Why should the government buy toxic assets?

The current idea is to establish an "Aggregator Bank" that will munch up the mortgage-backed securities that banks can't, or won't, sell. This is, of course, the original and much-maligned purpose of TARP. Skeptics abound. Here's Paul Krugman:

Financial institutions that want to "get bad assets off their balance sheets" can do that any time they like, by writing those assets down to zero — or by selling them at whatever price they can. If we create a new institution to take over those assets, the $700 billion question is, at what price? And I still haven't seen anything that explains how the price will be determined.

I suspect, though I'm not certain, that policymakers are once more coming around to the view that mortgage-backed securities are being systematically underpriced. But do we really know this? And how are we going to ensure that this doesn't end up being a huge giveaway to financial firms?

The problem is not that MBS's are hard to value; they shouldn't be harder to value than anything else. After all, people don't seem to have much trouble valuing the institutions—the banks, for example—that own them, whose shares they trade, albeit at much reduced price these days; so why should they have trouble valuing the assets these institutions own? The problem is not one of difficulty of valuation, but of the incentives of the banks. Banks apparently hold onto their MBS's because they would become insolvent or violate minimum capital adequacy requirements if they traded them for paltry sums of cash and/or marked them down to the market value that would be revealed as a result of trading. In the meantime, the banks can hope that the securities will appreciate either because the economy improves or the government intervenes.

If all this is true, the Aggregator Bank will be able to buy up the MBS's only by "overpaying" for them in the sense of paying more than the expected value of the stream of payments to which the holder is entitled—which the market says is lower than the banks claim. The Aggregator Bank has to pay the banks a premium that will reflect the loss of option value for the shareholders—whose shares would otherwise be worth zero rather than the value of the upside that would come with an economic recovery or other positive change in circumstances. This need not involve enriching bank shareholders, as Krugman suggests, though it could if done poorly—if the Aggregator over-overpays.

What is gained by this exercise? Not increased certainty or the discovery of the "real" value of the MBS's. It would have to be—if the intervention were to make sense at all—that the MBS's are worth more aggregated in the hands of the Aggregator than they are in the hands of banks and other investors scattered around the world. How could this be the case?

To see how, consider some of the litigation that has erupted as a consequence of the subprime crisis. This excellent blog, by Isaac Gradman, provides some examples. Debtors and attorneys general are suing loan originators like Countrywide for predatory lending practices, and winning settlements. Under the terms of these settlements, the mortgage loans are modified, with principal and interest reduced. The problem is that Countrywide is now the loan servicer for these loans, so if it agrees to lower payments, as it has, the holders of MBS's, not Countrywide, incur the loss. Can it do this? A definite maybe! Everything depends on the terms of the contracts between the loan servicer and the MBS holders. The contracts usually say that the loan servicer has to repurchase securities if the underlying loan was predatory, but there appears to be plenty of ambiguity, and indeed Countrywide's own obligations under the settlement seem to be contingent on its contracts with the investors. And even when predatory lending is not in issue, it is often in everyone's interests to renegotiate a mortgage loan when the debtor would otherwise default—because the house typically loses fifty percent of its value in foreclosure and would be worth more if the debtor gets to keep it.

But for every loan that has to be renegotiated, there are hundreds or thousands of investors who own securities that share in the proceeds of the loan, and each one of those investors—or a coalition of them if, as I suspect, the contract provides for a voting system—has the de facto right to block the loan renegotiation if they object to it. Even if, in the end, they lose, for the time being they can litigate, and little will happen until all the underlying legal issues are resolved. This creates a massive collective action problem, akin to corporate insolvency, the reason why we have bankruptcy law in the first place. The difference is—and this is crucial—that in your average corporate bankruptcy, where you have thousands of creditors and shareholders and other interested parties, millions or billions of dollars worth of assets will be at stake. Bankruptcy judges can twist arms, ensuring that too much value is not lost while parties squabble. Here, we again have thousands of interested parties, all with de facto "votes," but now all that is at stake is one person's home. MBS holders might, in theory, gain if loans are renegotiated but probably believe that since they can't, as a practical matter, guard their interests in millions of separate loan renegotiations, they do best by blocking all of them—especially in the current political climate that sympathizes with homeowners. It is hard to imagine a more inefficient system.

Two bills try to address this problem. A bankruptcy reform bill would enable bankruptcy judges to cram down mortgage loans in Chapter 13. MBS holders simply would have no way to block loan renegotiations that reduce the value of their securities. This is not necessarily a bad idea, but bankruptcy is always slow and costly.

Barney Frank's TARP-2 bill would require loan servicers and holders of mortgage-backed securities to revise the terms of their contracts. As Steven Davidoff describes it:

Significantly, the bill also abrogates the mortgage-backed security service agreements between servicers and investors in M.B.S.'s. Many of these agreements had required that in the case of a loan modification, the servicer was required to repurchase the loan. Instead, the bill voids these privately negotiated contractual provisions and allows these servicers to freely modify these loans if (i) the property is owner-occupied; (ii) default has occurred or is reasonably foreseeable and (iii) the servicer reasonably believes in good faith that the recovery on the principal obligation of the remodification will exceed the anticipated recovery on a foreclosure on a net present value basis.

This provision isn't as crazy as Davidoff thinks it is. If (iii) is really correct, and if individual MBS holders might hold out for a better deal rather than consent to loan modifications under the terms of the contract, then the bill solves the collective action problem. The difficulty is that all of these mushy words ("reasonably foreseeable," "good faith") have uncertain meaning, and can be exploited by the servicer to pummel the MBS holders. Servicers will do so to keep the TARP funds coming, though I hope the government realizes that as the price of MBS's fall, the government will have to send even more funds into the banks that own them, unless of course the whole crazy scheme works and the price of MBS's rises because of the reduced risk of hold-out.

Anyway, the Aggregator is an improvement on this scheme. Instead of trusting servicers to renegotiate wisely with debtors, with servicers acting as agents for unhappy MBS holders, the government steps into the shoes of the MBS holders and dictates the new terms of the mortgages. The theory behind the Aggregator must be that if the government owns all (or, anyway, enough) of the MBS's, then any remaining private MBS holders cannot hold out in negotiations to modify loans. No need to force loan servicers and MBS holders to renegotiate and no more litigation between loan servicers and MBS holders. As the primary owner of MBS's, the government will voluntarily ease the mortgages simply by waiving or relaxing its own contract rights. The theory here is not the TARP-1 view that if the government starts buying, the MBS market will be jerked back to life; it is that the government will break logjams caused by the dispersion of control rights among zillions of investors. Life is so much easier when the government owns everything.

To sum up, it's not that MBS's are being "systematically underpriced," in Krugman's words. It is that they are worth more together than apart, and only the government is big enough to own all (or enough) of them. That, anyway, is the only way I can make sense of the various reform proposals. If someone has a better theory, I am all ears.