By James Kwak
Andrew Lo’s review of twenty-one financial crisis books has been getting a fair amount of attention, including a recent mention in The Economist. Simply reading twenty-one books about the financial crisis is a demonstration of stamina that exceeds mine. I should also say at this point that I have no arguments with Lo’s description of 13 Bankers.
Lo’s main point, which he makes near the end of his article, is that it is important to get the facts straight. Too often people accept and repeat other people’s assertions—especially when they are published in reputable sources, and especially especially when those assertions back up their preexisting beliefs. This is a sentiment with which I could not agree more. One of the things I was struck by when writing 13 Bankers was learning that nonfiction books are not routinely fact-checked (Simon and I hire and pay for fact-checkers ourselves). As technology and the Internet produce a vast increase in the amount of writing on any particular subject, the base of actual facts on which all that writing rests remains the same (or even diminishes, as newspapers cut back on their staffs of journalists).
I’m not entirely convinced by Lo’s example, however. He focuses on a 2004 rule change by the SEC. According to Lo, in 2008, Lee Pickard claimed that “a rule change by the SEC in 2004 allowed broker-dealers to greatly increase their leverage, contributing to the ﬁnancial crisis” (p. 33). That is Lo’s summary, not Pickard’s original. This claim was picked up by other outlets, notably The New York Times, and combined with the observation that investment bank leverage ratios increased from 2004 to 2007, leading to the belief that the SEC’s rule change was a crucial factor behind the fragility of the financial system and hence the crisis.
Not so fast, Lo says (pp. 34–35):
While these “facts” seemed straightforward enough, it turns out that the 2004 SEC amendment to Rule 15c3–1 did nothing to change the leverage restrictions of these ﬁnancial institutions. In a speech given by the SEC’s director of the Division of Markets and Trading on April 9, 2009 (Sirri, 2009), Dr. Erik Sirri stated clearly and unequivocally that “First, and most importantly, the Commission did not undo any leverage restrictions in 2004”. . . .
[T]wo aspects of this story are especially noteworthy: (1) the misunderstanding seems to have originated with Mr. Pickard, a former senior SEC oﬃcial who held the very same position from 1973 to 1977 as Dr. Sirri did from 2006 to 2009, and who was directly involved in drafting parts of the original version of Rule 15c3–1; and (2) the mistake was quoted as fact by a number of well known legal scholars, economists, and top policy advisors.*
However, the statement that “the Commission did not undo any leverage restrictions in 2004” is true only in a very narrow sense. Lo is correct that the allowable leverage ratio did not change. He is also correct that the real issue for broker-dealer firms is not a traditional leverage ratio (assets to equity), but net capital (a measure of financial position). But the rule did change the way that broker-dealers were allowed to calculate their net capital; in other words, it changed the way you calculate the denominator. In fact, Sirri concedes this (quoted in Lo, p. 34, note 26.):
The net capital rule requires a broker-dealer to undertake two calculations: (1) a computation of the minimum amount of net capital the broker-dealer must maintain; and (2) a computation of the actual amount of net capital held by the broker-dealer. The ‘12-to-1’ restriction is part of the ﬁrst computation and it was not changed by the 2004 amendments. The greatest changes eﬀected by the 2004 amendments were to the second computation of actual net capital.
You can read the rule (and I did, while writing 13 Bankers): the SEC rule change is at SEC release 34-49830, and the current rules are here. In particular, Rule 15c3-1(c)(2) defines net capital as “net worth” subject to various adjustments. Paragraph 15c3-1(c)(2)(vi) says that you have to take deductions (“haircuts”) for different types of securities; conceptually, it’s like the risk weightings for Basel capital adequacy ratios.
The 2004 rule change said that certain broker-dealers could stop using the haircuts in 15c3-1(c)(2)(vi) and, instead, could use their own internal mathematical models to calculate haircuts, according to the rules in what is now Appendix E to Rule 15c3-1. The Summary to the rule change (p. 34428) says, “This alternative method permits a broker-dealer to use mathematical models to calculate net capital requirements for market and derivatives-related credit risk.”
And the whole purpose of the rule was to allow broker-dealers to take smaller deductions when calculating net capital. It’s also in the Summary (p. 34428):
These amendments are intended to reduce regulatory costs for broker- dealers by allowing very highly capitalized firms that have developed robust internal risk management practices to use those risk management practices, such as mathematical risk measurement models, for regulatory purposes. A broker-dealer’s deductions for market and credit risk probably will be lower under the alternative method of computing net capital than under the standard net capital rule.
(We quoted the first sentence of that passage as the epigraph for chapter 5 of 13 Bankers). How much smaller? Well, the SEC worked that out, too, when estimating the “benefits” of the rule change (p. 34455):
A major benefit for the broker-dealer will be lower deductions from net capital for market and credit risk that we expect will result from the use of the alternative method. . . . In the Proposing Release, we estimated that broker-dealers taking advantage of the alternative capital computation would realize an average reduction in capital deductions of approximately 40%. We estimated that a broker-dealer could reallocate capital to fund business activities for which the rate of return would be approximately 20 basis points (0.2%) higher.
In summary, a broker-dealer could increase its net capital, for the same portfolio of assets and liabilities, by switching to the new calculation method. Since it now had excess net capital in the broker-dealer business, its holding company could transfer capital out of the broker-dealer and into other businesses. So without raising more capital, it could now expand its operations in those other businesses, and hence its balance sheet.
How much did this rule change contribute to rising leverage ratios between 2004 and 2007? I don’t know; perhaps not that much. In any case, while high leverage contributed to the fragility of the big investment banks and hence the fragility of the financial system, the banks could also have done plenty of damage to the world without high leverage—simply by manufacturing toxic securities and selling all of them to investors (instead of eating their own dog food, as Citi and Merrill notably did). So I would not argue that the SEC rule change caused the financial crisis all on its own. But I also would not say that the rule change did not affect leverage at all.**
Now let’s go back to the “mistake” that people allegedly made describing this rule change. This is what Pickard said, as quoted by Lo (p. 33, emphasis added by me):
[Before the rule change] the broker-dealer was limited in the amount of debt it could incur, to about 12 times its net capital, though for various reason broker-dealers operated at signiﬁcantly lower ratios. . . If, however, Bear Stearns and other large broker-dealers had been subject to the typical haircuts on their securities positions, an aggregate indebtedness restriction, and other provisions for determining required net capital under the traditional standards, they would not have been able to incur their high debt leverage without substantially increasing their capital base.
I’m not sure there’s a mistake here, except perhaps for the word “substantially,” since I don’t know how big an impact this rule change had. There might be one, but I don’t see it.
Now, it’s entirely possible that other people picked up the story, repeated it, and added their own errors. It’s also possible that the rule change has been blown entirely out of proportion. That’s one of Lo’s arguments (pp. 34–35): leverage was as high in 1998 as in 2007, so what’s the big deal? That’s the essence of this chart from the Economist article:
But two things stand out for me here. One is that leverage did go up after 2004 for every bank. Second, 1998 was when we had the LTCM crisis, so for me a return to 1998 leverage levels is bad—not a justification for 2007 levels.
Still, at the end of the day, we have correlation without causation: the rule change very well might have contributed to higher leverage, but we can’t tell how much. It was certainly not the sole cause of the financial crisis. But if I were looking for a clear factual “mistake” that got picked up and circulated by the press and academics, I would not have chosen this one.
* I should point out here that Lo does not include Simon and me among the people quoting this “mistake.” We did refer to the SEC rule change—probably because we referred to it (correctly) as a change in the way net capital was calculated, not an increase in the allowable leverage ratio. So I have no dog in this fight.
** On p. 34, note 26, Lo says, following Sirri, that the 2004 rule change was mainly about increasing regulatory supervision: “By subjecting themselves to broader regulatory supervision—becoming designated “Consolidated Supervised Entities” or CSEs—these U.S. ﬁrms would be on a more equal footing with comparable European ﬁrms.” Lo quotes Sirri saying that the rule change added “an additional layer of supervision.” We all know how that ended up—with the SEC’s Inspector General calling the CSE program a complete failure—although that’s neither here nor there for this blog post.