This dissertation analyzes the patterns of business elites' offshore usage and its impact on domestic politics. I advance two theoretical claims. First, the availability of offshore zones makes autocracies stronger by facilitating patron-client exchanges that keep regimes in power. Second, I argue that vulnerability to political risks prevents firms from using tax havens for tax avoidance.

After providing some historical and economic background on the evolution of tax havens, I present a simple theory of patron-client exchange in autocracies that rely on a tax haven as a commitment device for a ruler. I argue that the availability of such a device makes elites more docile to the regime and prevents movements towards a more open and constrained regime.

Then, I offer cross-national evidence that speaks to some of my theory's implications. In particular, I use previously untapped country-level data on the size of offshore holdings to test if offshore utilization prevents democratization. I find that the countries with higher offshore wealth in the year 2007 were less likely to democratize between 2007 and 2016. I show an array of placebo tests and robustness checks to demonstrate that my results are unlikely to be driven by previous regime dynamics, coup risk, or other potential confounders.

Then, I move on to explore another important implication of my theory: an exogenous marginal decrease in the safety of offshore assets should trigger an increase in the spoils the ruler shares with the elite. I use recent rounds of international sanctions against the members of the Russian business community linked to the Russian regime to explore this empirical prediction.

I also discuss corporate tax avoidance through tax havens. In particular, I ask why some firms avoid local taxes by establishing affiliates in tax havens while others do not. I argue that firms that are more vulnerable to political risk are less likely to engage in income shifting since complying fully with the tax law makes firms more valuable to the host government. I explore the empirical implications of my theory using a registry-based dataset that allows tracing connections between firms and tax havens. My findings improve our understanding of tax capacity and can inform international efforts to combat tax avoidance.

I investigate the determinants of bank failures after the financial crisis of the years 2007 - 2009 to build a predictive model of bank failures.

I use two paradigms for prediction: accuracy-maximization and Neyman-Pearson paradigm. Accuracy-maximization implies that Type I errors and Type II errors are equally costly, thus out-of-sample predictive accuracy is the most important parameter for evaluation. Neyman-Pearson paradigm implies setting an upper bound for Type I errors and minimizing Type II errors within that bound. In this case, the costs associated with Type I and Type II errors can be different.

I find that, because the bank failures are rare events, many of the accuracy-maximizing classifiers tend to assign all the observations to the class of non-failing banks. This achieves out-of-sample predictive accuracy of 96 percent but misses all the failures. Two algorithms, post-Lasso logit, and random forest tend to have a relatively low level of Type II errors.

The classification with the Neyman-Pearson paradigm performs better in terms of minimizing Type II errors while containing Type I errors. All of the algorithms, in out-of-sample testing, were able to identify at least 50 percent of the failing banks, while having a false positive rate below ten percent. The minimum share of Type II errors was displayed by Ada-Boost algorithm (24 percent), while GLM with LASSO penalty and sparse LDA did not perform much worse (the level of Type II errors were 27 percent).

My analysis produces additional substantive insights. I find that low profitability and a high proportion of impaired loans are the most important factors for bank failures.