The 2007–09 global financial crisis led policy-makers to require that banks hold more capital. Higher minimum capital requirements reduce losses to stakeholders in case of bank failures, but they may constrain intermediation during financial crises. On the other hand, capital buffers do not constrain intermediation if the bank is willing to limit payouts instead. Buffers are macroprudential rather than microprudential—guarding against sudden decreases in economy-wide intermediation activity rather than against losses from a bank’s failure.

Using a model of optimal regulation of bank capital, this paper highlights fundamental inefficiency in economies with financial intermediation. It presents novel implications for the cyclicality of bank capital regulation. Instead of looking at inefficiently high bank risk-in the run-up to financial crises, I focus on inefficiently low bank risk-taking during financial crises. Such low risk-taking is inefficient because regulators could mitigate the risk of binding bank funding conditions by increasing a bank’s future profits to offset decreases in bank equity.

One policy implication of my findings is that banks should build up capital buffers in normal periods. Making banks more resilient to loan losses reduces the forecasted severity of financial crises and lowers their frequency. Another policy implication is that banks need time to rebuild capital buffers after a financial crisis, along with regulation to increase their profitability. Raising the prospect of future profits during the financial crisis will increase a bank’s access to outside funding and reduces the actual severity of the crisis.