Paul Krugman used his column today to tell us that any Democrat in the White House will take a tough line on regulating Wall Street. I hope that he is right, but am a bit more skeptical given past associations. But beyond the speculation, there is one factual matter where I would differ his assessment.

At one point he argues that the implicit "too big to fail" (TBTF) subsidy for large banks has mostly disappeared due to the Dodd-Frank reforms. He cites a blog post by Mike Konczal, which in turn relies on a study by the Government Accountability Office (GAO). The GAO study does seem to suggest that the TBTF subsidy has largely disappeared.

It uses 42 different models to estimate the size of the subsidy year by year. While its models get highly significant results showing a large subsidy at the peak of the crisis, most find no subsidy in 2013. This can be seen as a victory. But if we look at the results more closely, we find that the study also finds little evidence of a TBTF subsidy in 2006. While 28 of the 42 studies did get significant results indicating a subsidy, compared to just 8 in 2013, the average size of the subsidy looks to be very small. From the chart it appears to be less than 10 basis points (a tenth of a percentage point).

Obviously, the big banks did enjoy too big to fail protection in 2006, since only Lehman was allowed to fail in the crisis, yet the GAO analysis implies that this held very little value. The problem here is that interest rates spreads, between more and less risky assets, tend to collapse in normal times. The basic story is that fire insurance is not worth much if no one thinks there can be a fire.

In the GAO analysis it is difficult to distinguish between a situation in which big banks don't pay much less interest than anyone else because people no longer believe the government will bail them out in a crisis and a situation in which the big banks don't pay much less interest than anyone else because no one thinks that anyone is about to go out of business. In the latter case, TBTF insurance may still exist, it would just be difficult to measure by these techniques.

It is worth noting that Mike's blogpost also referred to a study by the I.M.F. which found a TBTF subsidy of 25 basis points. That may not sound like a very big deal, but 25 basis points on $10 trillion in big bank assets comes to $25 billion a year. That's about 0.6 percent of the federal budget, more than we are spending on TANF.

I wouldn't say the I.M.F. methodology is necessarily better, but I would say that I am not convinced the TBTF insurance is history. If Goldman sinks itself, I would not bet that the Treasury and the Fed would be prepared to let the market work its magic.