Three essays on financial institutions and risk management

Abstract

This dissertation is comprised of three chapters on financial institutions and risk management, which revolve around portfolio theory, asset pricing, financial networks, and banking. My dissertation underlines the importance of asset allocation under uncertainty for banks and institutional investors. Additionally, it sheds an important light on the implications of asset allocation in terms of forming financial networks and achieving financial stability. The investigation undertaken in this dissertation is both theoretical and empirical, with the exception of the second chapter in which the empirical analysis is conducted using a simulation study due to the confidentiality of the underlying interbank data.

In the third chapter, I study the conjecture that there may exist an implicit value in forming tracking error portfolios relative to mean-variance (MV) optimal portfolios. From an ex-ante perspective, tracking error portfolios are deemed MV sub-optimal. However, MV portfolio decisions are ``contaminated'' with estimation error that adversely affects their performance ex-post. I raise the question whether market orientation helps to mitigate estimation error and, hence, results in a better portfolio performance ex-post - even though such practice is deemed sub-optimal ex-ante. To investigate this, I derive a number of analytical conditions under which an implicit value does exist. %I show that such implicit value improves with investors risk tolerance, asset betas, and the number of underlying assets; but it decreases with market volatility and sample size. An empirical design with bootstrapping strongly supports my analytical conclusions. Moreover, I show that a market tracking long, but MV optimal short, portfolio provides evidence of excess risk-adjusted returns.

Chapter 2 contributes to the recent macroprudential policy addressing the resilience of financial systems in terms of their interconnectedness. I argue that beneath the overnight market there is a fundamental latent network that affects the funds distribution among entities. I propose a framework that identifies such latent network using a statistical learning procedure. Specifically, the framework reverse engineers overnight signals observed as banks conduct their reserve management on a daily basis. My simulation-based results show that possible disruptions in funds supply are highly affected by the interconnectedness of the latent network. The proposed framework serves as an early warning system for regulators to monitor the overnight market and to detect ex-ante possible disruptions based on the inherent network characteristics.

In Chapter 1, I raise the question whether the difference in risk-adjusted returns between big and small banks is reflective of an ex-ante premium or an ex-post inefficient risk-taking. Proposing a unique diversification index to capture the risk-taking behavior of banks, I find robust evidence to support the latter view. My findings indicate that too-big-to-fail (TBTF) subsidies not only disproportionately benefit large banks by granting them what amounts to a free insurance policy, but also distort the market in such a way that they incentivize large banks to take excessive risk and, hence, invest inefficiently. From a conventional point of view, diversification should be beneficial. However, there are limits of diversification that larger banks are more likely to encounter. I attribute these limits to the greater complexity and interconnectedness that larger banks more likely to face.