Overvaluation Models Transcript

Jekyll Island Conference

November 6, 2010

Charles Davidson: Welcome to the podcast of the Jekyll Island Conference sponsored by the Federal Reserve Bank of Atlanta and Rutgers University. We're in Jekyll Island, Ga., at the Jekyll Island Club Hotel with conference panelists who gave presentations on the subject of asset bubbles. Moderator Eric Rosengren, president of the Federal Reserve Bank of Boston, talks with presenter Narayana Kocherlakota, president of the Federal Reserve Bank of Minneapolis, and panel discussant Anil Kashyap, professor of economics and finance at the University of Chicago.

Eric Rosengren: Could you describe in some detail what your model describes and what kinds of implications it has for policy?

Narayana Kocherlakota: Sure, Eric. Thanks a lot. So, I think that the thrust of my model is that when you have credit constraints in an economy such that the borrowers have difficulty fully capitalizing their future income, that tends to drive down interest rates. And when those interest rates get driven down, we can have, bubbles can emerge in the prices of assets. Those bubbles have a good feature in that they're providing a vehicle for savings that otherwise wouldn't exist. The creation of borrower constraints means there aren't as many people to lend—there's not as many funds for lenders to access, and the bubbles create a larger pool for that. Savers are generally made better off by the existence of bubbles by having a larger pool of savings and having a higher interest rate on those savings.

The bad part of bubbles is they're unstable. So the way the bubble is working is, I'm saving today by holding this asset. It's overvalued with respect to its present value of future dividend payments. I'm only going to hold it, though, because someone is going to buy it from me at some higher value in the future, or I expect them to, at least. But that expectation can be unstable. If people in the future lose confidence in their ability to roll that asset forward, then you're going to have instability in the value of the bubble.

A bursting bubble can have deleterious consequences for an economy. Essentially what it's going to be doing is shutting down the supply of savings that I referred to. That's going to lower the interest rates; that's going to cut down labor supply, also making it more difficult for entrepreneurs and others to self-finance valuable investment projects.

My paper points to, I think, a natural fiscal response to a bursting bubble, where the government steps in by issuing some debt. Essentially, that debt can substitute for the bubble—the government can roll that debt over by repaying it, by just issuing new debt, just as the bubble, essentially, is rolled over in the same fashion. That means the government, essentially, gets a free lunch, in some sense—in the model, at least. And they're able to issue debt without having to raise new taxes in the future. That debt issue can be used for a variety of ways to enhance output in the current period, and I describe several of those in my paper.

So I think the punchline for my paper is that bubbles are a natural response to certain kinds of frictions that exist in credit markets. They have benefits. On the other hand, they also have costs. And in the aftermath of a bursting bubble, I think there is a natural fiscal response.

Rosengren: Anil, in Narayana's work and as well in your work, you've highlighted the role of leverage and the importance of leverage in thinking about asset bubbles. Do you want to talk a little bit about your own views and how it works in Narayana's model as well?

Anil Kashyap: Yeah, well, in his model—well, there are two models; he was talking about the second one. In the first model, he had a pretty interesting environment where the fact that some people have access to subsidized credit can lead them to, essentially, overpay and drive up the price of an asset, and the mechanism by which they overpay is to take on leverage and to borrow. And I think that observation is pretty general and can arise in a lot of models besides the one that he sketched.

I think the challenge is that we don't really have a consensus on whether the Fed would want to get into the business of regulating leverage. I mean, it's going to be much, much closer to doing credit allocation policies that the central bank has historically steered clear of, but we've seen that the consequences of a busting bubble are pretty severe, and failing to pay attention to this can be very, very costly, so I think this is a direction that we're going to have to go. There's also no consensus as to how best to do it. Right now I think of our regulatory system as being set up to track institutions, keep track of institutions as opposed to functions or assets. So, if you arrange to buy a house through a bank, where the bank holds the loan on its balance sheet, there's one set of regulatory restrictions that have to be satisfied, and there's a certain amount of capital that has to be present, but if you end up having that thing securitized—the same loan, securitized—and then perhaps sold off, maybe to a hedge fund that's going to hold it and finance it through a short-term repurchase agreement, the capital charges are very different. Yet the underlying risk in the house is all still the same, and the potential for instability, if there's a big number of these housing loans that fail, are just as real. And I think we're going to need to move to a regulatory system that recognizes that. I think a shorthand way of thinking about it is to say maybe we need to have kind of asset-specific, rather than institution-specific, capital charges. I think that's coming, I hope, but it's going to take a lot of further research and a lot of discussion in the policy community.

Kocherlakota: If I could jump in. I couldn't agree with Anil more on this. I think the first model you mentioned was about how guaranteeing bank creditors can lead to the overvaluation of land, which is like four or five links removed from them, in particular. They're lending to a bank, the bank is lending to a buyer, and then the buyer is buying a house. It is absolutely true that what you want to know about is, what's the cumulative leverage on any given asset? And that's going to be the key variable to think about in terms of doing regulation. That's a challenge for the Fed, I think, but it's a challenge more generally for the country when you're thinking about the Dodd-Frank apparatus, to be thinking about regulation in this, I think, more imaginative way.

Rosengren: So, when people look at asset bubbles, some have highlighted the role of monetary policy, but your model seems to focus more on supervisory policy as a possible way to think about addressing asset bubbles. Do you want to talk about that a little bit more?

Kocherlakota: Sure, as Anil mentioned, I had two models in my paper. But I do think they work together, so that one kind of bubble that emerges—the one I was describing to you earlier, just due to the scarcity of savings vehicles—and that, I think that the right response to that, if there is one, is coming from the fiscal authority and not so much from the central bank. That kind of bubble can be magnified, I think, by the kind of guarantees that creditors might imagine themselves to have when they're lending to a bank or other financial institution. So I think that you want to keep both sides of this in mind.

The second paper, the second model I should say, is more a macro kind of approach, and there is a macro response. I don't see that as being naturally coming from monetary policy. It's more about the total of government liabilities, which is really a fiscal decision, not a monetary decision. But then, given that, I think there is a real need to keep track of leverage and think about that. And wherever the source—basically, the first paper emphasizes whatever the source of tail risk in land values, be it coming from a bubble or any other kind of, maybe even from news, whatever that's coming from, that tail risk, is something regulators have to keep track of, and not just, as Anil would emphasize, not just at an institution-specific level, but more at an asset level.

Kashyap: Can I add something to that? I think Chairman Bernanke and the concluding panel made I think a very good and wise point, which is that we've kind of seen that equity markets can collapse without doing nearly so much damage as when you get a big expansion in the credit creation mechanism, when you get that kind of buildup. And I think that's my reading of the evidence as well.

The problem, of course, is equity markets are usually trades between two consenting adults—you know there's somebody selling and somebody buying, and nobody really thinks very much about regulating that, or maybe we restrict the ability to borrow on margin to buy stocks, but that wasn't central in the dot-com experience. It's just looked like euphoria or something that drove these prices to levels that weren't sustainable, and when it collapsed, it was okay.

The difficulty with monitoring credit booms is there are a whole bunch of entrenched interests that really want to see the boom continue, and taking the punch bowl away from the party is always harder, I think, when you're trying to slow down the credit expansion. I think that's something that's going to be a communication challenge for the regulatory community to remind everybody that while you're doing it, it's going to be incredibly unpopular. Unlike monetary policy, there's going to be clear winners and losers who are going to suffer when they can't get the credit, and they're going to complain, and it's going to be a challenge to say, you know, "We're trying to save you from yourself." I think that's going to be a big challenge moving ahead.

Rosengren: Final question for Narayana, which is, based on what you've already done and what we've seen over the last three years, where would you take research as it relates to asset bubbles going forward?

Kocherlakota: Well, that's a great question. I think that we have to be willing to explore a wide range of possible models. I wrote down one. Anil mentioned something in his discussion based on heterogeneity of beliefs and how that can feed into, lead into overvaluation as well. I think we're in the early stages, frankly, of understanding bubbles and how we can think about them best by embedding them into macroeconomic frameworks. I think Chairman Greenspan alluded to it in his discussion today that I foresee many PhD theses being written [on]. I think trees will be saved because it will be virtual, but I do see a lot of PhD theses being written on this topic. I think, you know in a five- to ten-year span, we'll have more perspective from an empirical point of view, but also I think a lot more knowledge and start to winnow down a class that we can agree on.

Rosengren: Anil, some final comments?

Kashyap: Yeah, I agree with what Narayana just said, that it's too early to tell, but I've thought for some time that, just as a broader challenge for macroeconomists—for a long, long time, we've been working with kind of these representative agent models, and somebody will get a Nobel Prize for coming up with a tractable way to model heterogeneity. Just even two types of people that we can carry around in some way without complicating the models too much. Usually what happens when you do that is then you have to keep track of something like the wealth distribution of the two types, and it becomes impossible, essentially, to get easy analytic results, and so you end up going to a computer and solving the model numerically and that kind of thing. That has limited some attempts to model phenomena because you understand you need the heterogeneity and you don't have a good way to do it.

I think that will perhaps be the next big revolution in macro, is when somebody figures out how to do that, and we can go back and revisit all these questions where we've essentially been shoehorning the model with a representative agent to analyze a question where we inherently want heterogeneity. So I think Narayana's right that that's going to be a topic for research going ahead, and it can link up to a bunch of other areas.

Rosengren: Anil and Narayana, thank you very much for a very exciting session.

Kashyap: Thanks a lot, Eric.

Kocherlakota: Thanks, Eric.

Davidson: This concludes our podcast of the Jekyll Island Conference sponsored by the Federal Reserve Bank of Atlanta and Rutgers University. Thanks for listening, and please return for more podcasts. If you have comments, please e-mail us at podcast@frbatlanta.org.