Capital Markets

Marking the Dealer’s Cards

What economists can learn from currency traders.
Economist StaffNovember 30, 2005

You can make a career as an exchange-rate economist without ever crossing paths with a currency trader. The dealers who set the dollar’s price and the professors who theorize about it are separate breeds, speaking different tongues and operating at quite different metabolic rates. This is how Richard Lyons, an economist at the Haas School of Business (part of the University of California, Berkeley), describes a day spent with a friend of trading stock.

Most of the time my friend was reading tea leaves that were, at least to me, not so clear. The pace was furious — a quote every five or ten seconds, a trade every minute or two…He looked over in the midst of his fury and asked me, “What should I do?” I laughed. Nervously.

If asked that question at the start of the year, most economists, watching the steady deterioration of America’s current-account balance, would have told a dollar trader to sell. That would have been costly. The greenback reached two-year highs against the euro and the pound this month — before slipping back a bit after the president of the European Central Bank indicated that higher euro-zone interest rates were on the way.

The currency market routinely confounds economists. A classic 1983 study by Richard Meese and Kenneth Rogoff, then both at the Federal Reserve, concluded that macroeconomic models could not explain a currency’s direction of travel, let alone how far it would go. One would do just as well to assume that next month’s exchange rate will be the same as this month’s. After another 20 years of interrogation, the macroeconomic data has confessed little more of value. A new review of the evidence finds that some models, in some periods, beat tossing a coin. But not by very much.

Testing the Tea Leaves

If traders can learn nothing from economists, can economists learn anything from traders? No, is the customary answer. Economists traditionally assume that everything worth knowing about a currency is known by everyone; and that anything new is quickly embodied in the price. They imagine the market governed by an auctioneer, who finds the price that will square everybody’s bids and offers before any actual trading takes place. Economists study the auctioneer, not the traders.

After his visit to the trading desks, however, Lyons decided that dealers merited closer examination. He and Martin Evans, of Georgetown University, are part of a wider effort to model the currency market from “the trenches.” We cannot understand how currencies behave, they argue, unless we study the thinking and behavior of those who trade them.

The mythical auctioneer is notably absent from the models they construct. In his place stand the big currency dealers, who act as marketmakers, willing to buy or sell any amount of currency at their quoted prices. The marketmaker will take orders from a mix of clients — from American multinationals repatriating profits, to Asian central banks manipulating their currency — whose beliefs are as diverse as their motives.

These orders represent opinions, backed by money. As such, they convey useful information about what clients believe and how strongly they believe it. The marketmaker is privy to all of these opinions as they accumulate on its order books, while its clients know only their own. It is like a poker game in which only the dealer is allowed to see what each player throws into the pot.

What can be learned from peering over the dealer’s shoulder? Messrs Lyons and Evans have access to six and a half years of data, ending in June 1999, from Citibank, one of the largest dollar dealers. The figures show the balance of orders flowing into the bank: on any given day, customers may buy more dollars from Citibank than they have sold to it, for example. This positive “order flow,” as it is called, is valuable evidence of buying pressure that is visible only to the dealer.

Any such pressure will eventually be reflected in a stronger currency. What is surprising is how long it takes. Given that a trade is made every few minutes, a quotation every few seconds, one would think that any information in today’s order flow would be swiftly incorporated into the dealer’s quotes and the dollar’s price. Whatever forecasting power the flow of orders possesses, it should be quickly exhausted.

It turns out, however, that the information in today’s order flow will still be percolating through the market weeks afterwards. It takes time for a marketmaker to make sense of his orders, to distinguish signal from noise, plot from happenstance. It takes more time still for that insight to spread to the market as a whole, as dealers reveal what they know through their quotes and trades with each other. This hiatus creates an opportunity for forecasters. Messrs Lyons and Evans reckon that Citibank’s order flow can predict almost 16 percent of the dollar’s bobbing and weaving four weeks hence. That may not sound like a lot. But the macroeconomists cannot even explain what is going on today, let alone 16 percent of what will happen a month from now.

The two economists invoke Friedrich Hayek, who long ago pointed out that the economy runs on “dispersed bits of incomplete and frequently contradictory knowledge.” These scattered insights, Hayek stressed, are communicated to everyone through shifts in market prices. But with the dollar, a big dealer’s order books provide a sneak preview. If you can look into those books, you can infer something useful about the dollar next month. If not, then, like the rest of us, you can just flip a coin.