A few days ago I criticized Dean Baker’s estimation that government is subsidizing large banks to the tune of $34.1 billion dollars by implicitly designating them “too big to fail”. Dean was kind enough to provide me with the FDIC data he used in his paper, and after looking at the data I’ve concluded that his estimates are even more speculative and unsound than I first though; it is unfortunate that they seem to be becoming conventional wisdom.

To recap, the gist of his paper was the when the government made it apparent that some banks were too big to fail, it allowed them borrow at a lower rate than small banks, which is in effect a subsidy for being too big to fail. He attempted to quantify the size of the subsidy by looking at average borrowing rates for large banks compared to the rates for  smaller banks, and how the gap between them has grown since TBTF policy became established. He used this gap in borrowing rates to back out a dollar value for the subsidy.

As far as I can tell, the specific number Dean used was the quarterly total interest expense as a percent of average total liabilities for two groups of banks; those with total assets above $100 billion, which are too big to fail, and those with assets below $100 million, which are not.

The graph below shows the data he used to get the $34.1 billion dollar estimate. Each bar represents the difference in quarterly cost of borrowing for large banks relative to small bank.  The period which he deems as being post-TBTF policy are the green bars depicted below, and the control period of pre-TBTF policy is in red. Dean attributes the difference in the average values in these two periods to the large banks becoming TBTF in the period in the green.

To Dean’s credit, he recognizes that the change in relative borrowing costs may result from general economic uncertainty, so as a more conservative estimate he compares the post-TBTF gap to another period of economic uncertainty: the end of the 2001 recession, which is depicted below. The difference in the average value in these two periods is his more conservative estimate of the TBTF subsidy, which he uses to come up with a dollar value of $6.3 billion. This estimate is ignored by most commenters, who tend instead to focus on the more shocking $34.1 billion estimate.

In addition to the problems with this that I pointed out before, what I learned from looking at the data is that the spread between large bank and small bank borrowing costs was actually higher during the fourth quarter of 2001 than it ever reached during the current financial crisis. Is it reasonable to assume that the gap in 2001 was attributable to large banks being too big to fail? I don’t think so, and I doubt if anyone can point to serious speculation that any of the countries largest banks might fail and that they would bailed out by the government during the fourth quarter of 2001.  This suggests that a rate spread resulting from people flocking to larger banks during times of economic uncertainty, like the period immediately preceding 9/11, can be as big as or bigger than the rate spread we are currently observing. So why should we believe that that is not what is currently occurring? There is no reason to assume this; to do so is pure speculation.

The graphs also illustrate that the spread actually begins rising gradually from end of 2006 to the first quarter of 2008, rising from slightly negative up to 0.40%.  This large increase predates the period when Dean claims TBTF policy was  established. The increase in spread that occurs from the end of 2008 on could easily be a continuation of a general trend that began in 2007 and largely followed the worsening financial market and economic conditions. He presents no evidence that this is not the case.

It’s unfortunate how widely cited this estimate has become. In addition to the ones I mentioned in my previous blog (James Kwak and Felix Salmon), I’ve found that Baker’s estimates have been cited approvingly by Zero Hedge, The Huffington Post, Matt Yglesias, USA Today, Gretchen Morgensen in the New York Times, and even Wikipedia. I dare say this estimate has become conventional wisdom.

The reason these number are getting mileage is not because they are reasonable estimates derived from a sound methodology, but because they put a shocking and concrete number on something that most people believe is true anyway: banks are benefitting financially from their too big to fail status. This number is much closer to pure conjecture than it is to a reasonable estimate, and it should be treated as such.