Sizing Up the Hole in the Banks: Nationalization is Not Inevitable
The time-honored principle of public sector intervention to support banks is that of Walter Bagehot, who wrote in 1873 that the central bank should provide lender of last resort (LLR) lending to solvent banks in a panic, but should not lend to insolvent banks. However, because of the too-big-to-fail effect in which failures of large institutions cause systemic damage (as demonstrated by Lehman Brothers), public policy in the current crisis has tended to support the key financial institutions without making too fine a distinction between whether they are illiquid or insolvent.
Nouriel Roubini has estimated that credit losses for US institutions could reach $3.6 trillion, of which half would be in banks and broker dealers. He observed that if so, the US banking system would be "effectively insolvent."1 Paul Krugman wrote a column coyly describing a "hypothetical" bank ("Gothamgroup") with $2 trillion in assets, $100 billion in capital, and $400 billion in troubled assets that might be worth only $200 billion, leaving it a "zombie bank" with negative net worth but still $20 billion in market capitalization only because of the expectation of a government bailout.2 The implication is that some large US banks may already be insolvent. If so, because they are too big to fail, the Bagehot principle would not apply and instead they might need to be nationalized.
The key question for the solvency of the major banks is whether their core capital exceeds their prospective losses. From mid-2007 through the end of the third quarter of 2008, large US banks had write-downs of $413 billion and raised new capital of $399 billion.3 Through this period their new capital thus approximately offset their losses. More recently, however, the plunge in bank shares has ruled out the raising of new capital through stock issuance. Instead, new capital has had to come from the public sector.
Both the International Monetary Fund and Goldman Sachs have recently estimated cumulative prospective credit losses on US residential real estate, commercial real estate, credit cards, automobile loans, and commercial and industrial loans at about $2.1 trillion, of which about half is from residential real estate.4 The Goldman Sachs authors calculate that the share of US banks in these losses will amount to almost $1 trillion. The implication is that the US banks have another $600 billion of losses that will need to be recognized, beyond amounts already declared by the end of the third quarter.
A bank's net worth can be broadly approximated by its Tier 1 capital.5 This is mainly the sum of original purchases of stock issued plus cumulative retained earnings. For the 17 large US banks (including the large investment banks now converted to bank holding companies), which represent about 90 percent of US banking assets, at the end of the third quarter of 2008 total Tier 1 capital amounted to $588 billion.6 They subsequently received approximately $180 billion in new Tier 1 capital from the Troubled Asset Relief Program (TARP) in the form of preferred shares, boosting their net worth to about $770 billion.
The further prospective losses would thus eliminate about 70 percent of the net worth of the large banks.7 The implication is that the large US banks as a group are not yet insolvent, although they are skating on thin ice. The situation could be worse for individual, relatively more exposed banks; and it could be much worse for the large banks as a group if the Roubini estimate turns out to be right.
To the extent that public support is given to insolvent banks too big to fail, the result is to "socialize" the losses. Because there could be sizable efficiency losses from public rather than private operation of the banks, some amount of loss socialization may be warranted to avoid nationalization. The large banks have assets of about $10 trillion. Suppose they were all nationalized. Suppose that lesser efficiency of government management and diversion of lending toward lower-return politicized objectives were to cause a reduction of just 1 percent in annual return on assets. The consequence would be a social loss of $100 billion per year. If the nationalization lasted a decade, the total social cost would be on the order of $1 trillion.
This thought experiment suggests that it might be socially worth somewhere on the order of $1 trillion to keep the large US banks under private ownership and management. The Obama administration's effort to avoid nationalization seems to be premised on such a view. It is reportedly on the verge of launching either a "bad" (or "aggregator") bank that would buy up questionable assets (the original TARP idea); or guaranteeing additional large blocks of such assets (as it did for about $300 billion for Citigroup and $120 billion for Bank of America); or doing a combination of the two. In either case the contingent loss to the government would be undertaken because the alternative social loss from nationalization would be even larger.
The bottom line is that it will require some form of troubled asset alleviation, combined with further recapitalization, to consolidate the process of shoring up the systemically important banks. After recognition of pending losses, to restore Tier 1 capital to 4 percent of the $10 trillion in assets would require recapitalization on the order of $170 billion.8 Providing additional public sector capital injections on this scale through the TARP and/or its successor could be a bargain if the illustrative $1 trillion is in the right ball park for the social inefficiency costs of major bank nationalization.
1. Bloomberg, January 20, 2009.
2. New York Times, January 23, 2009.
3. Bloomberg, December 15, 2008.
4.International Monetary Fund, Global Financial Stability Report: Market Update, January 28, 2009; Hatzius, Jan, and Michael A. Marshoun, Home Prices and Credit Losses: Projections and Policy Options. Global Economics Paper No. 177 (New York: Goldman Sachs Global ECS Research, January 13.)
5. In contrast, so-called Tier 2 capital includes long-term subordinated debt, representing resources that may provide a handy war-chest but ultimately must be repaid.
6. Calculated from quarterly reports for 17 large US banks.
7. Applying the 90 percent share of the total, additional losses would amount to $540 billion for the 17 largest banks, or 70 percent of the $770 billion in third-quarter net worth (after enhancement by TARP disbursements).
8. For the 17 largest banks: $770 billion Tier 1 capital at end-third quarter 2008, less $540 billion further losses to recognize, would leave $230 billion net worth. Topping them up to $400 billion would require recapitalization of $170 billion.