Too Unexpected to Fail: Bail-Out Policy and Sudden Freezes

I present a mechanism that relies on the interaction of coordination and ambiguity (Knightian uncertainty) and makes precise how a loss of confidence can arise in loan markets, leading to a systemic liquidity crisis. The paper studies a simple global-game coordination model among lenders to a financial intermediary and shows how a market haircut arises in equilibrium. I show how the haircut responds to a variety of parameters. In particular, I show that coordination is non-robust to ambiguity in investor signals and becomes fragile in an environment with ambiguity. This leads to the haircut jumping up suddenly, possibly to 100% when enough lenders are ambiguity-sensitive. Further, I show that the fragility of coordination implies that in such an environment, policy itself becomes a systemic trigger. If the regulator fails to rescue an institution that the market expects to be saved, which in turn changes market expectation about policy for other institutions even slightly, an immediate systemic collapse of liquidity ensues. The results explain both the contagious run on liquidity markets at the advent of the recent crisis as well as the liquidity market freeze after the Lehman collapse. While what matters for the possibility run is whether an institution is too-unexpected-to-fail (TUTF), it is likely that institutions typically considered too-big-to-fail (TBTF) are also likely to be TUTF. The results then show that TBTF institutions limit the spread of crises, and breaking up a TBTF increases systemic vulnerability. Further, the results cast some doubt on the efficacy of the ring-fencing policy proposed by the UK banking commission.