By Kindred Winecoff
Felix Salmon asks when the Fed will start caring about banking regulation, and points to this column by Jesse Eisinger. I'm not sure whether the question is serious or facetious, but I have an answer: Never. And it's not really the Fed's fault.
What do I mean? As it happens, I wrote my thesis on this question. The paper jumps off of previous literature that has established that monetary policy and regulatory policy have a natural tension. Monetary policy is counter-cyclical; regulatory policy is pro-cyclical. The principal-agent dynamic that exists when central banks are also bank regulators is such that central banks cannot credibly commit to either let struggling banks fail, or to tighten monetary policy when that might be damaging to banks.
As a result, banks know they'll get liquidity support from regulatory central banks when needed, so they act more riskily than they otherwise would. It's a form of moral hazard that is distinct from the typical TBTF hazard, because the mechanism is monetary rather than fiscal, but it operates similarly. And it may be worse: It applies to all banks, not just TBTF banks.
To answer the question, I compare overall bank capitalization ratios across OECD countries and time (1992-2007) using a standard time series cross-sectional econometric model. These countries were all in compliance with the Basel accords by at least 1999, so they all subscribed to some broadly similar regulatory guidelines, probably as far back as 1992. I use fixed effects to isolate changes within countries (so the stats aren't biased by idiosyncratic variation) and look at two interventions into the time series: The introduction of the euro in 1999, which removed monetary authority from a number of domestic central banks and gave it to the ECB; and the reorganization of domestic regulatory institutions in five countries, all of which shifted regulatory authority away from their central banks in 1999-2000.
I find that where monetary authority and regulatory authority were split, banking systems had higher capital ratios than when they were unified (the only exceptions being the PIGS -- actually, just Portugal, Greece, and Spain). The coefficients are substantively large and statistically significant at the usual levels. The results are robust to the inclusion of controls and alternative specifications.
The takeaway is that institutional design is important here. If you give one institution conflicting mandates -- one to act counter-cyclically, and one to act pro-cyclically -- then at the relevant margin one of those two has to give. Private sector actors are smart enough to know that, and adjust their behavior accordingly.
Update: This paper from Douglas Diamond and Raghuram Rajan, uploaded to NBER this week (ungated version), appears to be making a similar theoretical argument, although I haven't had time to go through it carefully yet.