The Federal Reserve is mulling a new set of tougher banking rules to boost the odds that the financial system will remain sufficiently liquid when the next crisis strikes. It’s a worthy goal, if only because one day another event will surely arrive. But all the usual caveats apply when it comes to engineering outcomes in economics, starting with the sober reality that any efforts to sidestep implosion due to banking turbulence are forever linked with the potential for blowback by way of moral hazard. In fact, there can be no permanent solution for managing bank risk if the goal is simultaneously keeping the threat of bank runs to a minimum while maximizing prudence in matters of investing and lending in the private sector. The ideal regulatory prescription that promotes each of these two competing interests is in constant flux due to the ebb and flow of the business cycle.

Walter Bagehot long ago identified this dynamic (some might say unstable) nature in the realm of the bank-macro relationship. There are times when a lender of last resort—a central bank—“should lend early and freely,” he famously advised in his 1873 book Lombard Street: A Description of the Money Market. But he was also careful to point out that any emergency lending should be extended to “solvent firms, against good collateral, and at ‘high rates.’”

The lesson is that a solution can easily turn into a problem because the macro profile is in a state of perpetual change. Accordingly, if the banking industry assumes that it will always be bailed out, no matter how many loans it extends to unworthy borrowers, well, that’s a recipe for trouble. Sound familiar? Something along these lines unfolded in the last housing bubble, when credit was easily and freely extended to high-risk clients–moral hazard in the extreme, circa 2006.

The potential for trouble by going off the deep end with lending should be minimized, but not to the extent that bad loans are allowed to take down the economy when ill-advised decisions in banking inevitably rise to a critical level. And, yes, poor decisions in banking in the aggregate will always reach a tipping point… eventually, as Hyman Minsky pointed out. Stability ultimately begets instability, and vice versa. Perhaps that’s partly because it’s impossible to find the right mix of regulations across long time horizons that a) keeps banking-induced panics to a minimum without b) promoting dubious lending and speculative activity.

Even if we could identify such an optimal mix, it would be constantly changing as economic conditions evolved. In the early stages of growth, the economy needs more speculation and lending; after an extended run of expansion, the opposite blend is timely. The idea that we can write one perfect set of banking regulations for all time is just fooling ourselves. What’s required today for current banking/macro conditions will be sub-optimal next year.

That doesn’t mean that the search for intelligent banking regulation is a waste of time. Quite the contrary: this is a critical aspect of public policy. But it’s a work in progress, now and forever. There are no optimal choices in macroeconomics as a practical matter, but there may be a productive chain of policy adjustments that, in real time, are the worst possible decisions available… except when compared to everything else.