Optimal bank capital

Having its roots in the financial system, the world economic downturn at stake since 2008 has revealed the importance of capital requirements for banks. Reinhart and Rogoff (2009) have shown that financial crises have substantial economic impact, reducing the GDP of 10% or more, with a non- neglecting part being permanent. This highlights the importance of a stable financial market for the entire economy. As a consequence, the new Basel III agreements have substantially increased capital requirement, which has generated important debates on the cost of raising capital for banks. In particular, banking associations have argued that this new regulation would increase funding costs for banks and therefore increase the cost of bank-credits. However these arguments generally focus on short-term feedbacks, whereas in order to define an adequate legislation and prevent new crises, a long-term perspective is essential. Using a large worldwide database of 98 banks over a period of 25 years, we have tested the Modigliani and Miller theorem, which states the neutrality of capital structure on the value of the bank - and hence on the weighted average cost of capital.1 Our main results reveal that financial market tends to underprice leverage risk compared to M&M predictions. Leveraging increases market value of banks making rational the shareholders’ decision of opting for very high leverage.2 We also show that the gain of leverage is much higher during the crisis period (after 2007) than during the preceding more stable period (1990-2006). This gain is also higher for systemically important banks, which are very likely to be bailed out by the government in case of distress. We argue that the underpriced transfer of risk from banks’ shareholders to governments is the most convincing theoretical argument explaining this departure from M&M theorem. The three other possible explanations - tax advantages, agency costs between managers and shareholders, and role of completing financ