Visiting Scholar R. Glenn Hubbard

When Henry Paulson, President Bush's Treasury secretary, first introduced the Troubled Asset Relief Program (TARP) in Congress last September, we cautioned against using government funds to buy mortgages and mortgage-related securities from banks. After the Emergency Economic Stabilization Act was signed into law in early October, the Treasury decided not to buy these assets. Instead, it used the first $350 billion of TARP funds to inject capital first into nine systemically important troubled banks, and later into insurer AIG (as part of a refinancing) and auto makers General Motors and Chrysler.

This approach seems to have achieved (albeit at a high cost for taxpayers) its principal objective of avoiding a massive collapse of the financial system. But it has not yet resulted in an increase in bank lending or the attraction of new private equity to the banking system, both of which are important to reviving the economy. There now appears to be active consideration of using TARP funds to buy "bad assets" from the banks. Major problems with so doing remain.

The central issue is how to price the assets. When the subprime crisis hit in the summer of 2007, the Treasury's first response was to encourage the private sector to create a fund--the so-called Super SIV (structured investment vehicle)--to buy mortgage-related assets. This proposal foundered due to the difficulty of setting a price for these assets, which come in complex and incomparable varieties. If Treasury pays close to par, it is paying far too much. If it pays current prices, no one will sell because of the adverse impact on their capital. If it pulls a price out of a hat, it will be acting arbitrarily.

Initial discussions focused on using a reverse auction with asset holders "bidding" to sell their mortgage-related securities to the Treasury. Such an approach raises significant problems--most significant is the risk posed by asymmetric information regarding the value of these securities. Because the holders of complex and incomparable mortgage-related securities have more information regarding their worth than does Treasury, Treasury is at a huge disadvantage and will likely overpay. Moreover, there will have to be many auctions of very different securities. All of this will take months to execute.

Reportedly, thought is now being given to only buying "bad assets" and putting them in a fund (called a "bad bank") owned by the government. This new variation raises additional problems. First, how should "bad assets" be defined? As the recession deepens, bad assets have multiplied and will continue to multiply from mortgages and mortgage-related securities to many other assets classes, including credit-card portfolios. We see little sense in defining bad assets simply as those that have been already significantly written down. The bank may be more exposed to losses from assets that have not been significantly written down, but could well be.

Further, as the potential class of bad assets expands so does the cost of purchase. Total mortgage-related securities and mortgages held by banks alone are estimated to be $6 trillion, of which mortgage-backed securities are $1.3 trillion. Total bank assets are $16.5 trillion.

Another proposal is to guarantee the value of bad assets rather than buy them. This outcome could be accomplished by a direct guarantee of assets that remained on the balance sheet of banks (or were brought in from outside conduits or SIVs), or into one government-run bad bank.

A version of this approach was used in the second round of TARP financing for Citigroup and Bank of America. In the case of Citigroup, a $306 billion asset pool was created in which Citigroup absorbs the first $29 billion of losses, the Treasury and FDIC jointly fund 90% of the next $15 billion ($5 billion from the Treasury through TARP, and $10 billion from the FDIC), and the Fed finally funds 90% of the remaining losses. In return, the government received $7 billion in preferred stock (with an 8% yield)--$4 billion to Treasury, $3 billion to FDIC. Under this approach, the problem of valuing the assets has been finessed into a problem of valuing the stock.

With more transparency, which Congress and its Oversight Panel would surely demand, that finesse would not work. A conservative estimate puts the value of the Citigroup option--i.e., the potential cost to taxpayers--at $60 billion, taking account of the stock warrants the government received. If one were to repeat this approach for all banks the cost would be as much as TARP. And, if the market for toxic assets were to fall further, the government could easily be responsible for trillions of dollars of losses. Last but not least, having the Fed become the residual risk bearer further undermines the Fed's financial stability and its ultimate independence, a concern recently voiced by the Committee on Capital Markets Regulation.

A more reasonable alternative would be to encourage banks to spin off the toxic assets into separate affiliated bad banks (as under a new reported German initiative). Ownership of these two entities would be allocated pro rata to all the financial investors as a proportion of the most updated accounting value of these assets. So a bank with $30 billion of bad assets and $70 billion of good assets will see its debt divided 30%-70% and its equity divided 30%-70%. Each $100 debt claim will become a $30 debt claim in the bad bank and a $70 debt claim in the good bank. The same would be true for equity. To limit the exposure of the FDIC to the bad bank, insured deposits and FDIC-guaranteed debt should remain in the good bank to the extent there are sufficient matching good assets. In addition, off-balance-sheet derivative contracts remain off balance sheet for the good bank to avoid the possibility that the failure of the bad bank would create systemic risk. Furthermore, convincing evidence would be needed before guaranteeing any of the liabilities of a bad bank because ordinarily a bad-bank failure should not result in systemic risk.

An alternative would be for the government to facilitate the injection of new, private-equity capital into banks by eliminating regulatory restrictions, such as bank ownership by private-equity or commercial firms, and providing protection against loss or dilution if there were to be subsequent government intervention. A subsidy for this private risk capital could even be given. As long as private capital holds most of the risk, it will certainly be allocated more efficiently than government money, minimizing the taxpayer cost of any subsidy. This idea would have to be more fully explored.

Whatever is done with respect to still solvent institutions, banks with dangerously low, or no or negative capital, should be taken over by the government through already established FDIC procedures, such as bridge loans. Just as with the Savings and Loan crisis of the 1980s, there is a significant risk that shareholders (and managers) with nothing to lose will roll the dice and lose even more. Existing shareholders with no equity should not benefit by government support. Further, a government takeover (just like a corporate bankruptcy) permits the restructuring of bank debt. As the S&L debacle shows, a decision worse than a nationalization of the banking sector is a nationalization of the losses, which still leaves the gains in private hands.

With still solvent institutions, there is no affordable and workable plan that will by itself assure that these banks will start lending and that private capital will return. We have stabilized the financial system against massive collapse, which is probably all the government can do. While we hope some more improvement may come from the bad bank or private-equity solutions outlined above, the road to further recovery of the financial system now lies principally with the economy's recovery and the success of the fiscal stimulus.

R. Glenn Hubbard is a visiting scholar at AEI. Hal Scott is a professor at Harvard Law School. Luigi Zingales is a professor at the Graduate School of Business at the University of Chicago.