Moral hazard is the shadow cast by every rescue. If losses are socialised — by deposit insurance, central-bank support, or the expectation of a bailout — the actors taking the risks no longer bear the full cost of failure. Rational players respond by taking more risk, and the safety net ends up breeding the accidents it was built to catch.
The tension is real, not rhetorical. The remedies that stop bank runs and contagion work precisely because they promise support in bad states of the world; that same promise dulls the incentive to avoid bad states. Crisis policy is therefore a permanent negotiation between stopping today’s fire and not subsidising tomorrow’s arson.
The Federal Reserve-brokered rescue of LTCM in 1998 became the canonical modern exhibit: a hedge fund so interconnected that its creditors were assembled to absorb it, teaching the market that sufficient size and entanglement purchase official concern. A decade later the 2008 crisis stress-tested the lesson — Bear Stearns was caught, Lehman was let go, and the ensuing chaos suggested that ex-post toughness has its own catastrophic price.
The practical consensus that emerged is to separate the actor from the system: save the function, punish the firm — wipe out shareholders, replace management, and lend freely only against sound collateral. Whether that separation holds under panic conditions is, on the historical evidence, an open question.