Macroeconomic models of nominal exchange rates perform poorly. In sample, R 2 statistics as high as 10 percent are rare. Out of sample, these models are typically out-forecast by a naïve random walk. This paper presents a model of a new kind. Instead of relying exclusively on macroeconomic determinants, the model includes a determinant from the field of microstructure-order flow. Order flow is the proximate determinant of price in all microstructure models. This is a radically different approach to exchange rate determination. It is also strikingly successful in accounting for realized rates. Our model of daily exchangerate changes produces R 2 statistics above 50 percent. Out of sample, our model produces significantly better short-horizon forecasts than a random walk. For the DM/$ spot market as a whole, we find that $1 billion of net dollar purchases increases the DM price of a dollar by about 1 pfennig.

No abstract

This paper tests whether macroeconomic news is transmitted to exchange rates via the transactions process and if so, what share occurs via transactions versus the traditional direct channel. We identify the link between order flow and macro news using a heteroskedasticity-based approach, a la Rigobon and Sack (2002). In both daily and intra-daily data, order flow varies considerably with macro news flow. At least half of the effect of macro news on exchange rates is transmitted via order flow. How Is Macro News Transmitted to Exchange Rates?In macroeconomic models of exchange rates, news maps directly into prices. The effect of news on currency demands in these models is common knowledge and transactions-though perhaps engendered by the change in exchange rates-play no role in causing the change. In microeconomic models of asset prices, in contrast, transactions do play a causal role in price determination (e.g., Glosten and Milgrom 1985, Kyle 1985).The causal role arises because transactions convey information that is not common knowledge. In this paper we test whether any part of the effect of news on exchange rates is transmitted via transactions and, if so, what share is transmitted that way versus the traditional direct channel.That transactions might play a role is motivated by recent empirical work demonstrating a link between signed transaction volume (order flow) and nominal exchange rate changes. 1 In the models employed by these papers, order flow affects exchange rates as a proximate determinant. The underlying determinant, which theory labels information, is not specified, nor is it directly tested. This leaves open the nagging question of what really drives the order flow. This paper is an attempt to address that question by examining whether macroeconomic news might be a determinant of order flow. This question is distinct from whether volume is determined by news, a well-established property of many speculative markets (see, e.g., Fleming and Remolona 1999). 2 1 See, e.g., Payne (1999), Rime (2000), Evans (2002), andLyons (2002a). Order flow-a concept from microstructure finance-is the cumulation over time of signed trades. Trades are signed according to whether the initiator is buying or selling. (The marketmaker posting the quote is the non-initiating side.) For example, a sale of 1 unit by a trader acting on a marketmaker's quote is order flow of -1. In rational-expectations models, order flow is undefined because all transactions in that setting are symmetric. A large empirical literature within finance shows that signing trades this way provides considerable explanatory power (see the review in Lyons 2001).2 A volume effect is consistent with idiosyncratic portfolio rebalancing in response to news. Under rational expectations, given the immediate unbiased adjustment in price that it implies, one would not expect good news for the dollar to produce positive (or negative) order flow in the aggregate; i.e., one would not expect a relative increase in executed transactions initiated ...