This thesis consists of three essays on bank capital structure and risk taking.
Chapter 1 theoretically investigates why non-dilutive CoCos are prevalent, even though advocates of CoCos suggest such securities should be dilutive to reduce bank risk-taking. In an agency model with two subsequent moral hazards, we show that while dilutive CoCos deter ex-ante risk-taking and prevent a bank from being undercapitalized, penalizing existing shareholders with dilution when the bank is already undercapitalized leads to risk shifting. CoCos' designs and risk implications depend on banks' equity capitalization, with non-dilutive CoCos particularly attractive to capital-constrained banks, because such securities can maximize the banks' financing capacity by tackling only the ex-post risk shifting.
Chapter 2 empirically documents how banks' risk appetite is affected by nondilutive CoCos. This chapter tracks the lending behaviour of G-SIBs after their first CoCo issuance. Regarding the pricing of syndicated loans, we find CoCo issuers on average charge a higher spread, controlling for borrower, bank and loan specific factors. Importantly, compared to undercapitalized CoCo issuers, well-capitalized issuers ask for lower spreads, implying a greater risk appetite. Additionally, there is no evidence suggesting banks intend to replace common equity with AT1 CoCo for lower financing costs. Demandable deposits commit banks to provide liquidity to retail investors when needed. Such a commitment rewards banks with higher financing capacity at risk of greater exposure to liquidity risk.
In Chapter 3, we develop a dynamic capital structure model in which the depositors concerns about banks' commitment problem to repayment and to future debt policy. We find banks have less incentive to cut leverage and set a higher long-term target when the withdrawal rate is low. We highlight that high withdrawal rates create a trade-off between excessive borrowing and insolvency risk.