2017 Financial Markets Conference—Lawrence Summers Keynote—On Market Valuation and Bank Regulation
Lawrence Summers, Harvard's Charles W. Eliot Professor and President Emeritus, asks: are major financial institutions significantly safer than they were before the crisis?
Lawrence Summers: Thank you very much for those generous words. You were kind enough to describe how I, over the course of my career, have done a certain amount of moving back and forth between academic life and experience in government. When I first moved from being a Harvard professor to being at the Treasury, people asked me, "Well, what was fundamentally different about being in government?" And I said, "There are two important answers to that question." The first is that, as a Harvard professor, the single worst thing you can do is to sign your name to something you did not write yourself. On the other hand, as a subcabinet official, it was a mark of effectiveness to do so as frequently as possible.
The other answer I gave was, when you're a researcher and you're working on a problem, and you find that the problem is intractable—there is a clear solution: you work on a different problem. [laughter] And when you're involved in public policy, you usually do not have that choice.
Then, time went on and I was fortunate enough to rise in the Treasury Department, and after a time I left the Treasury Department, and returned to Harvard as its president. And people asked me, "What is different about being Harvard's president than working in Washington at the Treasury Department?" And I gave an answer that, in retrospect, was breathtaking in its naïveté. I said, "Washington is so political." [laughter] I don't think I need to follow on the logic of that.
Henry Kissinger had a certain point when he said of academic politics that the fights are so vicious because the stakes are so small. [laughter] And I tried during that time to live up to Clark Kerr's admonition that there were three things that a university president had to do to succeed, three things he had to provide: ample parking for the faculty, ample football victories for the alumni, and ample sexual opportunity for the students. [laughter]
Anyway... [laughter] Let me make a more a more serious comment that, in a sense, frames the kind of talk I'm going to give today. I have always felt that, when in Washington, the right thing for me to focus my energies on, to the extent I can, is towards finding ways forward to implement solutions where we know what the solution is, and where there is widespread agreement as to what the solution is. And there are many problems like that—long-term budget deficits is an example, the deal between the Israelis and the Palestinians. Everyone knows what the deal is, the challenge is to get it done.
On the other hand, I have always felt, when I'm in academia, the right role is not to try to figure out how to play the inside game, where you're not in a position to do that on the outside, but is to step back to ask larger questions, to challenge conventional wisdoms, and to work on issues where it's not so clear what the right way forward is. And it's in that spirit that I want today to address the question of increased capital and financial stability, some doubts, and raise what I think is a fairly serious challenge to the general conventional wisdom in the official and private communities regarding our success in achieving financial stability through what has happened over the last eight years. I do not so much mean to suggest the wrongness of much of what has happened, as the insufficiency of much of what has happened with respect to the achievement of financial stability.
And I should say at the outset that my remarks today are very heavily based on joint work with Harvard graduate student Natasha Sarin that some of you may have seen, and it was published in the Brookings papers on economic activity a few months ago.
I think it is fair to say that it is conventional wisdom, in the official community—this part of it would be agreed in much of the private sector—that a great deal has been done that has increased financial stability, that has addressed the "too-big-to-fail" problem, particularly with respect to large institutions. And that has put our financial system on a stronger and more sustainable basis. And that the center of that effort, though by no means the totality of that effort, has been a dramatic increase in the capital that financial institutions hold. And it is felt that those big increases in capital put these institutions on a substantially stronger footing than they were previously, give them much greater capacity to absorb losses, and act as a preventative, with respect to the substantial risks associated with leverage.
I've chosen here two of many quotations from the regulatory community. In their exuberance, one can find senior members of the regulatory community speaking of increases in capital—in required capital—by a factor of 5 to 10, and describing a very dramatic change. This is of course not the only change that has been put in place—there have been changes with respect to liquidity, changes with respect to resolution, changes with respect to consumer protection, changes with respect to supervisory practice—but I think it is fair to say that increased capital has been at the center of policy concern.
What I want to do in the first part of this talk is call into question how important, and how significant, that achievement has been. Let me ask you to consider three separate facts, each of which should, I think, give pause about the "far more capital, far safer" system syllogism which is so central to current conventional wisdom.
First, if you look over time or across space, the capital ratio is distressingly unpredictive of failure. It is not the case that the institutions that failed during the 2007–2009 maelstrom had lower capital going into that crisis than the institutions that stayed afloat. Nor is it the case, over the longer term, that places and times where capital ratios were higher, systematically, had less failure than places where capital ratios were lower.
It should also give pause that, as Jeremy Bulow and Paul Klemperer have pointed out, in the vast majority of instances where banks fail and are resolved, the cost borne by the government, borne by the FDIC, is in excess of 15 percent of their assets. If their capital were really zero, in an effective sense, then there would be no cost. If their capital were really 10 percent, then they wouldn't be failing—at least, for reasons of lack of capital. The fact that institutions—all of which, on the month before they failed, were judged to have adequate capital, in fact had ex-post capital of negative 15 percent—should give pause about capital measures.
So first, there's just reason to be suspicious that these capital measures have the predictive power that they are supposed to.
Second, comes a basic application of straightforward financial theory—and I'll do this part of the talk relatively concisely. Imagine that an institution, any company—a bank, or anything else—becomes less levered. What would you expect? You would expect that its volatility of its equity would decline. That's, after all, what leverage is about, that's why they call it leverage. You would expect that its expected future volatility would decline. You would expect that its sensitivity to shocks would decline. You would expect that the required return on its risky securities would decline, as their risk went down. I would ask you to consider—and this is documented fairly extensively in our paper—that essentially none of those predictions are borne out in the data.
Remember, the claim is that there has been a large increase in capital. If there has been a large increase in capital, there should be a large decrease in volatility. There should be a large decrease in expected future volatility, as measured in implied volatility and option delta (∆). There should be a decline in beta (β), the sensitivity to the stock market. There should be a decline in the credit default spread. There should be a decline in the price-earnings ratio. And there should be an increase in the preferred stock price, reflecting greater safety and a reduced risk that that preferred stock would be unable to be paid because of a larger capital cushion.
There are plenty of grounds for quibbling and cobbling with this table. You can argue about what the right definition of the pre-crisis period is. You can argue about exactly how to measure volatility. There's much that you can add, and to some extent we have, and others will examine in the details of the data. But the conclusion that any of the major implications of the "increased capital, reduced leverage" view are borne out in the data is just not there. You can make as strong a case for the view that volatility measures have increased as you can for the view that volatility measures have decreased. Yes, it is no doubt true that if I had and we didn't quite get it—the data for 2016—they would be somewhat more favorable than this data. But you will not find anything that corresponds to a halving of leverage, in any data that's got anything to do with volatility.
Here's a third implication of the "increased capital" view, and I think it—the fact I'm about to talk about—helps to, as I'll argue, helps to illuminate this picture. Suppose, instead of looking at regulatory accounting measures of capital, you instead look at the market value of equity, and you divide it by risk-weighted assets. The market value of equity is a measure of the market's valuation of the capital that a firm has, including as capital the firm's claim on its future earnings. It's actually a significantly exaggerated measure, or a potentially exaggerated measure, of the firm's capital, because of the equity aspect, because of the option aspect. If the firm goes under, the equity owners don't have to pay. If the firm does well, the equity owners get the benefit. So the equity is actually an option on the total assets of the firm, and therefore is an overstatement of the actual value of its capital.
So what would you expect to see if, in an effective sense, a substantial increase in capital had been achieved? You would expect to see that if you divided the market value of equity by the total value of firms' risk-weighted assets, you would expect to see that that number had gone significantly up. What's the fact? The fact is that for—and, by the way, I'm stressing the big six banks, but essentially everything I'm talking about is a feature of big banks in the United States, mid-sized publicly traded banks in the United States, and global institutions, and institutions around the world. What you see is that the ratio of the market value of equity to risk-adjusted assets has come down from the pre-crisis period, not gone up.
Now, yes. It can be argued that the pre-crisis period had an element of bubble—though you might think that a bubble would go with increased volatility, not with decreased volatility. And so, again, there's plenty of room to argue with the details of these calculations, but I think you will have a difficult time making the case that there has been a large increase in capital of any kind that markets have noticed and judged likely to have a salutary effect. One way to look at that is to observe the rather substantial—in excess of 5 percent—yield that bank-preferred stocks are carrying. If, really, they had been effectively insulated by a common equity buffer, investors would not be demanding a 5 percent yield in the current environment to get them to hold preferred stock.
So if I don't convince you of anything else today, I'd like to unsettle you with respect to the conclusion that the system is superbly capitalized today and is vastly better capitalized than it was prior to the crisis. I would suggest in this regard, as Jeremy Bulow did in commenting on our paper, that if the Federal Reserve's stress tests have concluded that the unemployment rate could double, all real estate could decline by 30 percent in price, the stock market could nearly halve, and that no major bank would suffer to a degree that would require any kind of major regulatory action, that that may be a finding that reveals more about the stress test methodology than it reveals about the health of the financial system.
And if there's anyone in this room who is highly confident that the unemployment rate could go back to double digits, real estate prices could suffer more than they did in 2008, the Dow Jones average could halve, and that there wouldn't be any major issues in the banking system—well, I see the world very, very differently than you do. I am not sure that it is necessary for a healthy, functioning financial system that that test be met, but the claims that are met—the claims for that system—seem to me highly surprising, in light of this market evidence.
What is it that is going on? I think there are three elements that are important in explaining this, the last of which is by far the most important. But first is that the comparison may be misleading, to the extent that markets were not measuring things and appreciating risks properly during the pre-crisis period. This was a point that was emphasized when our paper was presented, and Natasha and I went back and excluded the data from the pre-crisis period after 2005 and compared the present with the period from 1995 to 2005—a period that included the sharp decline during the 2001 recession. Even doing the comparison that way, you do not find market measures suggest dramatic declines in volatility or dramatic increases in measured capital.
Second possibility—which is, I think, of substantial importance at particular moments—is that measures of regulatory capital—which are, after all accounting measures—are not really to be thought of as fully accurate measures of economic capital. At some level, this must be right at many moments. The head of the SEC opined a week after Bear fell that it had had a double-digit capital ratio the day before it fell. The institutions that collapsed in 2008, without exception, saw very dramatic declines in our preferred measure, the market value of their capital prior to their collapse, but in most cases did not see large declines in regulatory capital. When I see institutions in Europe trading at 30 percent of their book value, I ask myself whether their capital is being measured accurately and whether their assets are being appropriately valued.
But I think the most important aspect of this is an aspect that the regulatory community increasingly appreciates and is part of the motivation for the stress testing methodology, and that is taking a dynamic view of capital. The capital, really, of an institution reflects the excess of its current assets over its current liabilities, and the expectation of the future profits that it will earn—that afford it value, afford it the opportunity to borrow—offer assets on which lenders to that institution can rely. If there has been a substantial decline in the franchise value of institutions, as the observation of pervasive declines in the ratio of price to book value suggests, then capital in a dynamic sense has been substantially eroded.
So what I believe has happened is that there has been an increase in regulatory capital that has coincided with a very substantial decline in franchise value, leaving total capital, in an ultimate economic sense, not very different than it was before. Now, there are probably quite substantial benefits from capital's form changing from being future profits to being current assets, and so I believe that the run risk of the kind we saw in 2008 has indeed been attenuated. But that is a rather different thing than a suggestion that there has been a substantial increase in the effective capitalization of the system.
What are the implications of all of this for the officials here [who] work as regulators and supervisors? I would just conclude by making four observations. First, it's appropriate to take a dynamic view of capital that recognizes the role of future profitability and the role of future inability to make profits. Regulators should, along with studying measures of regulatory capital, study measures of the market value of capital. To be clear, it would be entirely inappropriate to focus only on market measures of capital. To do so would be to be at the risk of declaring "all is well" at the worst moment, simply because the market has decided to have a bubble.
On the other hand, failure to heed market measures, it seems to me, has been a consistent pattern of error. The most serious regulatory error associated with the 2008 crisis, in my judgment, was the failure during the first half of 2008—particularly the second and third quarters of 2008—to force banks to take action to replenish their capital, even though markets were sending very, very clear signals that that was necessary and appropriate.
Second: regulators need to recognize that destruction of franchise value is destruction of capital. No regulator should ever encourage the destruction of competition, fail to protect consumers, allow imprudent risk taking in order to increase franchise value. On the other hand, it should be recognized that the gratuitous and unhelpful imposition of costs on financial institutions, in addition to whatever risks it poses for the flow of credit, undermines financial stability by reducing effective capital.
This is not the place for a detailed rehearsal of possible areas where there are extra burdens. I will only recount one story that I believe to not be an isolated type of example. Some time ago, I had a very good student at Harvard. It was a student who I knew. I had no other relationship to the student, or no other great loyalty to the student. The student was graduating and was interested in getting a job in a certain segment of finance. I wrote a note to a senior person at a major U.S. financial institution suggesting that "John Doe," a graduating senior at Harvard, was a terrific student who had an interest in such and such, and perhaps their institution might want to consider John Doe.
I confess, I regarded this as an innocuous act. I learned subsequently that there had been more than one convocation of multiple attorneys at that institution to discuss the challenges posed by the fact that doing a favor for Larry Summers could raise questions and be scrutinized in a dark light at some point in the future.
On the other hand, ignoring and not seriously considering the application of a student to that bank could also be considered to have been a failure of fairness in hiring. And they worked their way through, and the person I wrote the letter to found it necessary to recuse from the hiring process, and the student ended up getting a job at a different place. But if—you know, we're laughing—but if that kind of thing is being imposed fairly constantly, and senior people at institutions are always living with that kind of paranoia, it is not, friends, without cost. And so, I think the impulse to regulatory simplification—particularly with respect to institutions that don't have the economies of scale to bear this kind of thing—receives another rationale from the kind of analysis that I have presented here.
The third implication that I have stressed, that comes out of this analysis, is the importance of focusing on prompt capital raising when things turn down. I am not convinced that the next time things turn down, there will be a judgment that, "This is really serious. We need to catch this early. We need to face up to the problem. We need to acknowledge that institutions are in potential jeopardy, and to force the suspension of dividends and the raising of capital." And it seems to me the single most important thing that the regulatory community should ponder is how it will respond when the next downturn comes, and the pressure to say that everything is OK, and that regulation is not on the verge of failing again, and to try to maintain confidence all the way through, will be very, very strong. And that seems to me to be a priority for the regulatory agenda as well.
Finally, the analysis that I've presented raises a prospect that I believe requires considerably more attention than it has yet received. It is a commonplace to speak of illiquidity without insolvency, requiring a lender of last resort to warn that illiquidity problems unaddressed, if they lead to higher borrowing rates, will over time lead to insolvency. It seems to me there is an inverse risk that needs to be considered. Institutions with long-term liabilities, with assets that come due only in the long term—but with substantial assets that can be borrowed against, and with substantial deposits that are guaranteed—could easily find themselves effectively insolvent in economic terms, but fully able to operate because of the combination of guaranteed deposits and the ability to borrow against assets as collateral. Such zombie institutions are problems waiting to happen, and are unlikely to be substantial contributors to effective economic performance. It seems to me hard to look at the landscape in Europe over the last few years and not wonder about this possibility.
And it seems to me appropriate to ask the question: can we rely on our current mechanisms to detect and act on such institutions? And the fact that I cited earlier—that when we close institutions, their capital gap has typically become negative 15 percent—suggests that our prompt corrective actions may be rather less prompt than we like to suppose.
These are all much more difficult problems than I have made them seem, and there are certainly many regulators who are aware of much—or all—of what I have said. But if by framing these concerns in relatively sharp and stark terms, I have done a bit to undermine complacency that we have a far better capitalized financial system and therefore can feel far better than we have in the past, I will have served my purpose this morning.
Thank you very much.
Paula Tkac, moderator: Thank you, Larry. We have time for one question, so I'm going to combine a bunch of questions. You alluded at the end in your slides to needing to be wary about illiquidity as opposed to insolvency, so I'm going to put this question up here, but it really gets to a variety of questions that were asked around other potential macroprudential policies that might be put in place for reducing systemic risk.
So, if an additional regulatory capital hasn't reduced it, then what other levers might be helpful in doing so, and what would help to make a more effective regulatory system?
Summers: So, I do not mean to say that increased regulatory capital has not been helpful. I suspect if we had not had regulatory capital increases, we would have seen larger increases in volatility, larger increases in the various risk premiums that I spoke about. What I meant to say—what I think I did say—was that increased regulatory capital had not, because of all the other things that were going on, achieved a big increase in capital. To put it in prosaic terms: if I exercise a lot and my weight doesn't go down because I eat a lot, too, it would be very much the wrong inference to suppose that my exercise program had been ineffective.
And I think that's the right way to think about this, so I'm not sure that regulatory capital requirements are quite as high in every area as they need to be. There are plenty of questions about the details of the regime, whether the current construct of operational capital is one that works, for example, but I'm not making a case for reduction in regulatory capital. If anything, perhaps there should be an increase in regulatory capital. I think there should be a careful look at the stress tests to reach a judgment as to whether the predictions of robustness of the system, in the face of economic catastrophe, are in fact realistic predictions. And if they're not, we should find a different way.
I think there should be a systematic review of regulatory burdens that reduce the level, and increase the volatility, of bank profits. I am skeptical that the Volcker Rule has been a net positive contributor to financial stability, in light of the questions about liquidity that it has raised, and the decline in reasonable business lines that has resulted—and in particular in light of the arbitrariness of the distinction between market making and proprietary trading. I think there are a variety of other areas where there has just been a degree of regulatory burden imposition that has reduced profits, therefore reduced franchise value, and therefore increased uncertainty and potential instability, that would be profitable to review.
I also think—and perhaps I should have said this in my initial remarks—that it is terribly important that our regulation be functional as well as institutional. And if the effect of our regulation is to burden some institutions and not burden other institutions who are able to compete without regulatory burden, and therefore to undermine the profitability of regulated institutions, the consequence of that may be not just to allow, as many stress, the increase in risky activity outside of the regulated financial system, but may ironically be to undermine, by undermining profit prospects, the regulated financial system.
So I would say that the arguments I've made in some strengthen the case for regulation of the shadow banking system, strengthen the case for review of unnecessary and excessive regulatory burdens, strengthen the case for hard thinking about the dynamic aspects of the regime—looking in particular at market values as triggers, market behavior as triggers for regulatory action—and, if anything, make a case for higher, rather than lower, regulatory requirements.
Tkac: Well, thank you. Given the questions here, I'm sure we could go on for an hour; but my first job is to get us out of here and keep us on time, so I want to [ask you to] join me in thanking Larry for his comments this morning, and we'll take a short break.