About That $83 Billion Bank Subsidy. We Still Mean It.

Our estimate that the largest U.S. banks get a taxpayer subsidy worth about $83 billion a year remains a subject of some interest. Notably, Dealbreaker's Matt Levine posted a response to our response to his response to our original editorial.

In the latest installment, Levine makes the interesting observation that even poorly rated small banks enjoy lower overall funding costs than big banks. Specifically, he selects a group of small banks with BBB-minus credit ratings, and finds that they pay about 0.2 percentage point less to borrow money than do the five largest U.S. banks by assets. From this, he concludes that the big banks don't get a subsidy, but actually pay more for their bigness.

Problem is, Levine's analysis doesn't address the question we posed in our analysis. He is asking whether big banks pay more for their funding than small banks. We were asking what big banks would pay for their funding if they didn't get government support.

This is an important distinction. As Levine rightly notes, big and small banks get their funding from different places. According to the Federal Deposit Insurance Corp., the small banks he chose -- Associated Banc Corp., TCF Financial Corp., First Horizon National Corp. and Zions Bancorporation -- get nearly 70 percent of their funding in the form of federally insured customer deposits, the cheapest type of unsecured financing available. By contrast, Bloomberg and FDIC data suggest that the five largest U.S. banks get less than 30 percent of their funding from insured customer deposits (they get 57 percent of their funding from deposits, only about half of which are insured).

In other words, the big banks use a mix of funding that naturally costs more than the mix used by small banks. Our point is that the big banks would pay even more without government support. Levine's analysis does not refute this.

Getting at the true value of the subsidy is much tougher. The study we cited in our original analysis attempts to do so by looking at two different kinds of credit ratings Fitch issued to the same banks: One that takes government support into account, and one that doesn't. The study then used the differential to estimate the benefit to banks when they go into the market to borrow.

Levine makes the point that it's silly to rely on demonstrably fallible credit-rating companies. Just because Fitch says a bank would have a BBB-minus rating in the absence of government support, that doesn't mean the market agrees.

We're under no illusions as to the reliability of credit ratings. That said, they are more likely to be skewed in favor of the largest banks which, after all, pay handsomely for the ratings. Why would Fitch, or any other rating service, risk being tougher on its customers than it needs to be? Indeed, Bloomberg's proprietary default-risk model, which is driven by market data, deems the largest U.S. banks more likely to renege on their obligations over the next year than Fitch's ratings imply. This means we might have lowballed the size of the big banks' subsidy.

Again, there's no perfect way to do the analysis. The broader point -- with which Federal Reserve Chairman Ben Bernanke and other senior officials appear to agree -- is that the subsidy is too big and should be ended.