James Surowiecki and Me

Back when I had time to read The New Yorker, I was a big fan of James Surowiecki. I would always look for his column; if it was there, it was usually the first thing I would read. Unfortunately, he’s no fan of mine.

Surowiecki makes three points about our recent long post on nationalization:

1. If the government were to take over a large bank like Citigroup, it would not be able to sell it into the private sector quickly, but would most likely own it for several years, which constitutes nationalization.
2. Recent U.S. history, by which he means the S&L crisis, shows that the right strategy is “exercising regulatory forebearance, cutting interest rates sharply (which raises bank profit margins), and helping the banks deal with their bad assets” – not bank takeovers.
3. We were misleading in citing the IMF’s $4.1 trillion number instead of the lower $1.1 trillion number for U.S. financial institutions. “I assume they used the $4.1 trillion number because it’s much scarier, and offers a much gloomier picture of the state of the U.S. financial system. Unfortunately, it also offers a much more misleading picture of the system.”

Sigh. I guess it’s impossible to make everyone like me.

I’ll take the points in reverse order.

3. I plead partially guilty to this one (I wrote that section of the post). $4.1 trillion is the IMF’s aggregate estimate of all writedowns by all financial institutions globally. To be honest, I used it because that was the number I remembered, and I was writing fast and late at night. (I do this in my spare time, remember?) The point I really meant to make was that the IMF’s estimate was going up, because the problem had spread into all sorts of lending. You’ll note that I didn’t actually compare the $4.1 trillion to any other number – now that would have been misleading.

When I dealt more specifically with U.S. bank capital levels in another post, I did use the relevant IMF number: $275-500 billion in capital needs. $275 billion is a much bigger number than $75 billion, the bottom-line total of the famous Table 3 of the stress test results. Surowiecki attempts to deal with this in his post called “The Fed and the I.M.F. Agree,” where he reconciles these two numbers. His reconciliation is correct as far as it goes. However, the stress tests were supposed to be for a “more adverse” scenario, meaning more conservative than the expected scenario; the IMF numbers, by contrast, are their expected scenario (see mainly PDF pp. 27-28 and 32-34). Who’s right? Well, Calculated Risk has a chart showing clearly that unemployment is already running slightly worse than projected in the “more adverse” scenario (see the second chart in that post), indicating that the “more adverse” stress test scenario/IMF expected scenario (which are similar, as Surowiecki notes) should be thought of as an expectation, not a conservative forecast. So if you believe that the stress tests were supposed to buffer against the possibility of a worse-than-expected outcome, they aren’t doing it.

There is another major difference between the stress tests and the IMF that deals not with the forecasting question, but with the capital adequacy question. The stress tests calculate capital requirements as a percentage of risk-weighted assets, while the IMF uses total assets (confusingly defined as “total assets less intangible assets” (PDF p. 34, note 39)), which ordinarily will be significantly higher (risk weights are generally less than or equal to one). So 4% of one does not equal 4% of the other. (The IMF’s $275-500 billion range reflects 4% and 6% thresholds.) Citigroup, for example, had total assets, excluding intangibles, of $1,897 billion as of December 31 (the snapshot used by both the stress tests and the IMF), while according to the stress tests it had risk-weighted assets of $996 billion. Who is right I will leave to others to debate, but the stress tests are allowing banks to get by with much less capital than the IMF report implies they should.

2. I am confused by Surowiecki’s interpretation of recent U.S. history. The “regulatory forbearance” he mentions happened early in the 1980s when, among other things, thrifts were allowed to invest in a broader set of assets and were allowed to offer higher interest rates to depositors, deposit insurance ceilings were raised, and capital adequacy requirements were relaxed for thrifts facing insolvency. This may have helped banks survive but, according to a later FDIC report, it also postponed and amplified the later crisis: “With respect to commercial mortgage markets, this legislation set the stage for a rapid expansion of lending, an increase in competition between thrifts and banks, overbuilding, and the subsequent commercial real estate market collapse in many regions.” The early 1990s saw precisely the opposite; the FDIC Improvement Act of 1991, for example, limited regulatory discretion in dealing with struggling institutions.

Forbearance can work, but it is not a cure-all. The FDIC report later contrasts beneficial and harmful forbearance programs, but it criticizes large-scale forbearance programs in no uncertain terms:

Longer-term, wholesale forbearance as practiced by the FSLIC was a high-risk regulatory policy whose main chances of success were that the economic environment for thrifts would improve before their condition deteriorated beyond repair or that the new, riskier investment powers they had been granted would pay off. The latter type of forbearance, which the FSLIC adopted against the background of a depleted insurance fund, is widely judged to have increased the cost of thrift failures.

In addition, the policies of the 1980s played out in a very different economic environment. The assets in question were primarily loans, rather than the complex securities we are dealing with today. And the global economic climate was considerably better than today, which helped banks earn their way out of their problems.

In any case, the bottom line of the S&L crisis was over 1,600 FDIC interventions between 1980 and 1994. (We’ve had 33 so far this year.) The Resolution Trust Corporation was charged with managing the assets of banks and thrifts that had become insolvent, not “helping the banks deal with bad assets.” One can argue about the lessons of the 1980s – particularly about what they mean for large banks – but it’s by no means clear that the policies Surowiecki highlights forestalled the need to take over banks.

1. I think this is one of the more serious criticisms of government takeover of a large bank – namely, that it would take a long time before it could be returned to the private sector.  But that doesn’t mean prolonged political control over lending decisions. The important thing about a takeover is that you can clean up the balance sheet, for example by transferring the toxic assets to a separate entity, without having to negotiate with anyone. Once you’ve done that and the bank is recapitalized (no one is saying this will be free), the government’s stock can be put into a trust with an independent board of trustees with long terms. This would insulate the bank reasonably well from political pressure, since the trustees’ legal obligation will be to run the bank for its own long-term benefit, and to privatize it when possible at the highest possible price for the taxpayer.

There is actually an example of this right now: AIG is majority-owned by the government, but the government’s stock is controlled by three trustees who are independent of Treasury and the Fed. This is why the Treasury is forced to negotiate with AIG like any other private party that is looking out for its own interests. On the other hand, Treasury had zero voting shares in Bank of America back in December – yet it felt able to threaten Ken Lewis with replacement if he didn’t close the acquisition of Merrill. My point is that you get government influence either way, regardless of the legal structure. With Bank of America, the source of the influence was the likelihood that B of A would need assistance from the government in the future. With AIG, the source of the influence was yelling and screaming through the media, plus the fact that AIG also needed additional assistance.

I’m not saying that the AIG situation is perfect, just that it is possible to insulate a government-owned entity from political meddling – as the government found out, to its frustration. The mistake with AIG, I believe, was giving it that independence without having fixed its fundamental problems – the ones that cause it to keep coming back for more money. Were the government to take over a large bank, it should clean up the balance sheet and recapitalize first – without negotiating – and then turn it over to an independent set of trustees.

This “big banks are different” issue is a serious one and worthy of consideration. But it doesn’t necessarily justify the application of a completely different set of principles than the ones that are routinely applied to smaller banks.