Chapter 3
One-on-One Interview with Oldrich Alfons Vasicek
By Nina Mehta
Financial Engineering News, May 2005
Oldrich Alfons Vasicek, a mathematician, is one of the leading lights in modern finance. His 1977 paper, “An Equilibrium Characterization of the Term Structure,” described the behavior of interest rates across maturities and paved the way for the development of the interest rate derivatives market. He is one of the co-founders of KMV Corp., the granddaddy of credit risk modeling firms, which Moody's Investors Service bought in 2002 (and renamed Moody's KMV). Throughout his career Vasicek has been at the forefront of credit risk modeling. Earlier this year he increased his stockpile of awards when he was named the 2004 IAFE/SunGard Financial Engineer of the Year. The author of more than 30 papers in mathematical and finance journals, Vasicek received a PhD in probability theory from Charles University in Prague, in what is now the Czech Republic, in 1968. This interview was conducted in February.
FEN:
You've worked on models to value credit risk for a large part of your career, but you said recently that valuing the plain debt of a single issuer is more challenging. Why?
Oldrich Vasicek:
It's a very basic question we need to ask and answer with a higher priority than asking how the derivatives are priced. Derivatives are secondary instruments. I'm not a great fan of so-called reduced-form models, which assume you know how to price debt—namely, that you observe the market prices of plain debt of various maturities—and once you have that, you price the credit default swaps and other options.
We need to understand how the more primary instruments are priced. In theory, we have the methodology available to do so—the Merton-Black-Scholes approach to valuing corporate liabilities. A company's debt is, after all, an option on the company's assets. If we know the terms of the option—that is, the financial structure of the firm—and we know the market value of the firm's assets, which we can infer from the stock price, then the derivative asset pricing theory should enable us to come up with the value of the debt. That approach been around for 30 years, but applying it is not a trivial problem because the firm's liability structure is usually very complex. There are various contingencies that are hard to describe and analyze mathematically.
FEN:
Has this problem become more pressing in recent years because companies can now issue complex hybrid instruments across the debt-equity range?
Vasicek:
It's always been a problem. A perfectly theoretically and practically complete solution has never really been available. It is more difficult because of new instruments constantly being invented and brought into practice. Financing now includes much more complex instruments than what was in use 20 years ago.
FEN:
Credit models have become more and more sophisticated as the market has grown up around them. In your view, is there some area of insufficiency in credit risk valuation?
Vasicek:
I'm not sure whether the existing approaches to credit valuation actually take into account correctly the possibility of very atypical catastrophic events that have not happened for the last 50 years, but that may be lurking in the future—such as a general collapse of the whole economy or a Great Depression or something of that nature. Empirical works that measure the default probabilities of various classes of credit risk typically use data that don't cover the occurrence of any such events. People may be a little too optimistic in measuring the probabilities of default if inferences are based just on data covering periods of relative calm.
FEN:
So model prices for debt instruments don't correctly include the possibility of extreme events occurring?
Vasicek:
No. Look at it this way: To protect himself against extreme losses, a lender may buy a way-out-of-the-money put on the company's stock. But if there is a general economic collapse, chances are that the issuers of the put will be unable to honor their obligation, since they will themselves be bankrupt.
On the other hand, I think that the market pricing of debt actually does take these kinds of things into account—that would explain the discrepancy between the probabilities of default, particularly on high-quality credits, and the actual pricing of debt for those companies. Typically, for high-quality corporate credits the spread over the riskless rate is higher than the measured probability of default. Part of the possible explanation is that the bond market incorporates the possibility of extreme events occurring in the future, while models do not.
FEN:
You're talking about high-quality debt because there are fewer other reasons that would explain the excess spread?
Vasicek:
Yes. The same thing would be true for low-quality debt, but it would add a relatively smaller amount to the risk premia.
FEN:
I know you've cut back your work in the credit area in the last few years, but let me ask you where you think credit modeling is heading. What kinds of questions should credit modelers try to address?
Vasicek:
A lot of attention goes to the pricing of various complicated debt instruments because those instruments are becoming more common. That's needed short-term. I think long-term it's important to understand the more basic problem we were talking about before—what exactly goes into the pricing of the straight debt of a firm. That's the economics of credit, not the valuation of assorted derivatives. There is too much mathematics and too little economics in finance nowadays. That may sound funny coming from a mathematician, but nevertheless that's my opinion. We must not forget that the subject of finance is economic decisions.
FEN:
You have a paper that's just coming out in the Journal of Financial Economics called “The Economics of Interest Rates.” That's an example of what you're talking about, since you're focusing on the economics rather than the behavior of interest rates, which is where you initially made your name in finance. What does the paper say, and why is it interesting?
Vasicek:
The area we somehow lost sight of, and should return to now, is economics. Twenty or thirty years ago it was the focus of a large portion of the work in finance. A prime example of an early work that went in that direction is the Cox-Ingersoll-Ross paper that was published in 1985. You can view this recent paper of mine as a generalization of that approach.
What Cox, Ross, and Ingersoll did at the time is they said, Where do interest rates come from? They had a formal description of the economic opportunities in a society on the one hand, and, on the other hand, a quantitative description of the preferences of the participants—preferences for current versus future consumption, the degree of risk aversion, and so on. If you then impose the market clearing conditions that ensure the existence of an equilibrium, you can see how the prices of various assets and in particular the prices of bonds—that is, the determination of interest rates—would happen in that economy.
But they assumed that everybody's preferences are the same, that you have homogenous investors in the economy, which is a limiting factor in the analysis. For one thing, there would be no borrowing or lending in that economy and the bond market wouldn't exist, because if everybody had the same preferences, everybody would hold the same investment portfolio. Besides, in reality, investors do have different preferences. So I have attempted to extend the general equilibrium models by incorporating heterogeneous preferences among investors. I'd been working on this problem for three or four years, but the actual breakthrough that allowed me to come up with the solution occurred about a year ago.
FEN:
What was the breakthrough?
Vasicek:
The main difficulty in dealing with heterogeneous investors is that the society's average, or aggregate, attitude toward risk and toward present versus future consumption shifts through time. This is because the investors' wealth levels change due to their different investment portfolios, and that gives different weights to their individual preferences. It turned out that the behavior of the individual wealth levels is driven by a single stochastic process, which gives you a mathematically tractable way of characterizing the development of the aggregate preferences.
FEN:
Let me go back to an earlier point in your career. What exactly were you hired to do at Wells Fargo?
Vasicek:
The Management Sciences Department of Wells Fargo was trying to look at the implications of newfangled theories such as the Capital Asset Pricing Model in the banking and investment practice. I was hired to be part of that because these models are highly quantitative and a mathematical background is definitely useful, even though they are essentially economic theories. CAPM was just making it into existence at that time. In fact, it was very slow in getting accepted. When I started at Wells Fargo in early 1969, most people in the bank still thought it was nonsense to measure risk by the variance of returns.
FEN:
You went on to do other research, and in 1977 you published “An Equilibrium Characterization of the Term Structure.” Were you already thinking about interest rates at Wells Fargo?
Vasicek:
No. It was when I was at the University of Rochester in New York, and then when I came back to California in 1976 and was teaching part-time at the University of California at Berkeley and doing part-time consulting for Wells Fargo that I got most of that work done.
FEN:
What drew you to the term structure of interest rates?
Vasicek:
At the time, there was very little available on the pricing of riskless debt and on interest rates in general. There was CAPM, which addressed the risk-return relationship of financial assets, but that was mostly applied to equities. There was the option-pricing theory for the pricing of derivative assets. But there was not much available on interest rates—how they behave, why they behave that way, what the relationships are among interest rates of various maturities. Some empirical work was available, but there was no theory of the term structure, so there was a void or gap. I was bothered by the gap, so I started to look into that.
FEN:
And bridged that gap. In 1983, you and John McQuown formed Diversified Corporate Loans, a high-minded but relatively short-lived venture. At that time mortgages were resold but commercial loans stayed put on the balance sheet. You were essentially trying to restructure a traditional business.
Vasicek:
Right. For some larger loans there were multiple lenders and there was some packaging and selling of loans, but that was not enough. We wanted to give banks an opportunity to diversify their portfolios.
FEN:
McQuown came up with the idea of pooling loans, right? Did he have to sell you on the concept?
Vasicek:
It was his idea but it made immediate sense to me. It was a meaningful thing. If you have an equities market where securities are efficiently and easily traded, any institutional investor can get a reasonably diversified portfolio. But it is much harder to get a well-diversified portfolio in a less-liquid market like bank loans because you do not have access through origination to the whole market. It's not good if a portfolio is concentrated in a specific industry or geographical region. Short of an efficient and extensive bond exchange, like a stock exchange—which was then and still is a ways away—the next means of allowing banks to diversify their loan portfolios would be for them to sell off a part of their portfolios to a pool and get back a share of the pool. That's what we were trying to do.
FEN:
But banks were reluctant to do that because they thought it would hurt their relationship with borrowers. Was that surprising to you at the time?
Vasicek:
I was not involved in visits to banks. I was trying to get a first crack on the valuation of the debt, trying to come up with what was later extended into the KMV approach. It was obvious that a condition for such pooling of loans across banks would be having a reasonable understanding of the value of those loans. These were privately held instruments by the originating bank. There was no secondary market. What was needed was some measure of the default risk for the pricing of these loans. I was spending my time trying to come up with a theoretical methodology for doing so.
FEN:
The modeling you did at KMV Corporation was a continuation of what you were doing at DCL.
Vasicek:
Yes, it was a continuation of that same effort. The early experience of the Black-Scholes-Merton approach to valuing debt had been so successful in actually giving an early warning of bankruptcies that when DCL closed, we thought a credit risk valuation methodology might have some value of its own.
FEN:
When you, McQuown, and Stephen Kealhofer formed KMV, was there an end point? How did you expect the market to evolve?
Vasicek:
It seemed to me that the approach made sense. If you develop something that makes sense, there should be an interest in it. KMV was a collective effort. I did my part on the theory of credit valuation. Steve made the approach applicable to real firms and real data. Mac kept track of the conceptual goal of these efforts and went to banks like a missionary to the native tribes to convince them that these crazy things could actually have some value for them.
FEN:
KMV's EDF (expected default frequency) measures and quantitative models have long been considered a counterbalance to the more qualitative models used by rating agencies. As rating agencies began buying credit risk modeling firms, there have been grumblings that there's been a reduction in terms of the different approaches to credit valuation. Is there any validity to that?
Vasicek:
I don't think that happened. Moody's has consciously kept a separation between the tradition of letter-rating methodologies and the KMV approach in order to have independent viewpoints on credit valuation. If anything, there have been lots of new models coming out in the industry.
FEN:
What do you think credit rating agencies and modelers learned from Enron, Worldcom, and other bankruptcies?
Vasicek:
The Enron case demonstrates the power of market-price-based measures of risk. The KMV model was giving warning signals before concerns found their way into ratings and bond market prices. Ratings often don't pay sufficient attention to market prices—by prices, I don't mean just the value of the stock but also its behavior, its volatility. If there's a lesson from the Enron case, it is that measures such as stock price volatility should be more explicitly incorporated into the criteria for forming letter ratings.
FEN:
I read something interesting a couple of days ago. CreditSights, an independent credit analysis and research firm, recently looked at Venezuelan and Brazilian sovereign debt and pointed out that Brazilian debt had higher ratings than Venezuelan debt. However, Venezuelan debt looked stronger, based on 13 out of 15 quantitative indicators. CreditSights observed that subjective factors often wind up trumping quantitative factors in assessing the credit quality of sovereign debt. What do you think?
Vasicek:
I do not know of any theoretically satisfying model for sovereign debt, frankly. So far, we've been talking about corporate debt for publicly held corporations. For those kinds of debt instruments we may not have perfectly worked out models, but we do have an approach—the derivative asset-pricing model. For sovereign debt, what's available right now is a combination of empirical statistical-type models and subjective judgment. There's no self-contained theory or model that exists for the valuation of sovereign debt.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.