Evidence from the financial crisesGlobal financial crisis of 2008 and european sovereign debt crisis

Resumen

This dissertation contributes to the empirical literature in banking and financial crises. The first chapter is co-authored with Gerald P. Dwyer, Maria Nieto and Margarita Samartín. In this chapter, we examine the implications of all three major European Central Bank’s unconventional monetary policies, long-term financing operations (LTRO), asset purchase programs and deposit facility rate for bank investments among loans, government securities and cash deposited in central banks. The second and third chapters are co-authored with Rebel A. Cole. The second chapter aims to contribute to the ongoing debate on the regulations of banks’ liquidity choices. In the third chapter, we develop empirical strategy to test the theoretical predictions of “A model of shadow banking” published in 2013 of the authors Nicola Gennaioli, Andrei Shleifer, and Robert W. Vishny. We want to show that holding securitized financial products on bank balance sheets diversifies bank’s idiosyncratic risk at the expense of increasing the systematic risk. The title of the first chapter in this thesis is “The effects of the ECB’s unconventional monetary policies from 2011 to 2018 on banking assets”. The Global Financial Crisis that started in 2008 transformed into Sovereign debt crisis in Europe in 2010. As a response to this severe economic downturn, the European Central Bank (ECB) implemented a diverse set of unconventional monetary policy measures that were effective in the period from 2011 – 2018: lending programs, asset purchases programs and low interest rate economic environment. The lending programs represent ECB’s facilities that provided direct liquidity to banks in a form of loans and this was the first time ever that ECB’s had taken this kind of initiative. They included two slightly different lending programs, long-term refinancing operations (LTRO) and targeted long-term refinancing operations ((T) LTRO). (T) LTRO represented the single policy in this toolkit of unconventional monetary policies that controlled to certain extent for the total amounts and structure of loans on banks’ assets portfolios. The primary research goal of the paper is to shed light on how effective are (T) LTRO for banks’ assets portfolio allocation with respect to the rest of the ECB’s unconventional monetary policies. We are interested in the effects of (T) LTRO on three banking assets that appeared to be of very specific interest to banks over Sovereign debt crisis and post-crisis period: loans, government securities and deposits in central banks. Furthermore, the literature on sovereign debt crisis provides evidence that the level of financial distress of a country has a particular role in banks’ choices between loans and government securities. This is why we are interested in how the effects of (T) LTRO and the rest of the policies differ due to the level of financial distress of the country. For this purpose, we conduct our analysis on two sets of countries: non-crisis and crisis euro area countries. A large body of research finds that a set of macroeconomic conditions and bank-specific factors explains bank assets portfolios. For the purpose of our study, we take into account country’s gdp growth, bank’s capital, and deposits from other banks as the most standard macroeconomic and bank-specific factors, respectively. The recent literature on European sovereign debt crisis suggests that banks’ incentives for investments between loans and government securities are shaped in large extent by the change in sovereign’s risk. This is why we consider the sovereign’s risk in our analysis as an additional important macroeconomic factor affecting bank assets portfolios and we measure it through the government’s level in credit default swaps. Bank-specific factors that we take into account in our analysis, bank’s capital and deposits from other banks, are the two fundamental financing sources for banks according to the banking theory. Thus, it is of essential importance to account for them in our setting. In our setting, the relationship between bank investment assets and bank-specific factors is endogenous as the bank level variables are simultaneously determined at the individual bank level. However, the relationship of the bank investment assets with the macro and ECB policy variables is exogenous because we construct the ECB’s policy variables at aggregate level. We are interested in the total effects of the macro and ECB’s policy measures on bank individual behavior. It is unlikely that these are correlated with bank confounding factors or that bank investment assets would affect simultaneously macro and policy variables. Thus, the basis of our empirical method is a multivariate framework in which we combine endogenously determined bank level variables and exogenously determined macro and policy variables. We address the issue of endogeneity within the bank level variables by applying Panel Vector Autoregression (PVAR) framework. Panel vector autoregression models have been increasingly used in applied research and especially along with the corresponding Stata estimation package of Love and Abrigo (2016). So far, researchers have been taking into account only a vector of endogenous variables in the main PVAR regression specification. In our model, macro and ECB’s policy variables are exogenous to bank investment variables and thus, they are not part of the vector autoregression setting of endogenous variables. Then, our empirical method could be also classified within the terminology of PVAR as a PVAR model with exogenous variables, PVARX. In our knowledge, we are the first to consider exogenous variables within a panel VAR setting. We document two important findings. First, the results suggest that (T) LTRO is the single policy being effective in turning around banks’ incentives for investments between loans and government securities in crisis countries, which represented a major issue in the onset of the European sovereign debt crisis. Second, we find that (T) LTRO plays the key role in enhancing the total credit output across the entire euro area, this means for both set of countries, non-crisis and crisis countries. We show that the existence of the first round of LTRO was particularly critical for enhancing total credit output by banks in non-crisis countries, as the standard factors affecting banks’ assets portfolios as well as the rest of the ECB’s policies stimulated banks’ investments other than lending. However, when we quantify the effects of one mean increase in the first round of LTRO and (T)LTRO, we find that , for instance, one mean increase in LTRO increase the total loans output by 0,22% while one mean increase in (T) LTRO increase the total loans output by 0.46 % in the case of non-crisis countries. There is similar double increase effect in the set of crisis countries, too. In addition, targeted LTRO achieved the desired effect on banks in crisis countries, too: increasing total loans outputs by banks. Hence, we find that targeted LTRO had the full effect on banks’ assets portfolios across the entire euro area, which was not the case with the first round of LTRO. The title of the second chapter is “On-balance sheet securitized assets and banking risks: Implications for the quality of liquid assets”. The study aims to contribute to the ongoing debate on the regulation on banks’ liquidity choices. We examine asset-backed securities that had the role of liquid assets on banks’ balance sheets in the period before the global financial crisis of 2008 (GFC). We find that this particular category of securities generates banks’ interconnectedness, an unattended risk exposure at the time. Therefore, we argue that the level of interconnectedness must be a determinant in regulating liquid asset categories and, consequently, banks’ private liquidity choices. Further, individual bank investments in certain assets must be limited when those assets are subject to interconnectedness with the rest of the financial system (or its segments), in addition to their underlying default risks. Put differently, the limitations of banks’ private liquidity choices should be dependent on the types of risks they generate. Our primary interest are the effects of a specific category of asset-backed securities on risk- ratio, where the risk-ratio is a ratio of bank’s interconnectedness over its total risk. The specific category of asset-backed securities of our interest is the other mortgage-backed securities (other MBS). Other MBS constitutes a central interest of this study because it represents an asset class that accounted for significant fractions of banks’ liquid asset portfolios in the preceding period of the GFC. Moreover, beginning from 2004, it significantly surpassed the holdings of residential MBS and treasuries and agencies securities on banks’ asset portfolios. By construction, other MBS is a category of asset-backed securities that are backed by real estate loans other than residential mortgages. The essential difference between the other MBS and the residential MBS is not just merely in the underlying asset that was used as collateral in the construction of these asset-backed securities. By construction, other MBS represents a collection of mortgage-backed securities backed up by commercial mortgages and, as well as various structured finance products, that are backed by commercial mortgages, whereas residential MBS represents a more homogenous category of mortgage-backed securities that is only backed by residential mortgages. We employ two different empirical methodologies. First, we employ regulatory discontinuity analysis (RDA) to document the effects of other MBS on a bank’s interconnectedness relative to its total risk on average. In our identification strategy, we use Enron collapse at the end of 2001 and a regulatory intervention that was related to the off-balance-sheet structured finance vehicles in July 2004. Both events directly affected the market reactions to securitization activities and products. We then apply this methodology to examine the heterogeneity of banks’ responses to the risk-ratio that is to a bank’s interconnectedness relative to its individual total risk. Therefore, we construct a pair of subsets on the basis on bank’s size, capital, and leverage. We examine the effects of other MBS on risks ratio separately on the subsets of large vs. small banks, top vs. bottom-capitalized banks, and top vs. bottom-leveraged banks. The regulatory discontinuity analysis allows us to estimate the average effects of other MBS on risk-ratio. Considering that we are primarily interested in determining if other MBS increased a bank’s interconnectedness relatively to its total individual risk, we need evidence on the effects of other MBS on the extreme quantiles of risk-ratio. The existence of the effects of other MBS in the upper or bottom quantiles of the risk-ratio distribution implies that other MBS affected bank’s interconnectedness rather than it exerted any effect on the bank’s total individual risk. Therefore, we employ quantile regression analysis to examine the effects of the other MBS on different percentiles of the risk-ratio. If the effects appear to differ between the extreme quantiles of the risk-ratio distribution, it indicates that other MBS significantly contributes or not to the bank’s interconnectedness with the banking system. Employing regulatory discontinuity analysis, we find that, on average, other MBS exerted a downward effect on a bank’s interconnectedness relative to its total risk following the Enron collapse in 2002. Further, we find that other MBS exerted an upward effect on the bank’s interconnectedness relative to its total risk on average after the regulatory announcement had been released in 2004. The results suggest that at different points of time and different market developments, the other MBS affects bank’s interconnectedness level differently. This implies that its liquidity features change with the market developments and consequently, their liquidity capacities should be adjusted accordingly. Furthermore, we find heterogeneity in banks’ responses to interconnectedness due to the holdings of other MBS with respect to a bank’s size, capital, and leverage. We document that other MBS drive bank’s interconnectedness relative to its total risk when banks are large and have high capital. With respect to the leverage, we find that other MBS drives bank’s interconnectedness in the case of low-leveraged banks only. The quantile regression analysis shows that the effects of other MBS are statistically significant at 1% significance level along the upper quantiles of the risk-ratio distribution. This result classifies other MBS as generators of bank’s interconnectedness. They develop bank’s interconnectedness with the system, exposing banks to a new source of risk rather than only affecting bank’s total risk. We consider this finding very important since indicates that asset portfolios are not constant but rather dynamic in the presence of financial innovation. Specifically, an increase in the exposure of