.563) 4.182** / (1.655) * / (4.333) / (2.804) 2.141*** / (0.547) Financial Regulation Uncertainty t / (0.353) / (1.734) / (4.484) / (2.893) / (0.350) Financial Regulation Uncertainty t ** / (0.300) / (1.608) / (4.160) / (2.650) 0.614** / (0.301) Constant *** / (0.017) *** / (0.044) / (0.100) / (0.075) *** / (0.017) Bank specific effects - F statistic Probability > F Durbin Watson d statistic Probability > Std Normal N (0,1) Hausman Test-Chi square Prob > Chi-square Breusch Pagan Test Statistic Prob > Chi-square Pooled OLS/Fixed/Random Effects a fixed random random random fixed R 2 within; between; overall ; ; ; ; ; ; ; ; ; ; Table 3 notes: Standard errors are reported in parenthesis. The symbols *, **, *** refer to the levels of significance of 10%, 5% and 1%, respectively. There are 4,336 small banks, 389 medium banks, 26 large banks, and 12 money center banks as of the beginning of the sample period. There are 124,191, 11,068, 775, and 360 observations, respectively. The All Banks column contains 4,763 banks and 136,394 observations 19