An interview with one of the savviest derivatives users on Wall Street.

By Joe Kolman

Stan Jonas, 48, is a minor legend on Wall Street. Currently a managing director at Societe Generale/FIMAT, Jonas runs a desk of about 20 people whose clients are mostly proprietary traders at other dealers and institutions. He first traded in 1979 for Acli, a government securities firm specializing in cash/futures arbitrages. In 1987 he headed the derivatives group at Shearson Lehman, where he directed proprietary arb trading and institutional business. He joined Societe Generale/FIMAT in 1991.

Jonas is recognized as one of the savviest users of exchange-traded and OTC derivatives, particularly internationally. "He's an unusual combination of trader and thinker," says Vince Bailey Jr., a portfolio manager at New York-based BEA Associates. "He understands the big picture and has a knack for translating economic issues into trading profits." A typical Jonas arbitrage spread trade might be based on the difference in the volatilities between Eurodollar options with different maturities.

In conversation, Jonas leaps easily between trading ideas, global historical events and options theory with a breathless machine-gun delivery. He spoke recently with editor Joe Kolman.

Q:You've lived through a lot of different markets over the years. What happened in February 1994 and how have things changed?

A:February 1994 was the end of the euphoric bubble.

The huge pump priming exercise by the Fed turned this market into a speculative free-for-all. The Fed perceived that the banking system was in terrible shape. So they kept the yield curve very steep and didn't raise rates in order to give banks a chance to rebuild their capital. Bankers were attentive. They got the signal that the government was giving out a free put. They figured: We'll buy the two year notes and make money on the capital gain and then on the carry profits.

So all through 1993, and particularly towards the end, we had this huge rally. When bonds broke at the end of '93, everybody was long bonds. As the Fed tightened, the market recognized how closely concentrated liquidity had been. Everybody was in the same trade. Everybody was long - US bonds, German bunds. And nobody imagined that so many people could be long in such huge size.

Q:What do you think was responsible for the huge capital movements we saw?

A:Think about what went on in the mind of a typical equity oriented hedge fund manager in 1992. In 1992, he's got $3 billion in stocks. The Fed has given the signal to the banking system and somebody says he should be long bonds. So he figures, "OK, bonds are one third as volatile as stocks so I should be long $10 billion bonds."

A little later on, the yield curve is steepening, so he figures he doesn't want to be long $10 billion bonds, he should be long two-year notes. How many notes? On a duration weighted basis his quant tells him that he should be long about seven times as much, so he figures he should be long $70 billion in two-year notes. That's more than exist.

In 1992, the value-at-risk guys, looking at past correlations and volatilities, were telling them that French bonds and two-year notes were a comparable exposure to their current U.S. fixed income positions: Why not move to Europe, where the easing cycle had not yet caught up to that in the United States?

Many of the hedge funds made that kind of decision. It all happened very quickly. All the equity guys decided to get into fixed income. And the "worst" thing is...they made a lot of money. The Fed eased 24 times and kept on easing.

Hedge funds were terrifically leveraged. Guys like Moore Capital perhaps had $3 to $4 billion of hot money at times leveraged 100 to 1. Michael Steinhardt had so many Canadian bonds that we thought he was going to have to file a 15D...

Q:...to take over Canada...

A: Right, to take over the country. Hedge funds had more investment capital than any major bank. When the Fed began to ease aggressively in 1992, the financial world was completely different than it had ever been. With derivatives, you could make bets that were impossible to make three or four years earlier. You could buy French bonds at the MATIF, or gilts at LIFFE or do structured products with pay-offs based on the difference between Spanish and German rates. The whole world essentially became a futures market.

If a Martian came to the U.S. and looked at the universe of hedge fund managers, he'd see the same person. Many of these managers are interrelated by blood, by hobbies, by education. They're all competing, checking what the other one's doing.

Q: So when it came time to sell...

A:...There was nobody to sell to. You owned everything. Or somebody like you did. So all of the statistical notions of diversification didn't make sense because diversification assumes there's somebody else, a truly "significant other." 1994 was the first year in history when all global bond markets moved in the same direction and I don't think that's coincidental. When it came time to liquidate, it was just impossible.

Q: But why were the selloffs so sudden?

A: Most hedge funds make money as trend followers. The key to trend following is that if the market goes up you keep buying more. If you're right, you end up a big hero. Normally, if you make money, you're supposed to take profits. But when they make money, they effectively double up on their positions, so on any big move down, they're going to have a big loss.

Anthropologically, it is interesting to note how hedge funds attempt to sell themselves to their potential investors. Many of the biggest ones promise that they'll never lose serious amounts of money. They have developed brute force stop loss systems, so that, hopefully, no matter what happens they try never to lose more than say 3 to 5 percent of their value in any month.

So what happens? They're not in fact diversified, they're not hedged. As they experience losses in one marketplace, they start shrinking their positions in every marketplace because they don't want their portfolios to lose more than 5 percent. So after the Fed tightened, the European markets began to suffer. Then the Latin American markets. Those were all profitable positions at the time. There was nothing wrong with them per se. But they began selling because they had to shrink their entire position.

It becomes "global triage." The position is pared of the most liquid securities first, leaving the fund with the most difficult move on their books. It's very much like our mortgage funds that, when forced to, sell their liquid securities, leaving them only with the "toxic waste," to the detriment of their shareholders. You begin to see the total market move globally. You start to see strange corollaries. One day you're down in Germany and the next day it spills over to Mexico and the Turkish markets in rhythms tied to investors' preferences and risk aversion, not to macroeconomic events in the marketplace. This is cheap relative to that, Mexico is cheap relative to Spain. This shrinkage quickly becomes self-exacerbating.

Q: So what does all this teach us?

A: The lesson from the banking side is that the static notions of risk and implied volatility are crazy. You can no longer provide infinite liquidity to the marketplace based on theoretical models. You can have mathematical structures that work on a very localized level, but in the real world they will almost always break down.

How do you manage your risk if some firm like Quantum wants to buy volatility on the Sterling/Mark cross, prior to Great Britain's leaving the ERM? They're going to do so much of that trade that if they're right, the volatility you've priced the trade at is not going to accurately reflect the realized volatility. In many ways its like arguing with a 600 pound gorilla.

The guy who first had the opportunity to make a lot of money doing this was Mark Rich. How did it work? I buy options. I go to this bright young dealer. He's got his options pricing model, and I want to buy a call. He gives me the price because he's got a model and he figures they're liquid instruments, and he can always buy more and more to complete the hedge.

But I'm the gorilla. So I can buy a lot of options from him and from other dealers just like him. I know one thing. Every time I buy more, it sets off a signal to buy more. It's a positive feedback loop. I'm the gorilla. I can control you. I can eventually make you buy a lot of it. If I can control the marketplace to make you buy more of it, I can dominate you because I know your recipe for replicating the options. As Nassim Taleb said in his Dynamic Hedging, "The market will follow the path to thwart the highest number of possible hedgers."

Q: OK, volatility can be affected by liquidity and can itself be volatile. Do you have the same problems with correlation?

A: Absolutely. Dealers have provided us with the opportunity to track a new type of risk embedded in the marketplace: correlation risk. The pricing is based on relative movements of previously stable historical relationships. All the hedging technologies based on those ideas are going to break down.

Take a look at a value-at-risk system. What's its greatest weakness? It says that we can historically identify certain probabilities of moves in the market, and that we have a stable correlation matrix.

Here of course is where the VAR or any statistical approach is vulnerable. When potential loss distributions are fat-tailed, simulation based critical value estimates show significant biases and have standard errors of substantial magnitude. This is particularly significant when a portfolio's positions contain options. These distributions are a mixture of different distributions, and it becomes virtually impossible to verify with any accuracy the potential losses associated with extremely rare events.

Q: Ultimately it all comes down to a set of assumptions...

A: Maybe even a set of historically updated assumptions. But history moves, and it doesn't move smoothly. By very definition, if everybody's using the same assumptions, and you have everybody in these trades, the correlation can't be appropriate, because things aren't going to move historically. If everybody has to get out at once, the markets are so finite and the historical parallels are so dynamic that you take on a tremendous amount of risk - risk that is unquantifiable.

Ironically, it's the worst sort of empiricism. More to the point, it's like the worst sort of technical trading. "I don't have to know anything about the fundamentals, the charts tell me all...," that's the refrain of the lazy technician, and it's been carried forward by many risk modelers.

Q: In the last few months, I think we've all noticed an enormous new interest in risk management. Why?

A: What's really moving the marketplace now is that more things are getting marked to market. And that's moving the marketplace dramatically toward risk management. When you have more and more liquid securities...

Q:...and accounting that supports mark to market...

A:...and accounting or technology that supports the ability and necessity of mark to market. All of a sudden, risk management becomes the only thing that counts. Because before, if you could value things at book value, what do you care? "The thing's a ten-year hold." The thing's eventually going to pay off. You didn't care about intermediate risk.

Q: Do you have any basic rules risk managers should follow through this minefield?

A: Well, risk management has got to start at the top. The first thing you have to ask is: What is my goal? How much risk do I want to take? What risks am I interested in pursuing? The calculus is relatively straightforward if you can develop the goal. Is my goal to replicate a T-bill? I may decide that my job is not to minimize risk, but to take on a given amount of risk for a given amount of reward. If I have a speculative component, I have to be careful how much speculative risk I'm willing to take.

Then you have to find a way to limit the risk you're going to take on. You could say, I never want to see more than x amount of variance in my positions on a given day... say three percent. Once you have that as your goal, then you can develop the tools and instruments to measure and control that risk.

In a way, risk management systems are only valuable if they can handle the "wings" of distribution. The outliers, those fat-tails, where somebody, unfortunately, has to live.

Q: How do you make sure traders abide by a particular risk management system? How should you pay them?

A: You have to devise a system that rewards the types of risks you want to pursue. You shouldn't try to overlay a value-at-risk system on top of a traditional payout system. If you tell a trader, "I don't care how much money you make, but I care how you make it," if you compensate people for the quality of the risk they take, not the absolute magnitudes, then the people who are working for you are not going to have an incentive to sell all these crazy options, literally and figuratively.

The flip side is that you should understand what your risk goals ultimately mean and have those goals supported politically throughout the institution. Let's look at a corporate treasurer. You have to determine: When is he doing a good job? The first rule is that if he hedges he's going to lose. Is the corporation willing to understand that by diminishing risk, it'll have losses on its books? Because if interest rates go down and he hedges against a rise in rates, somebody will always come back and say, "Interest rates went down and you hedged. Didn't you know interest rates would go down?"

Here's rule number two: Deal only with instruments that are liquid and where you can find a tradeable price from several different sources at the end of every day. Or even sooner, with live prices. Value-at-risk depends on being invested in structures with prices you can apply. Your value-at-risk system won't matter if you can't say what a particular thing is worth. You should be able to see price history, to be able to estimate volatility. That enables you to provide the feedback into your risk management system so your positions are easily understandable.

Rule number three: If you can't simulate the characteristics of more complicated instruments on an Excel spreadsheet, don't do it. You, not somebody else in your office. If you can't sit down and simulate what will happen in different scenarios, if you can't do that by yourself, you shouldn't be in the instrument.

Q: Now, of course, you can buy models that...

A: No, that's not enough. You have to be able to replicate it on your own. You have to understand the dynamics of the model, what the ultimate payoff is in that security. But good modeling programs are very helpful because they can help you see the risk in a security, and that's critical to being able to understand more complicated securities.

The fourth rule, the most important one in risk management, is making sure your portfolios are marked to market. When you don't have portfolios that are capable of being marked to market, risk management is irrelevant, a Platonic ideal.

Q: What is it that people like about derivatives? What's made derivatives so popular in the US?

A: A lot of it is the regulatory benefit. Derivatives have achieved such a rapid growth for reasons beyond their ability to compete in the marketplace. If we can replicate the payoffs of derivatives with a combination of linear securities and borrowings, why then is there a market for prepackaged derivatives? From a cynical point of view, there are three reasons, not in any particular order of priority: 1) regulatory; 2) the ability to disguise leverage; and, of course, 3) the recognition of economies of scale on the part of dealers willing to price and offer derivative securities.

Q: The capital requirements certainly encourage you to move all this stuff off balance sheet.

A: Yeah, but if your positions are not marked to market, you're not going to care about value-at-risk. When you understand that the market has a real impact on your positions - that's when you start to look at value-at-risk. Mark to market forces you to reevaluate your opinions. If Bob Citron had been long the equivalent amount of Eurodollar futures instead of these structured notes - if he was long 40,000 Euros and every day he got a margin call - that would force a different type of confrontation of the issues. It no longer becomes an intellectual/analytical debate, whether the losses were paper or real.

If you let a guy take on a position to sell OTC options without a mark to market component, that can lead to the greatest of human weaknesses, hope. "It can still come back. Maybe it'll come back." The irony is that net net, someone like Citron would be flat to up today but it doesn't matter - when you're leveraged, hope is the worst thing in the world because "eventually being right" doesn't count.

Q: You'd like everybody to trade futures...

A: Futures are not the only instrument. I'd like them to trade options, too. Here's a proposal originally developed by Sanford Grossman I'd like to see implemented. I'd like to arrange things so that all the major trades that take place over the counter are effectively known in the marketplace. Not the participants, not who and not why, but the price and structure.

Then the trade must be broken down so that we could determine somebody today traded this correlation at say .80. He bought a Deutsche mark/Spanish peseta option at x price. Then you'd analyze that x price. What implied correlation is that? What volatility is that?

Q: What would that do for you?

A: Rather than using some historical estimate of volatility and correlation, one would have the market's current estimate of those values. In times of stress, the market's judgment of implied volatilities will enable you to gauge your current risk as well as to price your derivatives portfolio properly.

Well, now if you had a value-at-risk system, and you had a similar instrument that you didn't know how to value, every day it would go on the books and you could mark it to market every day. Then it becomes public domain.

Q: But why would the OTC market ever agree to do that?

A: It would or could be mandated. I told you this was hypothetical. From their point of view, nobody would care. Or they could contribute to a generalized clearing function. This is where the world's going to go in any case.

It seems that the world is progressing to this point anyhow on a piecemeal basis. The current environment makes it imperative that firms provide this data to their customers, and there are literally dozens of risk management groups that collectively try to gather this data and provide pricing. Rather than having dozens of people all doing the same thing, let's economize. Real values will eventually be disseminated in the marketplace.

Q: It sounds sort of like FLEX options.

A: Yeah. I have a similar arrangement with swap shops right now. When we sign a contract to do a particular transaction - a swap, a cap, whatever - it stipulates that I get a tear-up value every night. They tell me the price every night. We'll argue about it, but we'll agree upon a price every night. If one transaction moves in my favor, they send me money. If it moves in their favor, I send them money. Eventually most transactions will be commoditized.

Q: You do that with everyone?

A: We do it with everyone.

Q: That's pretty unusual.

A: No, more and more people are doing it. It's to everybody's benefit because it cuts the credit risk down. We're turning OTC dealers into futures markets. The only people who don't relish it are the top credits. Because it diminishes the benefit of your triple-A. But the market is becoming so competitive now...

Q: You'd recommend that everyone do this.

A: Absolutely. It's imperative from a risk management point of view. One of the most important things that risk management people don't take into account are the problems of liquidity.

From an arbitrage perspective a totally balanced position could put me out of business. In principle, its not much different than MetalGesellschaft... My two positions are ostensibly identical, separated only by a difference in time, but at the margin that is a critical difference.

Take a typical trade we might do. I buy a cash bond, which is effectively receiving fixed from the Government, and I enter into an interest rate swap, in which I'm paying fixed. It's a simple asset swap. The bond begins to drop in price, so I have to give $100 million to my bond dealer because my government collateral is marked to market. How do I get money out of the swap?

I've got to go to that swap dealer and say I want to reprice that swap, which is worth more. We mark it to market, the swap dealer sends me money that I send to the bond dealer. That's what's going to happen.

Q: The Chicago exchanges are trying to set up something like that.

A: The dealers will effectively do it themselves. That's because we're arbitrageurs. We can't trade without liquidity.

Take the example of a $200 million fund I'm familiar with. We trade only these arbitrage relationships with 15 to 1 leverage. The trading is "risk-free," but if the market levels or moves 20 percent, we don't have enough money. So we develop tear-up arrangements with swap dealers. We say, "Listen, we're big customers, we're in every day, you want to do it or not?" They say, "No," we say, "OK, I'm sorry. We'll still come to you eventually. Because we'll go to a place that will do the trade, which is a place like General RE, which will do tear-ups, and later on we can transfer the position between you two guys. But we're not going to do business with you directly unless we can do tear-ups." That's going to be the uniform way and that's the way we're going to have the components of a value-at-management system.

Q: What do you think of RiskMetrics?

A: I think what JP Morgan has done is brilliant, a tremendous job. And they keep on doing it. The work is technologically very sound. It doesn't solve all the problems, but they address all the problems in a very clear way, and anybody who wants a good education in risk management should read the technical document that JP Morgan publishes. It's well written, it goes through all the math, it's not overly complicated, and it outlines what they're up to. And then as people talk about it and complain to them about the assumptions, they actually go back and they rebut and they argue. It's the best thing I've ever seen any institution do in this field.

They're very clever. If you have these positions, they give you the pricing for everything. One place, one stop. Maybe the exchanges do it, maybe Bankers Trust does it. You have everything. Interest rate swaps, French bonds. I've got currency positions. I've got Turkish lira forwards, and at 3:00 New York time you tell me how much the thing is worth and how much it's moved from yesterday and what my potential risk is. The first thing it tells you is how much you made or lost. But you also have some historical data to say exactly what your positions are, what the historical risk is. In a large number of these markets where we have options, there's all this information: what the implied valuation of your positions are. Then the risk management is basically arithmetic.

Q: What's going to happen when the market becomes more sophisticated about risk management? How is that going to change things?

A: You have to understand an important thing: Most financial engineering is a simply a technique of burying the cost and taking advantage of human nature. With derivatives products and options you can pick any point of the payoff distribution and sell off all the others.

The problem is not that people buy derivatives, it's that people tend to sell them. It can be very deceptive, and you have to understand what you're selling. Because implicitly what happens is that in order to pay for a high yield, they're truncating some part of distribution. They offer you high yield today in exchange for giving up something in the future, something that may seem very outlandish or realistic. But human nature is such that we always believe somebody else is going to get the tail of the distribution. People tend to look at things nearby as much more probable and they discount the probability of outside events much too much.

The sales pitch for some of these securities is that in the worst case you're going to get your money back. "You may lose the 6 percent return, but you're going to get your capital back." But that's deceptive. Because return of capital in our world is, of course, investing at no interest.

Value-at-risk is very critical because that kind of structure would show up as a very risky transaction. If you have a VAR mentality and you use the proper software vendors, you will eliminate in the next generation that ability to bury the risk in the books. You're not going to be able to bury those risks. They're going to show up, and that's going to change the way we look at what we're doing. It's going to stop 80 percent of the intellectual shenanigans.

Q: But is it going to stop trading losses?

A: No. The trading floors on Wall Street have much more sophisticated risk management systems than anything a corporation could invent. But you have the same problems. The same losses. There's no panacea.

Of course, it's those extremely rare events that either make or break us, and that's when we truly need a risk management system.

In the meantime, as more and more institutions adopt the VAR framework, which is not too different from what the Street has been doing for nearly a decade, we'll find traders becoming experts at arbitraging the failures of the VAR they are subjected to.

Q: So the conclusion seems to be that all the risk management modeling in the world isn't in itself going to protect you from the vicissitudes of the market.

A: George Santayana talks about the fallacy of misplaced concreteness. Look at all these option-adjusted spread models. They look so real and palpable and they produce real numbers. But people take them so seriously. The real issue is that over the past five years, risk management systems were in place and we still had disastrous consequences throughout the Street.

Look at mortgage-backed securities. There's been no area of the marketplace in which more analytical and computer work has been done. But it's all been to no avail. All these smart people, models, all of the risk control measures they thought they could do, the second-largest fixed income market in the world next to the government market - and you nevertheless had disastrous consequences.

You have to be aware that there are certain basic risks that can only be handled on a very simple level. How much money am I willing to commit to any strategy, and when the strategy goes wrong, am I willing to stop myself out and to look at it again? That's one thing value-at-risk doesn't let you do. The risk is not always continuous. You have to know when to decide if you're really wrong.

After all, one of the premises of trading is that there are driving fundamental factors in the marketplace. Life really isn't a draw from a random distribution with or without drift. When circumstances change, stable correlations disappear. When the Fed tightens, and my positions are long duration, it's not the volatility of the position that will worry me. A slow steady increase in rates will destroy me as well if I sit and wait for the conclusion. The speed of my demise is of less consequence than its certainty.