That's how we roll: an experiment on rollover risk

There is consensus that the recent nancial crisis revolved around a crash of the short-term credit market. Yet there is no agreement around the necessary policies to prevent another credit freeze. In this experiment we test the eects that contract length (i.e. maturity mismatch) has on the market-wide supply of short-term credit. Our main result is that, while credit markets with shorter maturities are less prone to freezes, the optimal policy should be state-dependent, favoring long contracts and lower maturity mismatch when the economy is in good shape, and allowing for short-term contracts when the economy is in a recession. We also report the possibility of credit runs on rms with strong fundamentals, something that cannot be observed in the canonical static models of nancial panics. Finally, we show that our experimental design produces rich learning dynamics, with a text-book bubble and crash pattern in the market for short-term credit.