The ideas are nicely summarized by Lacker as relating to two alternative theories of financial instability. If we observe a financial system that exhibits recurring crises, or even if we observe one crisis or panic, we might think that such a financial system is inherently unstable. That's the Diamond-Dybvig view of the world. Banks are about maturity transformation. That's socially useful, but leaves the banking system vulnerable to runs. The run problem, according to this view, can be eliminated or mitigated through interventions such as deposit insurance. We will need other regulations as well, for example capital requirements, to correct the moral hazard problem that is induced by deposit insurance.
Alternatively, we might think that financial instability arises from induced fragility. In this view, there are several dimensions to moral hazard. Deposit insurance induces excessive risk-taking by banks, too-big-to-fail induces excessive risk-taking by all large financial institutions, and the behavior of the Fed can also give rise to moral hazard. For example, suppose two alternative scenarios:
1) An increase in perceived risk in financial markets causes markets for short-term credit to "freeze." Financial intermediaries borrowing short and lending long face higher borrowing rates, with the spreads between these borrowing rates and safe rates of interest rising, and some investors are reluctant to lend to these institutions at any rate. The central bank does not intervene, so financial institutions have to be creative in adjusting to the increase in perceived risk. They develop new contracts and new modes of intermediation and exchange, that mitigate the problem. The knowledge they gain will then be useful the next time such an event happens.
2) The same event occurs as in case (1), but now the Fed intervenes through the discount window, or some related lending facility. Interest rate spreads have gone up, and the quantity of short-term lending has gone down, but now the Fed intervenes by lending to the financial institutions at the heart of the freeze, using the "undervalued" assets of the financial institutions to collateralize the loans. The concern here is that financial institutions come to expect this Fed intervention in times of stress, and they depend on it. Now they don't bother thinking about the alternatives. It's standard moral hazard, of course.
Alternatives (1) and (2) are not just hypothetical. Those were real policy choices that the Fed confronted in late 2007. Some recent discussion of the recently-released minutes of the 2007 FOMC meetings has focused on soundbite issues - some FOMC participant said something in 2007 that appears silly with hindsight. For example, this New York Times article quotes then-President of the St. Louis Fed Bill Poole:
My own bet is the financial market upset is not going to change fundamentally what’s going on in the real economy...
But the following is much more interesting. In its September 18, 2007 meeting, at which there was a reduction in the fed funds target rate by 50 basis points, there was a discussion of the Term Auction Facility (TAF), which was to play an important role in the subsequent financial crisis. The FOMC was concerned that a previous reduction in the margin between the fed funds target rate and the discount rate was not having much effect. The concern was that banks were not borrowing, in part because of "stigma," the idea being that borrowing at the discount window signals a troubled bank, which deters banks from borrowing when they should. If discount window funds are essentially auctioned off through the TAF, then the result could be to mitigate the stigma effect.
Here's what Lacker has to say in the September 18, 2007 meeting:
MR. LACKER. Thank you, Mr. Chairman. I’ve been thinking a lot about this since I heard about it last week. I want to start by complimenting the staff at New York and the Board who wrote the summary memo. I think it does a very good and balanced job of articulating the costs and benefits of this proposed facility. I was going to say that they undoubtedly did it in a compressed timeframe, but then I heard you guys have been working on it for weeks. [Laughter] But in any event, my hat is off to them.That's a pretty sophisticated take on the issues involved with the TAF. Anyone who thinks that economic theory is useless for policymaking, or that real policymakers don't bring it to bear on practical problems should read that. Here's what Bernanke comes back with:
I very much agree with the staff that weighing the costs and benefits to reach an assessment about the desirability of this is inherently a difficult judgment. For me the critical question concerns the normative implications of what we’re seeing in the marketplace for term funding and the normative implications of this proposed intervention. Banks that are borrowing at term now are paying up for insurance against the eventuality that their funding costs rise—for example, because of a deterioration in their perceived creditworthiness. Banks that have viewed themselves as more at risk are naturally willing to pay more for such insurance, and some reports suggest, as Mr. Dudley did this morning, the presence of an adverse-selection problem in the sense that borrowing at term reveals oneself to be a borrower of high risk, and so only high-risk borrowers are willing to
September 18, 2007 143 of 188
pay more. Banks that are reluctant to lend at term are placing a high value on being able to use their liquidity to accommodate assets that may come on their balance sheet soon. We had a lot of discussion about this in the morning. Balance sheet capacity appears genuinely to be a scarce valuable commodity these days. That’s consistent with the notion that raising bank capital is expensive in the current environment. I think the adverse-selection story is worth considering seriously in this context because it’s the interpretation of what we’re seeing that provides the best hope for this being an intervention that improves market functioning in the microeconomic sense of the term.
But if adverse selection is what has impaired the functioning of the term market in this normative sense, then there must be lower-risk banks that are unwilling to borrow at the same high rates as high-risk banks but that are rationed because they’re unable to distinguish themselves from high-risk banks. Now, if this is the case, the only way to improve market efficiency by lending is to lend more than the current volume of term lending because otherwise we’re just going to lend it to the current term-lending borrowers and none of these rationed-out, lower-risk banks are going to get access to it. In other words, if we do lend through an auction facility to draw in disadvantaged borrowers to try to reach them with credit, we can do so only by subsidizing the high-risk borrowers as well. Now, I’ll mention that, from the discussion this morning, my understanding is that we have very little idea what the volume of that term lending is. So I don’t see how we chose this number and how we can be confident that it’s going to do this and reach through the high-risk borrowers to pick up the low-risk borrowers.
More broadly, I’m not sure I see how this facility could improve the normative functioning in the market. We’re going to auction off only the same contracts that market participants are capable of offering now, only we’re also going to subject ourselves to the additional constraints September 18, 2007 144 of 188
imposed by our single-price auction format. So we’re not improving on any contract out there. The only unique attribute we would appear to bring is our ability to subsidize lending terms. We could conceivably improve market functioning if adverse selection is the right story here by doing something that market participants are incapable of doing, and that would be compelling borrowing by everybody or by a set of people to achieve a superior pooling allocation. But I don’t think we want to do that. Or we could conceivably improve market functioning by acting on information that’s superior to that of market participants—a knowledge of the creditworthiness of institutions, for example. But it isn’t clear that this is a key part of the proposal either, because institutions have to be rated 3 or above to get access and I think virtually all the currently affected institutions in these sorts of high-risk and low-risk categories are in the 3 or above categories already. A related point here is that, if we really think information constraints are at the heart of the problem, it might be better to address this problem by addressing those constraints directly by using our supervisory authority to encourage and facilitate greater transparency. So my sense is that this facility would just subsidize borrowing banks without doing anything to mitigate underlying informational asymmetries or any other type of market friction that I can think of. That means to me that this proposal raises the usual moral hazard concerns. The staff memo was very clear and articulate about those. I think there’s a danger with this facility of raising expectations that, in the future, significant increases in interbank funding spreads are going to be ameliorated by central bank intervention. If we raise that expectation, we’re going to undermine to some extent market mechanisms for assessing the relative risk of institutions.
I’m a little worried that if this does not produce a demonstrable effect on relevant market conditions, it could erode confidence in us, and I feel so especially in light of our previous change in discount window policy, which I think is widely viewed as having had little substantive effect so
September 18, 2007 145 of 188
far. I think that’s the view out there. I also worry that this could complicate the resolution of failing institutions whose condition, as Vice Chairman Geithner suggested, deteriorates while they’re borrowing from this auction facility. It would put us in a very awkward place. As Governor Kohn said, this isn’t like a one-day emergency kind of thing—it takes some time. But some institutions in questionable situations, some problem institutions, look for term funding and are willing to wait four days to get it and know enough about their condition to line it up ahead of time. I worry about this sounding like a cloak for the ECB, for us to give money to the ECB, and I worry about President Rosengren’s issues as well, and I’d be more comfortable with the swap line than I am with the domestic facility. If those foreign authorities want to extend credit and have the knowledge and capacity to do so, and it’s on their dime and they’re bearing the credit risk and they want to borrow the dollars from us, I see that as a reasonable step for a central bank to take. I also worry about valuing collateral. I don’t think that our mechanisms for doing that are robust and strong, especially in the current environment with at least standard haircuts.
Now, I can appreciate the broader problem articulated by the staff and others that banks that are constrained in the term funding market might tighten borrowing terms for consumers and businesses and that might have real economic consequences. But if that’s the problem, I think it would be better for us to just cut the funds rate rather than alter the relative funding costs of different banks. This is essentially what we did today. We cut the funds rate to offset the macroeconomic effects of higher credit spreads.
Just a final set of comments. More broadly, I’ve been hoping for some time that banking policy in our country was moving slowly but surely toward greater reliance on market discipline and away from forbearance and subsidization. I’ve been hoping that we as a central bank would gradually move away from things that are tainted with credit allocation. Times like these don’t September 18, 2007 146 of 188
come around very often—you know, once a decade—and my sense is that the precedent we set here is going to be remembered for a long time and it’s going to affect market behavior for a long time to come as well. In my opinion, we ought to look at these episodes of market stress as an opportunity to make some reputational progress on the time-consistency problem that is at the heart of moral hazard. So for me the balance of considerations weighs heavily against this proposal, Mr. Chairman.
CHAIRMAN BERNANKE. Let me echo what you first said and congratulate the staff on an enormous amount of work and a terrific presentation. I do want to just say a word, President Lacker. Most of your arguments are premised on the idea that markets are basically in some kind of equilibrium and we’re just messing with the equilibrium. There’s a longstanding tradition, both in central banking and in theoretical analysis, that you can have periods when there is insufficient cash in the market and prices are, therefore, driven away from fundamentals by the lack of cash. What central banks do is provide cash against collateral. I refer you, for example, to Franklin Allen and Douglas Gale’s new book, Understanding Financial Crises, in which they present a model with exactly that property.There you have it. Lacker is thinking about induced fragility; Bernanke is thinking about inherent instability. Bernanke's discussion is at a high level too (note the literature citation), though I think his characterization of Lacker as only willing to think about "some kind of equilibrium," is silly. I haven't read Allen and Gale's book, and I'm not familiar with their model, but I'm sure they are thinking about "some kind of equilibrium" too.
The moral hazard issue is the following: There is a good reason to allow some deviation of cash-based valuations from fundamentals because it creates incentives for providing sufficient liquidity and rewards those who have sufficient liquidity. But beyond a certain range, the models as well as central banking experience suggest that, when you have a fire sale type of situation, then the central bank can be useful by providing cash against that collateral. Essentially this process would take $140 billion of undervalued, hard-to-sell, or illiquid assets onto the central bank’s balance sheet and provide term liquid funding in its place, which I find totally consistent with Bagehot and the traditions of central bank lending. Now, the question arises whether the market is in sufficient September 18, 2007 147 of 188
distress. If it’s not, then we’re in the first regime where some deviation of values from fundamentals is legitimate. If the market is in extreme distress, I think—I repeat what I said before—there can be situations in which markets are simply not functioning well and there could be a lot of reasons for that in which case the central bank could help the market function better. I agree that some of these issues about bid size and so on do confuse the issue. We need to make a strong distinction between helping these markets to function better—we can address, for example, some of the counterparty risk because we have all this collateral in our discount windows—and helping or bailing out any individual institution. I agree with you that we certainly want to avoid that perception if at all possible.
Here's the key issue. Some people - a plethora in fact - have been willing to tell us two things. First, they claim that existing economic theory is useless in addressing financial crisis issues. Second, they argue that policymakers failed to do anything in advance of the financial crisis that would have prevented it or mitigated its effects. My claim has been that there is plenty of off-the-shelf economic theory and evidence that can be brought to bear in organizing our thinking about the financial crisis and the recent recession. Indeed, a good part of that theory is what Jeff Lacker discusses in his recent speech, and uses in the FOMC discussion from September 2007 that I quoted above. Further, we can see from Lacker's 2007 discussion that he's concerned about exactly the right issues - with the benefit of hindsight, he looks pretty good.
If we buy into the induced fragility view in a big way, then we have to take what Lacker says in his recent speech seriously. Moral hazard in financial markets is not just a long-term issue, but can propel events during a financial crisis. According to this view, in 2007 the Fed began to make decisions - such as establishing the TAF program - that caused large financial institutions to relax and let the Fed do the worrying. Indeed, the assisted purchase of Bear Sterns confirmed that view, and then everyone was surprised when Lehman was allowed to fail. Can we make the case that the Fed's behavior in 2007 and later contributed in a big way to the severity of the financial crisis and the recession? Maybe. We should at least be studying this seriously, though I'm sure some of you can lead us to some interesting work that is already doing that.
1. The Fed's actions in 2007 did not contribute to the crisis for by then the loans were already bad.ReplyDelete
2. It would be better to study how truly bad all the loans were. We still have no idea how much damage was done, but the departure from trend is a compelling argument that the damage was far deeper and clearly far deeper than understood by Lacker or Bill Poole.
3. There is no evidence for induced fragility. Take deposit insurance. Who in the private world has the tools to judge whether to leave deposits with a bank? Thus, TBTF behavior is going to happen with or without insurance, as it did before deposit insurance.
We have deposit insurance merely because it is convenient and is valuable and useful in its own right. It does not change bank behavior.
A general rule:
"For any given free put written by the Fed to markets, markets will take actions to maximize its value such that the cost of honoring the put eventually becomes too high."
A correction to my first comment.ReplyDelete
Hot off the presses we now have an estimate of how off Poole and Lacker were.
They only missed by $22 trillion, per the GAO
http://www.gao.gov/assets/660/651322.pdf, as read by the Huffington Post.
It seems to me that the first question that has to be asked, is, If the Fed had let it all go down in 2007-08, how much additional damage would have been done beyond the $22 trillion that had taken place under the watchful eyes of Greenspan, Poole, and Lacker?
Now, what about this warrants serious study?
I don't think I agree with anything you're saying. For example,ReplyDelete
1. "There is no evidence for induced fragility." I would say there is no evidence for inherent instability. Canada is the counterexample.
2. "We have deposit insurance merely because it is convenient and is valuable and useful in its own right. It does not change bank behavior."
You're in the very small minority who thinks this. What do you think the savings and loan crisis was about? Regulations on S and L's were relaxed. They had deposit insurance. Regulators did not watch what they were doing. They took too much risk. Simple.
the S&L crisis had nothing to do with asset quality. It was all about deregulation of interest on deposits and withdrawals to money market funds.Delete
"there is no evidence for inherent instability"ReplyDelete
How do you interpret the recurrent financial crises before the Federal Reserve, FDIC, etc.?
The Canadian banking system is remarkably stable - back to the 19th century. During the 20th century there were 3 bank failures in Canada - one in the 1920s, and the failure of two smaller regional banks in the 1980s. There were no bank failures in Canada during the Great Depression, and none during the recent financial crisis, which the Canadian financial system weathered quite nicely. So, if banks and other financial institutions are inherently unstable, then the Canadian banking system should have been subject to recurrent panics. QEDDelete
The canadian encyclopedia attributes the stability you mention to improved banking regulation:Delete
"... until the 1920s banks in Canada were generally unstable. Between 1867 and 1914 the failure rate of Canadian banks was 36% as opposed to 22.5% in the US, costing Canadian shareholders 31.2 times more than was lost to American shareholders; of the 26 failures in this period, 19 resulted in criminal charges against bank officers or employees. Improved bank regulation reversed these failure rates and Canada has had only 2 bank failures since 1923 while the US has had over 17 000."
If there is no inherent instability, why did Canada's banking system show a high level of instability in fairly deregulated conditions, followed by enviable stability after well-designed regulation was passed?
The numbers may be right, but their characterization is wrong. The banking system in Canada prior to the 1920s was in fact stable compared to what you see in the US. A lot depends on how you measure instability. Here's a paper I wrote with a couple of other people:Delete
In that paper, we make a big deal of comparing Canada and the US over the period 1863-1913 (the National Banking era). A key feature of Canada over that period is that it doesn't exhibit the recurrent panic periods that you see in Canada. In Canada, runs don't happen, and we try to explain that in the paper. The Canadian monetary system at the time was actually a private money system, for the most part, and a successful one.
Thanks -- I'll take a look at the paper. I don't suppose there's an ungated version posted somewhere?Delete
It's on mmy web page:Delete
First, let me bow my head in respect for an excellent post that highlights the important constributions of economic theory.ReplyDelete
On the issue, I am more sympathetic to the Diamond-Dybvig view (I made similar arguments in our discussion of why financial institutions may have a hard time intermediating across maturities) than to the alternative view that equates banks with infinitely-lived long-memory agents that learn from their mistakes and act on the information. Finding solutions to complex problems takes time. If an institution fails in the meantime it loses the opportunity to learn, and those that survive may find it preferable to focus on the next big thing rather than continue to contemplate what went wrong during an event that happens once every blue moon.
To study what banks will actually do we must incorporate a great deal more of organizational theory, with assumptions about incentive structures and asymmetry of information between stockholders and depositors on one hand and managers on the other clearly laid out. For example, here is the new Barclay's CEO, Antony Jenkins, talking about a change in culture, whatever that means, rather than specific innovations:
This is not to say that induced moral hazard is not a problem that should be taken seriously. In the end it comes down to how much one weighs the cost of one versus the other.
I agree with most of what you're saying. However, Diamond-Dybvig obviously has no predictive power. It's a multiple equilibrium story, so it tells you nothing about where and when you should see panics. But the problem is that there are regularities in the data. The pre-Fed panics have strong regularities associated with them, and we have cross country differences to worry about (e.g. what I'm saying about Canada). So we need better theory, and I think you agree with that.Delete
Yes, I absolutely agree!Delete
Diamond-Dybvig is an ad-hoc modelization. The question is whether incorporating ad-hoc modelization into general equilibrium models is predictive. But then you must add heterogeneity, distribution of wealth, hypothesis about financial institutional frameworks. This leads to multiple equilibria -like it or not. Se Bernanke et al (1996:4).ReplyDelete
Your position is that, during the bubble, we were already in a structural recession. What we see now is a lower growth path and you believe that this is the natural without-bubble path -or a little bit worse because of FED policies. But why? The crisis have an independent effect as ad-hoc modelizations describe (financial accelerator, debt overhang). Without FED policies, we would have depression-like dynamics, not the less dramatic shift in the growth path.
Your position is Austrian ideology in disguise -disguised by your deep knowledge of modern macro.
The multiple equilibrium story as you call it is your nemesis: you can't defend Austrian ideology a.k.a. Mellonism if you really believe in modern macro.
Williamson has written about Ron Paul. Our host is not an Austrian. Nor of course was Mellon, since virtually no Americans had heard about the Austrian school back then (Irving Fisher excepted). Spend more time attacking Williamson's ideas, and less trying to associate him with boogeymen.Delete
I think you mean a model.
You seem to like to put words in my mouth. I certainly would not assert that multiple equilibrium phenomena had nothing to do with recent experience or the state of the world today. I don't think we have enough evidence to say yes or no, and you have to write down the model and show it to me to convince me. You seem pretty sure of yourself though. How do you conclude this:
"Without FED policies, we would have depression-like dynamics, not the less dramatic shift in the growth path."
You seem convinced. Why?
Evidence means having a model? Strange definition of evidence? Well, however, we have models on financial acceleration and we have models (yours) negating financial acceleration. But none can be considered 'evidence', models are descriptive -also in physics in a certain sense. We have eventually some interpretations about historical experience (Friedman-Schwartz), but still no clear-cut evidence.ReplyDelete
Without FED policies... where can I find evidence for a what-if scenario? I simply can't. But clearly Bernanke made his decisions based on models -probably his own-, that is, what you wrongly call 'evidence'.
Ad-hoc modelizations: Diamond-Dygbig points to the possibility of multiple equilibria in a quite idealized setting. Bernanke starts with a partial equilibrium setting and notes how hard is to move to a general equilibrium framework in a more realistic scenario than Diamond. I call the first type of model 'ad-hoc models' -built to show how a certain outcome is possible. More realistic, and then more 'predictive' models are hard to build.
I am not saying I am sure, but I question how you present your opinion. If you really believe that we still do not know, you should be less skeptical about stabilization measures. That means it's not theory which is driving your analysis. And please you are not the only one who understands modern macro.
Still thanks for the answer.
"Evidence means having a model?"Delete
Yes, you need the model in order to structure how you think about the data. You can't just stare at the numbers. You need a theory.
"we have models on financial acceleration and we have models (yours) negating financial acceleration."
That's funny. Do you know where the financial accelerator came from? You're thinking Bernanke-Gertler-Gilchrist, I think. In the 1980s there was Bernanke-Gertler and some other guy Williamson whose published work predates B-G by a couple of years. Gertler taught me, in part, but we learned financial intermediation together, and Bernanke, Gertler, and I used to hang out at the NBER in the summer in Hubbard's group, that also included Gary Gorton.
"Yes, you need a model in order to think about the data".Delete
In the previous post, you said something else: "I need EVIDENCE, give me a model". Which is either a mistake or a sign of epistemic closure.
To which, I replied. There are enough models out there (sorry, I am not paid to produce models, I have to feed my family).
So why don't you give credit to B-G-G or to your previous works? And, although important, financial intermediation is part of the problem (yes I know it's your area of expertise). How do we incorporate findings on financial intermediation -which are highly dependent on ad-hoc assumptions- in macroeconomic models and have reliable predictions of real fluctuations?
I never said that we don't need models, but models are not evidence. Otherwise you would have simply to decide which models you prefer eventually, for internal coherence.
The problem -which is at the epistemological level- is that models cannot be easily rejected. And I mean, not only micro-founded models, also non-structural ones. Models are hardly predictive in terms of behavior of complex systems because they are 'models'.
By the way, we have a lot of theories and models which makes sense of post-crisis growth rates and most point to the existence of welfare losses because of the size and intensity of a financial crisis. This convergence is quite suspect, it runs from Fisher to Schumpeter, from Friedman to Bernanke (or, as I learned today, young Williamson).
So, where's your evidence, sir?
P.S.: I read Gary Gorton on the financial crisis and my first reaction was "he is advising us to abolish shadow banking or what?" and "why is he against QE?".
By the way, Bernanke cites also Bagehot, quite revealing I think.Delete
We have two ad-hoc models within one paradigm, one pointing to fragility and moral hazard, one to instability and panic. Who is right?
Well, Bernanke says: economics did not start in the 70s, old Bagehot rules may still be valid. Let's also look outside the paradigm we helped to build.
What I like in this position is the absence of epistemic closure. Bernanke can be wrong, but he shows a more pragmatic attitude and a more mature epistemology.
"By the way, we have a lot of theories and models which makes sense of post-crisis growth rates..."Delete
"...old Bagehot rules may still be valid."
I once read Bagehot and did not find it helpful. For some reason people like to throw his name around. No idea why.
Put it this way, if Bagehot could read recent works on financial crisis, he would say "cool, this is exactly the same dilemma I was aware of" or, alternatively, "that's not much more helpful than what we already know, why people throw these models around".Delete
I opt for the first answer.
Anyway, your replies -and your post- really helped me a lot. I am the guy who reads everything and finds everything helpful. I think it is the right scientific attitude.
Thanks for your time.
Can we make the case that the Fed's behavior in 2007 and later contributed in a big way to the severity of the financial crisis and the recession? Maybe.ReplyDelete
Almost certainly -- but which behaviors had what effect?
The biggest, and still most un-discussed Fed/ Regulation/ Market failure, was in allowing AAA credit ratings to so many trillions of MBS / CDS / CDO "financial products".
a) There should be some upper maximum limit on AAA non-gov't debt. So that, at the limit, new AAA debt can only get issued by the rating agencies downgrading other debt.
b) There should be more liquidity/ turnover input in debt valuations -- less traded, less liquid debt shouldn't be AAA / eligible for use as Tier 1 capital.
But these debt issues, which I find convincingly important, seem to not come up in either the Fed notes nor most economist analyses of the pre-Crisis, Crisis, and post-Crises periods.
Williamson is referring to 2007 and later Fed behavior. He means QE. Sure, pre-crisis regulation -or absence of- can be considered mistakes. Or maybe not, financial innovation will always find its way and it can be destabilizing. The most interesting part of Williamson's analysis is, however, the assumption that there was a fall in real growth before the crisis and the system was trying to cope with that in its own destabilizing way. The question is what caused such a fall in the first place and the answer can't be financial innovation. Or maybe it was financial innovation -the growth of the finance sector (see Greenwood, Scharfstein).Delete
If only we could get rid of ideology...
Bill Poole is a known wingnutReplyDelete
WEDNESDAY, FEBRUARY 20, 2013
He'd gone full wingnut (on economics) when I was his TA, so this isn't really new to me.