February 5-7, 2013 | St. Louis Mo.
February 7, 2013
Breakfast Keynote Address
Introduction: James Bullard (3:30)
Keynote: Governor Jeremy Stein (46:29)
Keynote Q&A (11:22)
Session Four – Household Balance Sheets: Deleveraging and Economic Growth
- What's Driving Deleveraging? Evidence from the 2007-2009 Survey of Consumer Finances
Karen Dynan and Wendy Edelberg (22:08)
Session Four, Part 1 Discussant: John Krainer (20:16)
Session Four, Part 1 Q&A (14:47)
- Household Balance Sheets, Consumption, and the Economic Slump
Atif R. Mian, Kamalesh Rao and Amir Sufi (21:57)
Session Four, Part 2 Discussant: Brian Melzer (18:51)
Session Four, Part 2 Q&A (15:44)
Session Four Moderator: Daniel Davis (2:14)
Closing Plenary – Facilitated Panel Discussion: Household Balance Sheets and Economic Growth
Panelist: David Buchholz (10:24)
Panelist: Steven Fazzari (11:41)
Panelist: Deniz Igan (12:13)
Discussant: Barry Cynamon (20:03)
Panel Discussion Q&A (27:29)
Ray Boshara (2:09)
Below is a full transcript of this video presentation. It has not been edited or reviewed for accuracy or readability.
Governor Jeremy Stein: Thanks very much. Thanks, Jim. Thanks for that very nice introduction. And thanks to the organizers for having me, so it’s a pleasure to be here. So let me start by apologizing for the imperfect alignment of my talk with the theme of the conference. I’m absolutely going to talk about financial stability, but as you’ll see, the household part is going to be a little suppressed. So it’s going to be about financial stability with a little bit more emphasis on financial markets and on financial intermediaries. So in particular, the question I’d like to address is the following. What are the factors that lead to overheating episodes in credit markets? In other words, why is it that sometimes we see credit booms when it looks like credit standards get laxer when, you know, from that point forward, returns to investors seem to be lower? You know, we’ve seen at least in some instances that these sorts of credit booms can have adverse spillover consequences to the rest of the economy. And I’m going to focus quite a bit of attention on the why question. You know, why do we see these things with the hope that if we understand the why, it will help us do a little bit better job on, you know, measurement issues and ultimately on policy responses.
So let me start by kind of laying out, and this is kind of a straw man exercise, by laying out two distinct, no, not necessarily mutually exclusive, but two distinct views of the mechanisms that might lead to variation in the pricing of credit risk overtime. So one view I’m going to call a primitive preferences and beliefs, and the other I’m going to call institution agency on incentives. And so the first, I think, you know, is for any economists who want to sort of start with the basics. The first is kind of the natural starting point, but my kind of pitch here is going to be that if you really want to understand credit markets, you’ve got to go not only to the first, but also to the second and start thinking really quite seriously about the institutions that are involved.
So according to the primitive’s view, if you want to think about changes in the pricing of credit overtime, they’re going to ultimately reflect the preferences and beliefs of the kind of people that Chris Carroll was talking about last night; the households. Okay, I’m going to talk about households for one minute here. So you know, they’re the ones that are ultimately the intermediaries and all that are just kind of doing their bidding in this view of the world. And so they’re the ones that matter. So if you wanted to have an account of why credit is cheap, you would say, well, it must be credit is cheap now because household risk tolerance is high. Maybe wealth has gone up, or maybe it’s a beliefs thing exactly as Chris talked about. Maybe their beliefs are more optimistic about the future, and they’re willing to lend—you know, the lenders are more optimistic about the future. They’re not thinking about low-probability risks and so forth. So that would be a sort of household primitive centered view.
I think this view is absolutely on track when you want to think about pricing of the equity market. So if you want to think about things, you know, why are expected returns on stocks sometimes high, sometimes low? Internet bubble? Things like that. It’s clear, and we have a fair amount of evidence that the beliefs of households are at least partially involved. So very nice survey evidence which shows that when retail investors are most optimistic about the future returns on the stock market, that’s when actually a statistical model tells you the stock market’s going to do poorly. All right? So that’s a sort of irrational beliefs kind of story.
So by contrast, I’m a little bit more skeptical that one can really tell the story of credit markets as fully without focusing on the role of institutions and incentives. So basic premise is pretty straightforward. If you think about who makes credit decisions, they’re almost always delegated to agents inside banks, mutual funds, hedge funds, pension funds, and so forth. And the idea of just being that if you really want to understand how credit gets priced, and how credit decisions get made, you have to really think quite specifically about the incentives and the rules facing the people making the decisions. And these incentives, and you know, the rules include sort of a whole host of things. So it’s compensation schemes, it’s corporate governments, it’s very importantly regulation, it’s accounting, it’s everything we use to sort of measure and, you know, create motivation in governance around these delegated decision makers.
So let me say, you know, at an abstract level, you can think of the decision makers, the agents, and the people who write the rules—again, this is both regulators and internal corporate governance. You can think of this as sort of an ongoing evolutionary process. And each is sort of evolving overtime. So take the rules as given at any point in time. You’ve got a given set of rules. You as the agent are kind of trying to do your best, and sometimes you will find vulnerabilities in these rules, and you’ll do things that effectively, intentionally or unintentionally, exploit the vulnerabilities in the rules, and end up doing things that are not optimal for the principle, where the principle can either be thought of as your firm, if you’re a trader, say, or society in some larger sense. Now, of course, if the damage caused is significant, then there’s going to be adaptation on the other side. In other words, the rules will get revised. That can be either internal, you know, governance will change the incentive schemes, so that we’ll start having more clawbacks, things like that. Or it will be regulation. Nevertheless, you know, because it’s sort of a dynamic process, we’re going to get time variation, and we’ll get to see periods when sometimes the rules are containing things pretty well, and other times when, you know, one side gets the upper hand, and the rules do not as good a job of containing sort of this agency based behavior.
So let me get a little bit more specific. I think one of the fundamental challenges in this kind of delegates investment management is many of the rules that we use, either regulatory or incentive based rules, are vulnerable to a particular thing, which is agents who act to boost their returns by effectively selling insurance against unlikely events. That is by writing, you know, deep out of the money put options. So a classic finance textbook example is if I hire you to manage my equity portfolio, and I say, “I’d like you to beat the S&P 500. That’s your job.” You can actually do a very, very good job of that by literally doing nothing more than holding the S&P 500, and writing put options against it. It’s a theorem in finance that you can beat the S&P in mean and in variance by writing deep out of the money options. Now of course you do that by taking on tail risk that is not well measured by mean and variance, and that may make me, as the person who hired you, sad after the fact. But it’s something that, you know, just illustrates the point that give me a measurement scheme, and finance is very flexible, very pliable, and there are ways to sort of create the appearance of that performance.
So credit, why is this relevant? You think about credit. Credit is really fundamentally all about as an element of put writing. Right? You’re sort of short an option when you take default risk. So it’s always going to be somewhat challenging in an agency context, especially to the extent that the risks associated with the put writing can be structured in a way that is sort of partly of age the current measurement scheme. So think about the Triple A rated tranche of a subprime CDO, where in that case, maybe the measurement scheme is the credit risk model used by the rating agency. So you know, they have a model, to the extent that that model is a little bit behind the curve, and it was designed, say, for first generation securitization. If the model doesn’t fully understand the imbedded leverage, or the correlational issues associated with a second generation product, it’s going to be behind the curve, and somebody who’s designing against that model, you know, will find vulnerabilities in it. So that’s sort of the general, you know, mindset I’m trying to outline. Some of these agency problems are further exacerbated by competition among intermediaries and relative performance evaluation. So for example, you see this in the money market mutual fund sector where, you know, empirical evidence shows you can gain a lot of assets under management by just having a slightly higher yield than your competitors. Now, that’s a good thing if that extra yield reflects true ability. It’s less of a good thing if that extra yield reflects risk taking that’s not fully appreciated. So again, and the competition to sort of take market share away can potentially exacerbate that sort of risk taking.
Okay. So that’s the general thing. Now, I recognize, it’s not all that satisfying, either intellectually or from a policy perspective, to just list ways that delegated money management can go wrong. Right? It sort of doesn’t make sense, because this is the system we have. It obviously works in some larger sense, because you know, we have very, very large, deep, wide credit markets based almost entirely on this sort of delegated thing. So it must be that on average, we have evolved the evolutionary process as led us to a place where the combination of internal governance and external regulation has managed to contain these problems reasonably well. Again, on average. So this is not meant to be, sort of, you know, saying this stuff doesn’t work. That’s totally not the point. The more interesting set of questions is sort of the time series questions. Why is it that sometimes occasionally, things get a little out of balance and you know, these agency problems become more pronounced, and we get a certain kind of risk-type taking that we don’t like? You know, what is this whole institutional view say from a kind of positive economics perspective about why sometimes credit markets overheat?
So let me suggest three factors that can contribute to overheating. The first kind of flows naturally out of this perspective, is financial innovation. So obviously, you know, financial innovation has provided important benefits to society, but this point of view suggests a dark side, which is sometimes financial innovation can be about giving new ways for agents to write stealthy put options that are not well captured by the existing rules. And so one message is, I think, you know, policy makers probably should be on alert anytime there’s rapid growth in a new product that’s not yet fully understood. So back to these subprime CDO’s second generation securitizations. If you ask yourself why did they exist in the first place? Sort of a, you know, a reasonable conjecture is they arose precisely because of this kind of evolutionary process where there was a set of rules, and eventually something is going to arise that finds a weakness in the rules.
Going back further, you know, you can tell a similar story when payment and kind is their pick. Payment and kind interest features were first used in leverage buyouts in the late 1980’s. You know, why did these things suddenly get invented? Probably not a coincidence that many of them were sold to savings and loan institutions at a time when both savings and loans had a very particular set of incentives to generate income, to boost their measured capital and so forth. So I think you can often understand something about financial innovation by considering the agency problem on the side of the buyer of the investment. Okay?
Second related factor on this list is changes in regulation. New regulation is going to tend to spur further innovation as people try to sort of work their way around the private cost that the rules create, and it may also open up new loopholes that existing instruments may be able to exploit.
Third factor I think that’s important here is change in the economic environment that changes the incentives of the agents making the credit decisions. So for example, a period of low interest rates of the sort that we’re experiencing today can create incentives for agents to take either greater duration or greater credit risk in an effort to, you know, reach for yield. So an insurance company that has guaranteed minimum rates of return on some of its products might find its solvency threatening by a period of low rates, and they find that it is in a position where it feels compelled to take on added risk. You can tell a similar story for a bank that has a deposit taking franchise. That deposit taking franchise’s profitability has been compromised by low interest rates, and in an effort to kind of keep its margins up it might feel that it has to take on risk. And obviously, these factors may not be independent of one another, they might interact. So low rates may create a demand for certain form of risk taking and financial innovation may in part arise to fill this demand.
Okay, so that’s—you know, that’s the story. And I’ve taken some paintings to draw a distinction, probably a somewhat overdrawn distinction, between this view of households risk preferences versus kind of agents and incentives. You know, why bother to draw the distinction? Why should you care? Why should you care? You know, if it’s markets overheating, it’s markets overheating, why make such a fuss about the underlying mechanism? One reason is that I think your view of the underlying mechanism shapes how you go about measurement. Okay? So let’s consider a very specific question. Suppose I asked you, let’s look at the high yield market. Does the high yield market look overheated in the sense that as an investor, one might expect disappointing returns on a going forward basis. You know, very concretely, what would you look at? What data would you look at to help inform that kind of a forecast? I think if you’re thinking in a kind of households primitives thing, I think the most natural thing is well, you’d look at credit spreads, and say, well, if investors have a lot of risk tolerance, they’re going to want to buy a lot of these bonds. It’s going to bid up the price of the bonds, and price relative to fundamentals is sort of a natural place to look. So you’d look at credit spreads. On the other hand, if you take this agency perspective, where agents are sort of trying to beat various incentive schemes, it’s less obvious that increased risk appetite is going to be as well summarized simply by credit spreads. In other words, I may, as an agent, be willing to sacrifice some return, but I’d rather do that via settle or non-priced terms, subordination, other things that allow me at the same time to show a higher stated yield, because I’m in the business of looking good needs to be some kind of a measurement scheme.
And again, I’m just going to come back to the example of pay and kind bonds in leverage buyouts. And this is just sort of reminiscing. A long time ago, many, many years ago, my buddy Steve Kaplin [phonetic 00:14:52] and I did a study of the leverage buyouts of the 1980’s. And one of the things that was really striking is, you know, there’s all this financial innovation at the time, the use of these pick bonds, which turned out to have, you know, a not happy ending for the people who bought them, those things did not have low credit spreads. They looked pretty attractive if you looked at the credit spreads. But of course, what they gave up was just tremendous amounts of subordination. So these things were due not to get paid for many, many years. They were, at the same time, in deals where there was a slud of bank debt. The bank debt was getting paid back very, very quickly. The companies were often selling assets from underneath the junk bond investors to pay back the banks. So you know, the junk bond investors had moved very far to the back of the line, and it was that observation that helps you understand why these guys got low returns. And I think there’s some broader example there that’s worth bearing in mind.
It’s interesting to think about, there’s some recent work—much more recent work by Robin Greenwood and Sam Hanson of HBS that I think you can at least think of through this lens. They show that if your goal is to forecast the returns on credit—so you’re just trying to do a forecasting model to think about how credit investors will do. Credit spreads are helpful, but you get a lot of extra incremental explanatory power from a non-priced variable, namely the high yield share. The high yield share being the fracture of total bond issuance that is in the high yield or junk bond category as opposed to investment grade. When that’s high, bond returns going forward tend to be low. And there’s not only statistically, but sort of a strong economic magnitude there. And one interpretation, this is just me putting an interpretation, is that there are all these non-priced terms that issuers are responding to. So when they see covenant protection going away, and other things like that, they think this is an attractive time to issue. And the issuance picks it up, and it’s sort of a summary for all these non-priced terms. So that’s one implication of this view. Sort of one thing you get for thinking explicitly from an agency perspective.
A second implication that I’d like to highlight is what you might call the tip of the iceberg caveat, which is that quantifying this sort of risk taking in credit markets is going to be inherently difficult to do in real time, precisely because the risks are often taken in opaque ways that escape conventional measurement practice. So one should be humble about one’s ability to ever see the whole picture, and when you do see clues, you should interpret them a little bit accordingly. So for example, like I’m going to keep mentioning junk bond market, and I want to be clear, it’s not because I think it’s the necessarily leading venue for the sorts of risk taking that has systemic implications. Rather, I think it’s a useful barometer. You know, precisely because as I said, we have a longtime series, we know how to benchmark the credit spreads, we know how to benchmark the covenants, all that kind of stuff. So it’s a useful barometer that might give you clues, and there are other forms of risk taking that are kind of more innovative and more novel, and it’s harder to benchmark them. They may be ultimately more interesting, but the junk bond market may be kind of an interesting clue. Okay?
So with all this sort of prelude, what I thought I would do next is just take you on a very brief tour of a handful of recent developments and credit markets, and we can kind of think about them through this lens. So this tour is going to draw very heavily on work done by fed staff as part of what we call our quantitative surveillance efforts. And this is under the offices of our office of financial stability. So just to sort of go through sort of a few facts and figures, so the first stop on the tour is going to be the market for leverage finance. So this includes both public junk bonds, and so-called syndicated leverage loans. Let’s see, and I’ve got a slide here. So first thing to just point out is these markets have been pretty robust. The junk bond market set a new record for issuance in 2012. So there’s just sort of been a lot of activity, and you guys know all that. In terms of, you know, variables you might look at to start gauging whether there’s anything that feels like overheating, I would say the picture is decidedly mixed. So first thing again you might look at is credit spreads. So credit spreads, you know, have come down. Have come down pretty markedly in recent months. Nevertheless, if you try to benchmark it, by absolute standards, I would say they’re sort of moderate. They’re median. So in fact, the credit spread as of today on high yield is roughly 400 basis points, and that is within, you know, a few basis points of the median of the ten years leading up to the crisis. So 1997 to 2007. So credit spreads I would say look, you know, plus or minus ordinary. So you know, you wouldn’t take much inference from that. On the other hand, this high yield share that I alluded to a moment ago, the fraction of issuance that’s junk was somewhat elevated. Not remarkably elevated. I think it’s at sort of the 65th percentile of its distribution over the last however many years. So that’s a little bit more pessimistic, again, based on the research of Greenwood and Hanson.
Non-priced terms are interesting. So we’ve seen interest in movement in the past year, and especially in the past quarter. And I don’t know how clearly you can see this. I’ve got a few different of these things. The top upper left panel is payment in kind. Again, it’s the use of, you know, pick interest. This is non-cash interest. You know, quite substantially elevated in the fourth quarter, starting to run at levels, you know, approaching highs in 2007. The panel on the top over there is covenant light. These are loans where basically covenants get dropped another non-price adjustment. And again, you can see a sort of similar pattern that’s been elevated in the fourth quarter. Similarly, bottom left panel dividend recapitalization, these are loans that are issued to take out basically, you know, where the firm borrows to pay a dividend to the equity investors, which is again, it’s a situation where the bond holders are subordinating themselves a little bit. So that, too, has gone up a bit. And finally, and least ocularly significantly in leverage buyout deals, there has been sort of slightly more leverage. I would say that that’s a trend, but a somewhat new trend. So you know, that’s the picture. You can kind of weigh these different pieces of evidence a little bit subjectively. My own reading overall is that we are starting to see a somewhat significant pattern of reaching for yield starting to emerge, I guess I would say, in corporate credit.
Now, here’s the important qualification even if this were true. So this is my conjecture based on, again, a bunch of evidence assembled in a subjective way. Even if this were true, and even if that doesn’t bode well for the expected returns to the investors themselves; the junk bond investors, the leverage loan investors. Importantly, it need not follow that this has ominous systemic implications. That is to say, even if at some point some of these investors have disappointing returns, absence spillovers to the rest of the system, the losses may be confined, and therefore not a particularly pressing policy concern. So in this regard, one lesson from the crisis is, it’s not just bad credit decisions that creates systemic problems. It’s bad credit decisions combined with excessive maturity transformation. Okay? So you know, a badly underwritten subprime loan is one thing. A badly underwriting subprime loan that is then collateral for an asset back commercial paper vehicle, which in turn, the commercial paper is acquired by a money market mutual fund. That’s something else, and potentially more dangerous. So I think one wants to start trying to pull these ideas together to even begin thinking about systemic implications.
And this observation suggests a measurement principle, which is in principle, what you’d like to know for any asset class is not just, you know, your forecast of expected returns. You’d like to try to understand its capital structure. That is to say, you know, if you could sort of take all the subprime mortgages, or all the junk bonds, you’d like to say, ultimately, who’s holding them? And are there short term claims against it? What fraction of all the junk bonds in the universe are being held by people who have the kind of short term claims that they can pull out, you know, on short notice or when things start to go bad? And it’s that short term financing that creates the potential for fire sales, because they can liquidate and then force these illiquid assets to be sold, and that, you know, fire sales and deleveraging, that’s in some sense the mechanism that makes things like this become a little bit more systemic.
So now this is a, as I said, it’s an idealized measurement construct. It’s a little hard to operationalize. Here is one sort of faint picture, which I think should give you a little bit of comfort. So this is dealer financing of corporate debt. This is meant to capture, you know, corporate debt that has been ultimately financed by broker dealers, you know, lending on a repo basis against the stock. So the basic picture here is that, you know, there was a pretty pronounced increase in this short term funding of corporate debt in the years leading up to the crisis. It drops very abruptly during the crisis, and then basically doesn’t go back up. Okay? So you know, relative to where we were several years ago, there appears to be significantly less short term leverage against this particular class of instruments, which given kind of the story I’ve been laying out, is I think cause for some comfort. In other words, you know, all else equal, the same amount of credit risk taking is going to cause less systemic harm if there’s less leverage against it. Okay? So I think that’s one kind of direction you could go in.
You know, nevertheless, as I said, our ability to measure what we would ideally like to measure is limited, so I want to be cautious, and just stress how hard it is, you know, because you were talking about sort of you want better data, you know, how hard it is to get the data to line up with kind of the theory that we want. Ideally, what we want for any asset class is the totality of short term claims that can pull the plug when things start to go bad. So repo financing is one example, but there are other things. And crucially, I want to make the point, these short term claims need not be debt claims necessarily. In other words, for example, if you have illiquid junk bonds or leverage loans, and they’re held by open end investment vehicles like mutual funds or ETFs where it’s demand—it’s equity, technically, but it’s demandable equity, so they can pull the plug at the first sign of trouble. That can have the same fire sale generating properties as short term debts. So I think we’re naturally, based on a sort of experience in the crisis, we’re naturally inclined to look at repo, but precisely because we’re looking very, very careful at repo, it may be that next time the short term financing takes a slightly different form.
So with that in mind, I’ll just show you a picture of inflows into—so this is just for the high yield category. This is inflows into mutual funds in the top panel, and exchanged traded funds or ETFs in the bottom panel. And interestingly, it’s almost the mirror image of the pattern you saw with the dealer financing. That is to say, pretty flat and uninteresting in the years leading up to the crisis, but then, you know, reasonably sharply increasing in the last few years. So you know, don’t want to make too much of it, but it suggests, I would say loosely it suggests that there may be some substitutability between these different forms of demandable finance, and just underscores this message about data, which is you don’t want to get too obsessed with focusing on a particular category. You want to think about the data a little bit broadly.
So continuing on with the broad theme of maturity transformation, here’s the next stop on the tour. It’s maybe a little esoteric. It’s something called the agency mortgage real estate investment trust. Or agency REIT sector. So what are these guys? They’re guys that are growing pretty fast. So they’ve gone from 152 billion year end 2010 to 398 billion at the end of 2012. What do they do? They basically buy agency mortgage back securities. Same stuff that the fed buys when we do our asset purchases. They buy these and they fund them very short term. So this is a pure maturity transformation activity. They’re buying these long term mortgage back securities and funding them in the repo market. Okay, so again, not to suggest that this is at a point where it’s systemically threatening, but this is the kind of thing I think one wants to be paying attention to, and thinking about. Again, if you think that the mechanism of systemic problems has to do with maturity transformation, this is an interesting class of activities. And of course, one thing to note about this business model is its viability is very much dependent on conditions both in the NBS market and in the market for short term repo finance, because these guys live off the spread between the two. So in other words, if you can borrow at the short end cheaper, you know—if you can borrow at the short end at a lower rate, then the long term NBS yield, you’ll earn a spread, and that spread can be passed on to your investors. If the yield curve flattens, it gets harder all else equal unless you start leveraging more, it gets harder all else equal to just generate the same returns. Okay, and I’m going to come back to this a little bit later.
Another place where this desire to generate yield can show up potentially, is in the securities holdings of commercial banks. So I did some recent work a year or so ago with Sam Hanson where we found that—one thing we documented is that the duration of banks non-trading accounts securities, that duration tends to increase when the short rate goes down. So as the short rate goes down, banks seem to be holding longer maturity securities, so our hypothesis is just a hypothesis is that it was due to a particular form of agency behavior, which is that given the conventions of GAAP accounting, you know, you can boost your reported income all else equal if you move out of low yielding short term stuff into higher yielding long term stuff. Of course, if the expectation’s hypothesis is true, there’s going to be some countervailing capital gain or capital loss, but from a pure measurement perspective, it looks like you’re doing better when the yield curve is steeper to push out into longer duration securities.
It looks like, I don’t want to overstate, it looks like there’s something along these lines happening today, the maturity of securities in banks available for sale portfolios appears to be near the upper end of its historical range. It’s what you might expect given low short term rates. I would say this is noteworthy on two counts. First, the added interest rate exposure may, itself, be a potential source of risk for the banking sector something that we want to pay attention to, especially since existing capital regulation, as many of you know, doesn’t explicitly address interest rate risk. It addresses obviously many forms of credit risk, but there’s not a capital charge for interest rate risks, so one wants to use other forms if it’s not capital regulation, be it stress testing or otherwise, to just be paying attention to this stuff. And second, just in the spirit of tips of icebergs, this is something we can measure. And if there’s evidence of a desired to reach for yield on this dimension, it might be suggestive of that desire elsewhere where we have a, you know, less easy time of kind of measuring it empirically.
All right. Final stop. Final stop. And this is the most financial engineering thing I’ll talk to you about. This is something called collateral transformation. Okay? So collateral transformation, this is a little tricky. This is something that’s been around for quite a while. I don’t want to be very clear. Right now, it’s not happening at a scale as far as we know. I mean, the data’s a little sketchy. It’s not happening at a scale that would raise any particular concerns. I’m highlighting it rather because it’s exactly the kind of activity where at least in principle our new regulations could be creating the potential for rapid growth in the future. So it’s just something that, you know, I just want to kind of put out there so people understand it, and you know, to think about being watchful in the future. So it’s a tricky thing. Let me try to explain it with an example. So imagine that you’re an insurance company, and you want to do a derivatives trade for whatever reason. To do that, you’re going to have to post collateral with a clearing house to do the derivatives trade. Now, suppose the clearing house has high standards, and it says, “I want you to post pristine collateral.” By which I mean, “I want you to post treasury securities.” Okay? Now, unfortunately, the insurance company that wants to do the derivatives trade doesn’t have any unencumbered treasury securities. All it has sitting around, basically, are some junk bonds. So here’s where the collateral swap comes in. The insurance company might approach a broker dealer and say, “Can we do a trade?” Basically, “I’m going to give you my—“ it’s a two-way repo transaction. I’m going to give you my junk bonds, lend them to you for 60 days, whatever. You’re going to then lend me some treasury securities. I can then take those treasury securities and post them as collateral. Okay? So now the insurance company can go ahead and do its derivatives trade. Now, of course, the broker dealer may not have either a bunch of these spare treasury securities sitting around. It may be a middle man. So in order to round up the treasury securities, it’s going to have to do a mirror transaction with somebody else. So maybe it finds a pension fund. So maybe it goes to the pension fund and does the reverse swap. That is to say the pension fund has the extra treasury securities and takes the junk bonds, and you know, so at the end of the day it’s basically the pension fund via the broker dealer has lent the treasury securities to the insurance company. Okay? Now, why would the pension fund want to be involved in this? Maybe it has, going back to kind of the theme, maybe it has something of a reaching for yield motive. It wants to show slightly higher yield, it will be, from the balance sheet perspective, holding treasury securities, but it’s got a little bit of extra income coming in from having done this transaction. Okay?
So again, this is a business that we don’t have any evidences happening at large scale. But you know, given that we’re moving to a world where there’s going to be a lot more demand for safe collateral because of Basel III liquidity requirements, because of clearing houses, because of increased margins for uncleared derivatives, all of that stuff. It’s the kind of thing that could exert a bit of a gravitational pull on this kind of business. And two points, and you know, we’re from a systematic perspective worth noting about these sorts of transactions. First of all, they reproduce some of the same unwind risks that would exist had the clearing house simply lowered its standards in the first place. That is to say, think about what happened. Suppose the junk bonds that have been posted in this transaction fall in value. Well, then the broker dealer’s going to say, “Wait, wait, wait. I’m going to impose a margin call effectively. I’m going to make you post a little bit more margin.” What will that do? That will force the insurance company to scale back its trade, and basically do some unwind much as it would have had to do had it used the junk bonds directly as collateral with the clearing house. So I think it’s helpful on sort of default risk. It’s helpful that way. It’s less helpful in combating what you might call unwind risk and the pressure to sell assets when asset values fall. Secondly is, you can see by just the complexity of the description I gave you, we’re creating some extra counterparty exposures in here, right? Because now we’ve got the insurance company with the broker dealer, and the broker dealer with the pension fund. And so there’s sort of extra layers of counterparty exposure.
Okay. Again, you know, this is not something that’s currently large. We have some very suggestive evidence of growth potential in that the fed does something called the senior credit officer opinion survey, and one of the questions we ask in the survey of dealers is about perspective transactions. So they indicate not much going on, or not so much going on in terms of current transactions, but you know, a greater degree of clients inquiring about such transactions going forward. So again, just as something just to put on the radar screen as something that merits some future monitoring.
All right. So let me now turn to policy implications. So the general question of how policy makers should respond to these various different manifestations of market overheating is a big difficult question. I’m not going to attempt to deliver a set of prescriptions. The one thing I thought I would try to do is provoke a little bit of discussion around one specific aspect of the question. Namely, what are the respective roles of traditional supervisory and regulatory tools on the one hand versus monetary policy on the other hand in addressing the sorts of market overheating phenomenon that we’ve been talking about? So just to be a little bit concrete, imagine, this is obviously just purely hypothetical, imagine that it’s 18 months from now, and that with interest rates low, all the trends that I’ve identified have kind of continued to build to the point where we believe that there could be some meaningful systemic implications. So in such a hypothetical setting, what if any policy measures might be contemplated. So I sometimes argue that in such circumstances, that policy makers should follow what you might call a decoupling approach. So what I mean by decoupling is that monetary policy should restrict its attention to the duel mandate goals of price stability and maximum employment, and the full battery of supervisory and regulatory tools should be used to safeguard financial stability. Okay? So you know, why decoupling? There’s several arguments you could make. First, monetary policy’s likely to be a blunt tool for dealing with financial stability concerns. Even if we stipulate, even if we just assume that, you know, low interest rates are part of the reason for, say, a worrisome boom in a segment of credit markets. They’re unlikely to be a whole story. And you know, would you really want to raise rates and risk choking off economic activity broadly in an effort to rain in a part of the market? Wouldn’t it be better to have a kind of more surgical and narrowly focused supervisory regulatory approach that doesn’t have this collateral damage to the rest of the economy? So that’s I think one argument in favor of decoupling.
Related concern is that if that monetary policy already has its hands full with the duel mandate, this is sort of a tools versus—you know, instruments versus objectives kind of issue. And that if you also make monetary policy responsible for financial stability, it’s going to have more objectives than instruments, and it won’t be able to do any of its tasks well. So you know, I think these critics are important and correct. You know, it’s clear that in some cases, supervisory and regulatory tools are better targeted and more likely to be effective. So back to my example of banks and duration risks, it seems like the fed is well positioned. Stress testing, supervision is well positioned to sort of try to deal with that as opposed to try to do something with monetary policy.
Nevertheless, you know, going forward, I think it’s going to be important to keep an open mind, and not to be too rigid about applying this decoupling philosophy. So that’s to say in spite of the caveats, I can imagine circumstances where it might make sense to enlist monetary policy tools in the pursuit of financial stability. So let me offer three observations in support of this perspective. First, despite what I would consider to be significant recent progress, supervisory regulatory tools remain imperfect in their ability to address many sorts of financial stability concerns. So if the underlying environment creates a strong incentive for institutions to say take more credit risks to reach for yield, it’s unlikely that regulatory tools can completely—they can help, but it’s unlikely that they can completely contain this behavior. That’s not to say we shouldn’t do our best with these tools. Absolutely we should. But we should also be realistic about their limitations. These limitations arise because of the inherent volubility of the tools in a world of regulatory arbitrage, because our authority doesn’t extend equally to all parts of the financial system, and as I said before, because risk taking often tends to be taken in non-transparent ways. So it’s somewhat out of the field of vision. Okay, so that’s one observation.
Second observation is, you know, while monetary policy in some sense may not be the ideal tool for the job, it has one key advantage, which is monetary policy gets in all the cracks. So what I mean by that is, you know, think of it, the one thing that a commercial bank, a broker dealer, an offshore hedge fund, an ABCP vehicle. The one thing these guys all have in common is they all face the same set of market interest rates. So market interest rates exerts sort of a broader and less, you know, arbitrageable force; set of influences, and they may be all to therefore reach into corners of a system that supervision regulation cannot.
Third, and finally, in response to this sort of concern about adequate number of instruments, we’ve seen in recent years that the monetary policy toolkit consists of more than just a single instrument. That’s to say, we can do more than just literally adjust the funds rate. So concretely, by, for example, changing the composition of our asset holdings as in the maturity extension program, we can influence not only the expected path of the short rate, but also term premium, and the shape of the yield curve.
And you know, we’ve done so thus far to provide further accommodation, but it may well be that once we move away from the zero lower bound, and we don’t need this in a purely accommodative sense, the second instrument may continue to be helpful, not simply, again, in providing accommodation, but as a complement to other efforts on the financial stability front. So to see why, recall the central role; that maturity transformation . The funding of long term assets with short term run prone liabilities. The fundamental role that this plays in propagating systemic instabilities. And we’re over, as illustrated by the mortgage REITS sector that I mentioned a little while ago, the economic appeal of maturity transformation hinges quite importantly on the shape of the yield curve. Okay? In that particular case, that has to do with NBS spreads and the GC repo rate at which you could fund in the short term. And you know, think about what our policies have done. Again, in the name of accommodation, you know, we have at times put pressure on these sorts of spreads, right? And think about what is the twist in operation twist? Right? Is buying long term securities thereby pushing down long term yields selling short term bills, there’s considerable evidence. I think that, you know, there’s almost indisputable evidence that we have successfully managed to push down long term yields. There’s I think more tentative, least suggestive notion that when we were selling bills during the end of maturity extension program, it was putting upward pressure on so-called GC; general collateral repo rates. In others, we were flattening term premium. We were flattening the yield curve. As I said, that’s the twist in operation twist. And I think once you think of it that way, you can see a potential financial stability angle as well as a possible response to questions about number of instruments.
So just, you know, imagine that at some point in the future we were away from the zero lower bound, and we found ourselves after our traditional duel mandate reasons, we wanted to ease policy. It was appropriate to be more accommodative. And suppose that at the same time, there was a rising in financial markets, a general concern about excessive maturity transformation. Okay? And we were having a hard time. We were doing the best we could with our supervisory and regulatory tools, but we felt like we were having a hard time doing it as well as we would like. It might be at the right combination of policies there would be to, on the one hand, lower the path of the federal funds rate, which would be providing accommodation in the traditional manner, while at the same time doing the type of asset swaps that we did during the maturity extension program to flatten the yield curve and thereby, reduce the appeal of doing this sort of maturity transformation. So there you have two tools kind of working on two different problems.
All right. Let me wrap up just by saying, I hope you take this last example in the spirit in which it was intended. This is not meant to be even close to a currently actionable policy proposal. It’s just a hypothetical that’s, you know, intended to lend a little bit of concreteness to a broader them. And so the broader theme. And so the broader theme is the following. Is that one of the most difficult jobs that central banks face is in dealing with these sorts of episodes of credit market overheating that could potentially pose a broader threat to financial stability. As compared with inflation or unemployment, the measurement here is harder. So even recognizing the extent of the problem in real time is a challenge, and on top of that, the supervisory and regulatory tools that we have, while helpful, are far from perfect. So that suggests two principles. Principle one is decisions are inevitably going to be made—have to be made in an environment of significant uncertainty, and you have to calibrate your standards of evidence accordingly. If you wait for decisive proof of market overheating, that may amount to an implicit policy of inaction on this dimension. And second, I think we just want to be open minded about how we can best use the fuller rate of instruments at our disposal, and in some cases, it may be that the only way to achieve a meaningful macro prudential approach is by allowing for some greater overlap in the goals of monetary policy and financial regulation.
Thanks very much. If we have a few minutes, I’d be delighted to take some questions.