This paper compares the out-of-sample predictive performance of different early warning models for systemic banking crises using a sample of advanced economies covering the past 45 years. We compare a benchmark logit approach to several machine learning approaches recently proposed in the literature. We find that while machine learning methods often attain a very high in-sample fit, they are outperformed by the logit approach in recursive out-of-sample evaluations. This result is robust to the choice of performance metric, crisis definition, preference parameter, and sample length, as well as to using different sets of variables and data transformations. Thus, our paper suggests that further enhancements to machine learning early warning models are needed before they are able to offer a substantial value-added for predicting systemic banking crises. Conventional logit models appear to use the available information already fairly efficiently, and would for instance have been able to predict the 2007/2008 financial crisis out-of-sample for many countries. In line with economic intuition, these models identify credit expansions, asset price booms and external imbalances as key predictors of systemic banking crises.

n modern finance, financial decision making of households plays an important role. At least the recent global financial crisis recalled the role of the household sector in financial stability. Besides the debate how, for example, the introduction of financial sector regulations and reforms can enhance the stability of financial systems and prevent future crises, research focuses on how household behavior contributes to financial stability and the performance of the economy. By allocating their re- sources, e.g., making decisions about labor supply, consumption, savings, and debt, households directly affect market production and prices and, thus, make a relevant contribution to financial stability. The exposure to the financial sector enables house- holds to influence the overall economy (Santoso and Sukada, 2009).

When an external shock such as the economic crisis in 2008/2009 occurs, the interconnectedness of sectors can be affected. This paper investigates whether the R&D sector experienced changes in its sectoral integration through the recession. Based on an input-output analysis, it can be shown that the linkages of the R&D sector with other sectors remain stable. In some countries, the inter-sectoral integration becomes even stronger. Policy makers can be encouraged to use public R&D spending as a means of fiscal policy against an economic crisis.

Reserve requirements, as a tool of macroprudential policy, have been increasingly employed since the outbreak of the great financial crisis. We conduct an analysis of the effect of reserve requirements in tranquil and crisis times on long-run growth rates of GDP per capita and credit (%GDP) making use of Bayesian model averaging methods. Regulation has on average a negative effect on GDP in tranquil times, which is only partly offset by a positive (but not robust effect) in crisis times. Credit over GDP is positively affected by higher requirements in the longer run.

Does private equity (PE) contribute to financial fragility during economic crises? The proliferation of poorly structured transactions during booms may increase the vulnerability of the economy to downturns. During the 2008 crisis, PE-backed companies decreased investments less than did their peers and experienced greater equity and debt inflows, higher asset growth, and increased market share. These effects are especially strong among financially constrained companies and those whose PE investors had more resources at the crisis onset. In a survey, PE firms report being active investors during the crisis and spending more time working with their portfolio companies.

This paper compares the out-of-sample predictive performance of different early warning models for systemic banking crises using a sample of advanced economies covering the past 45 years. We compare a benchmark logit approach to several machine learning approaches recently proposed in the literature. We find that while machine learning methods often attain a very high in-sample fit, they are outperformed by the logit approach in recursive out-of-sample evaluations. This result is robust to the choice of performance measure, crisis definition, preference parameter, and sample length, as well as to using different sets of variables and data transformations. Thus, our paper suggests that further enhancements to machine learning early warning models are needed before they are able to offer a substantial value-added for predicting systemic banking crises. Conventional logit models appear to use the available information already fairly effciently, and would for instance have been able to predict the 2007/2008 financial crisis out-of-sample for many countries. In line with economic intuition, these models identify credit expansions, asset price booms and external imbalances as key predictors of systemic banking crises.

Understanding how banks react to unexpected events has become a very important economic and social question, especially since the financial crisis (Ivashina and Scharfstein, 2010; Puri et al., 2011). Whereas previous financial crises had largely stayed in the realm of finance, or very limited areas of the economy, the financial crisis of 2007-2008 demonstrated that unexpected financial shocks can have severe implications for the real economy in general, impacting the lives of a large cross-section of the population, for example through general reductions in employment (Chodorow- Reich, 2014; Popov and Rocholl, 2017). This new realization has led to an extensive literature on how banks react to unexpected events, especially if and how they transfer such shocks to firms and households. As a result, understanding exactly how shocks are transferred not only between banks (Popov and Udell, 2012; Schnabl, 2012), but also between banks and firms has become a crucial aspect of financial research (Peek and Rosengren, 2000; Gan, 2007; Ongena et al., 2015; Acharya et al., 2018; Gropp et al., 2018; Huber, 2018). It has returned into focus the idea that a functioning connection between banks and firms constitutes a crucial part of a well-functioning economy. This thesis aims to contribute to the understanding of how this bank-firm relationship functions and what pitfalls it might entail.

We isolate the direct bank-to-sovereign distress channel within the eurozone’s sovereign-bank-loop by exploiting the global, non-eurozone related variation in stock prices. We instrument banking sector stock returns in the eurozone with exposure-weighted stock market returns from non-eurozone countries and take further precautions to remove any eurozone crisis-related variation. We find that the transmission of instrumented bank distress, while economically relevant, is significantly smaller than the corresponding coefficient in the unadjusted OLS framework, confirming concerns on reverse causality and omitted variables in previous studies. Furthermore, we show that the spillover of bank distress is significantly stronger for countries with poorer macroeconomic performances, weaker financial sectors and financial regulation and during times of elevated political uncertainty.

Zehn Jahre nach der Lehman-Pleite werden die Finanzmärkte besser kontrolliert denn je. Das kann böse Folgen haben.