In an Uncertain World (6 page)

The first lesson is that our ability to address economic crises beyond our borders is limited. The money we lent to Mexico could not have had the desired effect without the policy choices the Mexican government made. This was the crucial element, both because of the effects of individual policies—especially on interest rates—and because of the confidence engendered by the more amorphous cumulative sense that the Mexicans were serious about getting their act together.

As an episode in public policy making, our decision making in the face of a highly uncertain situation and considerable political pressure showed that the probabilistic thinking that I internalized so deeply in the financial world had real applicability in Washington. And that process was ongoing, as we reevaluated options and policies when the facts changed on the ground in Mexico and in the financial markets. I think, too, that our work demonstrated the value of robust and open intellectual interchange in making government decisions.

Yet in other ways the episode showed me just how challenging decision making is in the context of government. Good decisions are much more difficult to make when disagreement is not just about means but about objectives. The private sector often focuses intensely on customers and employees, but in the final analysis everything comes back to serving the overriding objective of profitability—except perhaps for the relatively small portion of corporate activity devoted to philanthropy and other public purposes. The public sector, by contrast, operates with many equally legitimate objectives. For many in Congress, narcotics and illegal immigration mattered far more than economic issues in dealing with Mexico, and these legislators were not persuaded by our argument that the former problems would get far worse if Mexico defaulted and suffered from severe and prolonged economic duress.

Mexico also demonstrated the difficulties our political processes have in dealing effectively with issues that involve technical complexities, shorter-term cost to achieve longer-term gain, incomplete information and uncertain outcomes, opportunities for political advantage, and inadequate public understanding. Unfortunately, many of the most important economic, geopolitical, and environmental challenges of today's complicated world fit this profile, raising the question of how effectively our political system will be able to deal with them.

Having said that, the Mexican crisis also showed the strength of our system. Congress, while not able to act itself and often complicating our efforts, also induced greater focus on some important issues, such as moral hazard, and helped assure that all points of view were considered, a value often lost in a more monolithic system. In addition, some individual legislators were tremendously helpful. As I discovered, finding effective legislative allies is key to navigating our system successfully. At one point, Senator D'Amato had proposed measures that would rule out future Treasury use of the ESF in this type of situation—which would have severely hampered us in dealing with the Asian crisis two years later. But Senators Dodd and Sarbanes, who had a deep understanding of the benefits and risks of the global financial system, filibustered D'Amato's language, which led to a more limited constraint. I also remember an act of graciousness of the kind that occurs too seldom in any walk of life. Frank Murkowski, a Republican senator from Alaska and a former banker, who had opposed our rescue package as unlikely to work, went out of his way when I was testifying later at a hearing on another matter to say that he had been wrong—a gesture unusual in Washington and most other places.

However, Murkowski's prediction could have turned out to be right. Our program could have been undertaken only by a President—and an administration—willing to take a major calculated risk, substantive and political. We could have failed because of a mistake in our analysis, but also because of unforeseeable circumstances, or simply the foreseeable risk actually occurring. If the odds are calculated accurately at three to one, you'll lose one time in four. Unfortunately, Washington—the political process and the media—judges decisions based solely on outcomes, not on the quality of the decision making, and makes little allowance for the inevitability of some level of human error. This can easily lead to undue risk aversion on the part of public officials. The same issue exists in the private sector—in my own experience, most seriously in judging trading and investing results. But the private sector somewhat more frequently recognizes the need to look beyond the outcome to reach the most sensible and constructive evaluation.

Some years later, Paul O'Neill, the Bush administration's newly appointed Treasury Secretary, said he liked the Mexican program because it worked. “We gave them money, it stabilized their situation, and they paid back the money ahead of schedule,” he said. “I like success. I'm not a real fan of even well-meaning failure.” Where O'Neill said he liked what worked, my view was that decisions shouldn't be evaluated only on the basis of results. Even the best decisions about intervention are probabilistic and run a real risk of failure, but the failure wouldn't necessarily make the decision wrong.

Finally, what concerns me most is how little the public understands the impact that all of the issues around globalization and economic conditions elsewhere have on jobs, living standards, and growth in this country and how critical U.S. leadership is on these international economic matters. The result, as I realized over and over again during my six and a half years in Washington, is that public support—and thus political support—for trade liberalization, international financial-crisis response, foreign aid, funding for the World Bank and the IMF, and the like—is at best very difficult to obtain.

At one point during the second term, Secretary of State Madeleine Albright and I discussed holding joint public meetings around the country to try to improve public understanding of how global issues, both economic and geopolitical, affect people's lives. Regrettably, we never did this, but some kind of ongoing public education campaign is badly needed to change the politics around all these concerns, which are so critical to our future. On trade, for example, dislocations are very specific and keenly felt—and lead to strong political action—but the benefits of both exports and imports are widely dispersed and not recognized as trade-related, and thus haven't developed the level of political support they require.

We also face significant challenges when it comes to the international politics of economic leadership. In Mexico, and later in the Asian financial crisis, U.S. leadership, exercised in correlation with the G-7, the IMF, the World Bank, and others, was necessary for effective response. But even our closest allies are ambivalent about the role of the United States. We are criticized if we don't lead and resented if we do. At Treasury, the lesson we took was to work all the more energetically with other countries to reach consensus whenever practical, which often meant making accommodations on our part. But we also recognized that at times we would feel a need to push beyond where others wanted to go.

In 1995, I referred to the Mexican crisis as a “very low-probability event.” But my view later changed. The likelihood of a contagious crisis emanating from problems in any one developing country may ordinarily be small. But modern capital markets—with their many interrelationships, size, and speed—combined with the inherent human tendency to go to excess, create a seemingly inevitable tendency toward periodic destabilization that is difficult to anticipate and prevent. Indeed, only a couple of years later, I found myself immersed in another global financial crisis—one far more threatening in its scale, complexity, and potential consequences than what had happened in Mexico.

CHAPTER TWO

A Market Education

IWAS AN ODD CHOICE for Goldman Sachs when the firm hired me, at the age of twenty-eight, to work in its storied arbitrage department. Nothing about my demeanor or my experience would have suggested I might be good at such work. The stereotypical personality type of the arbitrageur was, in those days, forceful and confrontational. I was then, as now, a low-key, not manifestly aggressive person. As for my qualifications, I don't think I'd even ever heard the phrase “risk arbitrage” before I started the job search that led to Goldman Sachs. But as it happened, arbitrage and I turned out to be a pretty good fit.

Arbitrage in its classic form is nothing more complicated than attempting to profit by buying something in one market and then selling it at the same time in another market for a price differential. As practiced in the years before the Second World War, when communication advantages were still possible, classic arbitrage meant trying to capture discrepancies in different financial markets. To take a simple example, the British pound might have been trading at $2.42 in London and $2.43 in New York. If an arbitrageur could buy pounds in London and sell them in New York simultaneously, he would be assured a profit of $1 for every $242 he put up. The only risk in this kind of arbitrage is not completing the transaction fast enough. During the first half of the century, many firms made a steady income from the minor price differentials for the same currencies and securities trading in different markets.

As global communications improved, however, the profit went out of traditional arbitrage. Once everyone knew in real time what the pound was trading at on various markets, the discrepancies became, for the most part, too small to be worth exploiting. But in the years after the Second World War, Gustave Levy, the man I would work for a couple of decades later at Goldman Sachs, helped develop a new business known as “risk arbitrage.” In classic arbitrage you buy and sell the same thing simultaneously. In the simplest form of risk arbitrage, you buy one stock—call it A—that will be converted into another stock—B—once an already announced event, such as a merger, is completed. At the same time as the purchase, you sell B in order to “hedge” the transaction and lock in your profit. There's an element of risk, because the conversion of A into B isn't certain—the deal might fall apart rather than close.

Since the 1950s, risk arbitrage on Wall Street has meant buying securities that are the subject of some material event, like a merger, a tender offer, a breakup, divestiture, or a bankruptcy. As a hypothetical example, Big Company might announce a friendly takeover of Acme Industries at the price of one half of a share of BigCo's stock for every Acme share. Say BigCo is trading at $32 per share. The stock of the target company was trading at $13 before the deal was announced and rose to $14.50 after the announcement, based on its being worth $16 per share once the deal is completed (one half of BigCo if it remained at $32). In a risk arbitrage transaction, you would buy shares of Acme and “sell short” the number of BigCo shares you would receive when the takeover closed. Short selling in this context means selling something now that you don't yet own to hedge against market risk—in other words, to protect yourself against the possibility that by the time the item you're buying (A) is converted into the item you're selling (B), B will have gone down in value. (To sell something you don't own, you have to borrow it for a fee.) Then when the deal closes, you simply take the shares of BigCo you received in exchange for your Acme shares and deliver them against the short, replacing what you borrowed and closing out the position. Your profit is the difference between the transaction price and the price you initially locked in. Movements of the BigCo stock subsequent to your short sale don't matter—if BigCo goes down 5 points, it doesn't affect you because you've already sold the BigCo stock short, locking in the spread against the Acme stock you've bought.

However, you receive the profit only
if
the transaction goes through. If the deal breaks up, you are left with a position that you bought at a deal premium (Acme) and a short position (BigCo)—with almost certain losses on one or both. In this type of transaction, the potential profit is much larger than in a classical arbitrage trade—$1.50 for every share costing $14.50 in my hypothetical example. But the risk is also much greater, since the takeover might fail to close for any number of reasons. In practice, such transactions become enormously more complicated and more interesting.

Gus, a great financial innovator with the gentle disposition of an active volcano, had developed this kind of transaction after World War II in response to anomalies produced by the wartime boom. During the Great Depression, a number of railroads had filed for bankruptcy, leaving the prices of their shares and bonds badly depressed. During the war years, however, the railroads had been operating at full capacity and, as a result, were flush with cash. Coming through bankruptcy court, they were due to be reorganized in ways that would unlock their real value. Arbitrageurs like Gus would buy the stock of such technically insolvent companies and wait for them to be restructured.

By the end of the 1950s, that kind of opportunity was also becoming rare. But in the mid-1960s, around the time Gus and his protégé L. Jay Tenenbaum hired me as the junior man in the arbitrage department at Goldman Sachs, the risk-arbitrage business was picking up again, thanks to a wave of takeovers and mergers. By the end of the decade, Goldman Sachs was making significant profits in the context of the times—several million dollars a year—using its own capital for these transactions. Because the work was risky, complicated, and highly profitable, it had also acquired a certain mystique. Firms like Goldman didn't want their competitors to know how they went about the arbitrage business. In 1966, the year I was hired, L. Jay was quoted in
Business Week:
“Asking about our arbitrage operations is like walking into a couple's home and asking about their sex life.” While arbitrage is still a big business on Wall Street, it has become much less secretive.

I'll try to explain what we did in those days by describing an arbitrage transaction I actually worked on in 1967. Although this deal was in many ways typical of the hundreds I was involved in during my first several years at Goldman Sachs, I remember it well for reasons that will become clear. It was a merger of two companies that were traded publicly: Becton Dickinson, a medium-sized manufacturer of medical equipment, and Univis, a somewhat smaller company that made eyeglass lenses. Under the terms of an announced friendly takeover, Becton Dickinson would buy all outstanding shares of Univis for about $35 million in stock. Shareholders in Univis would get a .6075 share of Becton Dickinson for each share of Univis they held.

At the time the deal was announced, on September 4, 1967, Becton Dickinson was trading at around $55 a share and Univis at around $24½. If the merger was to be completed, A, or Univis, would become B, or Becton Dickinson, and a Univis share would be worth $33½—at the price of Becton Dickinson when the deal was first announced (.6075 x $55). To decide whether to engage in arbitrage, we had to estimate the odds of the merger coming to fruition, what we would make if it did, and what we would lose if it didn't—my framework, you might say, for dealing with most decisions in life.

Such an announced merger could fail to be completed for any number of reasons. It might be called off after either side performed its “due diligence” of examining the other's books in detail. Or the shareholders of either company might reject the terms of the transaction as not favorable enough. The Justice Department or the Federal Trade Commission might decide that a combination of the two companies was anticompetitive. Regulatory issues might surface. One of the firms might have a history of announcing deals and not completing them, and simply change its mind or be too unwilling to make accommodations on specific matters that arose after the initial agreement in principle. We would weigh and balance the different factors to decide whether or not to take an arbitrage position.

The first order of business was rapid, intensive research. I had to examine all the publicly available information I could obtain. I had to talk to proxy lawyers and antitrust lawyers. Then I had to speak to officers at both companies, much as a securities analyst does. I almost never had all the information I would have liked. Seldom did I have enough time to think everything through.

But even with as much information and time as I might have hoped for, risk arbitrage would have fallen far short of science. Many of the notes I put down on my legal pad weren't quantitative or measurable points of data. They were judgments. And once I finished all my analysis and reached a point of relative clarity, a correct answer wouldn't simply present itself. The final decision was another judgment, involving my sense of a situation. We might pass up a transaction where the numbers looked promising simply because of a feeling that two companies didn't make a good match or because we didn't trust some of the people involved.

But recognizing the essential component of experienced feel in this kind of judgment is different from not having a framework and making decisions in a nonsystematic way or on the basis of instinct. Some arbitrageurs at other firms operated on a far more ad hoc and subjective basis, their decisions driven by bits of information, trading activity, and gossip. At Goldman, our decisions were driven much more by analysis. We always tried to think of everything that could possibly go wrong with a deal and then tried to evaluate how much weight to accord to such risks in our analysis. Despite the all-too-human tendency to lose sight of one's own disciplined framework, we tried our best to be cool and hardheaded. Emotion, like instinct not moored in analysis, could be misleading. If you became frightened easily—or were greedy—you couldn't function effectively as an arbitrageur.

In merger transactions such as Becton-Univis, our projected loss would typically be much larger if the deal fell apart than our projected gain if it went through. That meant that the odds had to be substantially in our favor for us to choose to participate. But how greatly did they have to be in our favor? Someone who had been to business school would have recognized the charts I made on my yellow pad as expected-value tables, used to calculate the anticipated outcome of a transaction. After a while, organizing my analysis according to these tables became second nature and I'd do them in my head. But I still constantly scribbled notes and numbers on a legal pad—a lifelong habit with me.

The basic inputs in an arbitrage expected-value table are the price you have to pay for a stock; what you will get for the stock if a deal goes through (the potential upside); what you will have to sell it for if the deal doesn't go through (the potential downside); and finally—the most difficult factor to assess and the heart of risk arbitrage—the odds that the transaction will be completed. With the help of some papers from Goldman's archives, I've re-created an expected-value table for Becton-Univis. After the merger was announced, Univis stock traded at $30½ (up from $24½ before the announcement). That meant the upside potential from an arbitrage trade was $3, because a Univis share would be worth $33½—.6075 of a share of Becton Dickinson—if the deal went through. If the deal didn't go through, Univis would be likely to fall back to around $24½, giving our investment a downside potential of around $6. Let's say we rated the odds of the merger being completed as slightly better than six to one (about 85 percent success to 15 percent failure). On an expected-value basis, the potential upside would be $3 multiplied by 85 percent. The downside risk would be $6 multiplied by 15 percent.

$3 x 85 percent = $2.55 upside potential
– $6 x 15 percent = – $0.90 downside risk
————————————————————
Expected value = $1.65

The $1.65 was what one could expect to earn by tying up $30.50 of the firm's capital for three months. That works out to a return of approximately 5½ percent, or 22 percent on an annualized basis. A lower rate of return than that would have been a red light. We figured that it wasn't worthwhile to obligate the firm's capital for a return of less than 20 percent per annum.

I'm simplifying in a variety of ways. You also had to factor in the risk that a merger would break up under conditions that would cause the target stock you'd bought—in this case Univis—to fall lower than its preannouncement floor or that would drive the acquiring company's stock—the Becton Dickinson shares you'd sold short—higher. Or, even worse, both could occur at the same time. And you wouldn't just make the decision to invest in this sort of deal and wait for the result several months later. The odds of a merger reaching closure changed constantly over time, as risks emerged and receded and share prices fluctuated. We had to stay on top of the situation, recalculating the odds and deciding whether to commit more, reduce our position, or even liquidate it entirely. And, of course, an arbitrageur would be involved in many such deals at any one time. You had to do a lot of them, because arbitrage is an actuarial business, like insurance. You expect to lose money in some cases but to make money over the long run thanks to the law of averages.

In the case of Becton-Univis, the positive expected value prompted us to take a position—we sold short 60.75 shares of Becton Dickinson for every 100 shares of Univis we bought. As I explained, selling short the acquiring company—which we'd do by borrowing shares for a fee—was a hedge against the market risk. If the stock prices of both companies went down while the merger was under way—perhaps because the sector or market weakened—our profit would still be locked in, as long as the deal went through.

Goldman's trading records—which the firm graciously made available to us for this example—show that on my recommendation, we initially bought 33,233 shares of Univis at an average price of $30.28 and a total cost of just over $1 million—a significant amount at that time. We also sold short 19,800 shares of Becton Dickinson, into which the Univis shares would be converted. After increasing our positions in the interim, we stood to make around $125,000 if the merger closed. By the end of the year, Becton had risen to around $60, causing Univis to climb to $33¾.