More TARP Programs, More Policy by Deal

Back on January 2, the Treasury Department announced something called the Targeted Investment Program. I missed this at the time, along with (according to a quick search – thank you Google Reader!) all of the economics blogs that I read. The press release admitted that this was a program announced after the fact to cover the second Citigroup bailout (the first was under the Capital Purchase Program, the main bank recapitalization plan). In essence, the program says that if Treasury thinks a financial institution is at risk of a loss of confidence, Treasury can invest in it under any terms they want. This is very similar to the Systemically Significant Failing Institutions Program, also announced after the fact (in November) to cover the second AIG bailout, which reads almost identically, except instead of talking about a “loss of confidence” it takes about the “disorderly failure” of a systemically important institution.

This isn’t a power grab by Treasury – they already had this power under the EESA (the main bailout bill passed in October, commonly known as TARP). And I happen to agree that if a systemically significant institution – the kind that whose failure would have a major impact on countless other institutions – is going to fail, it should be bailed out. However, I think these programs have two major failings.

First, they do only the vaguest job of specifying what types of institutions will be bailed out, making it difficult to predict when the government will step in. Shouldn’t the government be able to figure out at this point which institutions are systemically significant, and say what they are instead of periodically relaxing the criteria to let in, say, GMAC? Second, and more importantly, they are completely vague on the terms of such a bailout (as opposed to the Capital Purchase Program, which has predefined terms that happen to be quite generous to participants). This is a problem because of the incentives it creates. If you are a shareholder in a bank that may be in trouble, you cannot be sure whether or not it will qualify for a bailout. And if you happen to run a bank that may be in trouble, you know that if push comes to shove, you can negotiate a deal with Treasury at the last minute by threatening to blow your brains out on their nice carpet.

This is a case where it would be good for Treasury to tie its hands in advance by predefining the terms of a rescue operation (say, type of asset invested in, warrant amount, strike price, governance, executive compensation restrictions, etc.). First of all, it would enable public debate over the terms, instead of the usual second-guessing on Monday morning when Treasury announces the deal it struck on Sunday evening “before the Asian markets open.” Second, it allows Treasury to say, “This is the only deal on the table. Take it or leave it.” It is a commonplace in negotiations that you are better off if you can credibly claim that your hands are tied, because this gives you a valid reason why you simply cannot concede to your counterparty’s requests.

The counterargument will be that each failing institution is different and the rescue has to be tailored to its situation. But I don’t really buy this. The predefined plan could be, to take a simple example, that Treasury will buy as much common stock as is needed to inject the required capital, at a 10% discount to the price on at the previous market close. No matter how much capital a bank needs in a pinch, it can get it under those terms – but the more capital, the more the existing shareholders get diluted, which is exactly as it should be. This plan should have relatively harsh terms; the Capital Purchase Program is the one with easy terms that banks turn to first. Even these terms are better for shareholders than bankruptcy, which means that the bank’s board of directors has a fiduciary obligation to take them if they can’t find private capital and the only alternative is bankruptcy. This is obviously just a simplistic example (maybe convertible preferred stock would be better than common), but I don’t see why certain ground rules like this can’t be defined in advance.

Originally published at the Baseline Scenario and reproduced here with the author’s permission.