Exchange rate forecasting, order flow and macroeconomic information

https://doi.org/10.1016/j.jinteco.2009.03.005Get rights and content

Abstract

This paper adds to the research efforts that aim to bridge the divide between macro and micro approaches to exchange rate economics by examining the linkages between exchange rate movements, order flow and expectations of macroeconomic variables. The basic hypothesis tested is that if order flow reflects heterogeneous expectations about macroeconomic fundamentals, and currency markets learn about the state of the economy gradually, then order flow can have both explanatory and forecasting power for exchange rates. Using one year of high frequency data collected via a live feed from Reuters for three major exchange rates, we find that: i) order flow is intimately related to a broad set of current and expected macroeconomic fundamentals; ii) more importantly, order flow is a powerful predictor of daily movements in exchange rates in an out-of-sample exercise, on the basis of economic value criteria such as Sharpe ratios and performance fees implied by utility calculations.

Introduction

Following decades of failure to empirically explain and forecast fluctuations in exchange rates using traditional exchange rate determination models (Meese and Rogoff, 1983, Cheung et al., 2005, Engel et al., 2008), the recent microstructure literature has provided promising evidence, pioneered by a series of papers by Evans and Lyons, 2002a, Evans and Lyons, 2005a. These papers have theoretically motivated and empirically demonstrated the existence of a close link between daily exchange rate movements and order flow. Order flow is defined as the net of buyer- and seller-initiated currency transactions, and may be thought of as a measure of net buying pressure (Lyons, 2001).

In a macro–micro dichotomy of exchange rate determination, one may view the standard macro approach as based on the assumption that only common knowledge macroeconomic information matters, and the micro approach as based on the view that heterogeneous beliefs are essential to determine prices. However, given the lack of a widely accepted model for nominal exchange rates, neither of these extreme perspectives is likely to be correct. A hybrid view, as presented in the microstructure approach to exchange rates (e.g. Evans and Lyons, 2002a, Evans and Lyons, 2007, Bacchetta and van Wincoop, 2006), seems more plausible. In this framework, macroeconomic information impacts on exchange rates not only directly, as in a standard macro model, but also indirectly via order flow. Order flow becomes a transmission mechanism that facilitates aggregation of dispersed price-relevant information such as heterogeneous interpretations of news, changes in expectations, and shocks to hedging and liquidity demands.

Evans and Lyons (2002a) provide evidence that order flow is a significant determinant of two major bilateral exchange rates, obtaining coefficients of determination substantially larger than the ones usually found using standard macroeconomic models of nominal exchange rates. Their results are found to be fairly robust by subsequent literature (e.g. Payne, 2003, Marsh and O'Rourke, 2005, Killeen et al., 2006). Moreover, Evans and Lyons, 2005a, Evans and Lyons, 2006 argue that gradual learning in the foreign exchange (FX) market can generate not only explanatory, but also forecasting power in order flow.

The finding that order flow has more explanatory power than macro variables for exchange rate behavior gives some support to the importance of heterogeneous expectations (Bacchetta and van Wincoop, 2006). However, it does not necessarily imply that order flow is the underlying determinant of exchange rates. It may well be that macroeconomic fundamentals are an important driving force for exchange rates, but that conventional measures of expected future fundamentals are so imprecise that an order-flow “proxy” performs better in estimation. Unlike expectations measured by survey data, order flow represents a willingness to back one's beliefs with real money (Lyons, 2001).

Building on the recent success of the microstructure approach to exchange rates, a number of important hurdles remain on the route towards understanding exchange rate behavior. First, if one were willing to accept the existence of a link between order flow and exchange rate movements, economists are still awaiting for conclusive empirical evidence explaining where the information in order flow stems from. This issue is important in attempting to bridge the divide between micro and macro approaches to exchange rate economics.

Second, while the emphasis of the microstructure literature has primarily been on explaining exchange rate movements with order flow, there are only few empirical results on its forecasting power. The Meese–Rogoff finding that no available information is useful in forecasting exchange rates out-of-sample better than a naïve random walk model is robust and remains the conventional wisdom. This stylized fact implies that knowledge of the state of the economy available at a point in time is largely useless information for predicting currency fluctuations. However, if order flow does indeed reflects heterogeneous beliefs about the current and future state of the economy, and if currency markets do not discover order flow in real time but only through a gradual learning process (due to, for example, the partially decentralized nature of the FX market and its relatively low degree of transparency), then order flow should also provide forecasting power for exchange rate returns, as discussed in greater detail in the next section.

In this paper, we make progress on both these issues. We start from noting that theoretically order flow can aggregate macroeconomic information through two channels: (i) differential interpretation of news (currently available information); and (ii) heterogeneous expectations about future fundamentals. We provide evidence that the information impounded in order flow is intimately related to a broad set of macroeconomic variables of the kind suggested by exchange rate theories, as well as to expectations and changes in expectations about these fundamentals, implying that both channels suggested by theory are at work. Then, given the intermediary role of order flow for the relation between exchange rates and macroeconomic fundamentals, we investigate empirically the ability of simple microstructure models based on order flow to outperform a naïve random walk benchmark in out-of-sample forecasting.

The forecasting analysis relies on the use of economic criteria. Statistical evidence of exchange rate predictability in itself does not guarantee that an investor can earn profits from an asset allocation strategy that exploits this predictability. In practice, ranking models is useful to an investor only if it leads to tangible economic gains. Therefore, we assess the economic value of exchange rate predictability by evaluating the impact of predictable changes in the conditional FX returns on the performance of dynamic asset allocation strategies. Building on previous research by West et al. (1993), Fleming et al. (2001) and Della Corte et al. (forthcoming), we employ mean-variance analysis as a standard measure of portfolio performance and apply quadratic utility to examine whether there are any economic gains for an investor who uses exchange rate forecasts from an order flow model relative to an investor who uses forecasts from alternative specifications, including a naïve random walk model. Economic gains are evaluated mainly using two measures: the Sharpe ratio and the performance fee. The Sharpe ratio is the most common measure of performance evaluation employed in financial markets to assess the success or failure of active asset managers; it is calculated as the ratio of the average realized portfolio excess returns to their variability. The performance fee measures how much a risk-averse investor is willing to pay for switching from a portfolio strategy based on the random walk model to one which conditions on order flow. In addition, we calculate the break-even transaction cost, that is the transaction cost that would remove any economic gain from a dynamic asset allocation strategy relative to a simple random walk strategy.1

Using one year of data for three major exchange rates obtained from Reuters on special order, we find evidence that order flow is a powerful predictor of movements in daily exchange rates in an out-of-sample exercise, where an investor carries out allocation decisions based on order flow information. The Sharpe ratio of the order flow model is around unity and substantially higher than the Sharpe ratios delivered by alternative models, including the random walk. Furthermore, we find that a risk-averse investor would be prepared to pay high performance fees to switch from the random walk model to a model based on order flow. Consistent with leading microstructure theories, our interpretation is that order flow is a key vehicle via which fundamental information impacts on current and future prices.

The remainder of the paper is organized as follows. In the next section, we provide a brief literature review. Section 3 describes the data set and presents preliminary results on the link between order flow and exchange rates. The relation between order flow and macroeconomic fundamentals is examined in Section 4. The forecasting setup and the investor's asset allocation problem are described in Section 5, and the results on the economic value of forecasting models that condition on order flow are reported in Section 6. Section 7 concludes.

Section snippets

Related literature and motivation

The failure of conventional structural models to explain and forecast exchange rates has recently given rise to two different strands of research: one focusing on the implications of the standard present-value approach to asset pricing and the other based on the microstructure approach to the FX market. On the one hand, Engel and West (2005) demonstrate that the lack of forecastability of exchange rates using fundamentals can be reconciled with exchange rate determination theories within a

Data sources

The FX market is by far the largest financial market, with a daily turnover of US dollar (USD) 3210 billion, a third of which is in spot transactions. Electronic brokers have become the preferred means of settling trades, and 50–70% of turnover in the major currency pairs is settled through the two main electronic platforms, Reuters and Electronic Brokerage System (EBS) (Galati and Melvin, 2004).7

Order flow and macroeconomic fundamentals

In this section, we examine the link between macroeconomic information and order flow using the standard present-value exchange rate model:Δst+1=(1b)b(stEtmft)+ɛt+1,where ɛt+1(1b)q=0bq(Et+1mft+q+1Etmft+q+1). As discussed previously, in this model order flow may capture current fundamentals information (the first term in Eq. (4)) and changes in expectations about future fundamentals (the second term in Eq. (4)). We investigate empirically both links between order flow, expectations and

Empirical models and asset allocation: the framework

Recently, several banks have invested in technology that captures order flow information for forecasting purposes (e.g. the CitiFlow system by Citigroup and similar systems built at UBS, Royal Bank of Scotland and HSBC). The microstructure literature has used some of these data (e.g. Evans and Lyons, 2005a, Marsh and O'Rourke, 2005, Sager and Taylor, 2008) as well as data constructed from electronic platforms, Reuters and EBS (e.g. Evans, 2002, Payne, 2003, Berger et al., 2008). In this

The forecasting power of order flow: empirical results

We begin our economic evaluation of one-day-ahead exchange rate predictability by performing in-sample estimations of the four candidate models: MGEN, MPOF, MFB and MRW. The estimation is carried out over the period from February 13, 2004 to June 14, 2004, comprising about one third of the available observations. While the number of observations used in the in-sample estimation is relatively small, all models are particularly parsimonious linear models, with a small number of parameters. This

Conclusions

This paper makes two related contributions to empirical exchange rate economics. We show that order flow is related to current and expected future macroeconomic fundamentals, and can profitably forecast risk-adjusted currency returns.

Previous research has found that order flow has strong explanatory power for exchange rate movements, whereas macroeconomic fundamentals have weak explanatory power. We provide evidence that a significant amount of order flow variation can be explained using

Acknowledgments

This paper was partly written while Lucio Sarno and Elvira Sojli were visiting Norges Bank. We are grateful for constructive comments to Charles Engel (editor), an anonymous referee, Tam Bayoumi, Geir Bjønnes, Hilde Bjørnland, Alain Chaboud, Pasquale Della Corte, Frank Diebold, Martin Evans, Bilal Hafeez, Robert Kosowski, Richard Lyons, Ian Marsh, Lukas Menkhoff, Michael Moore, Carol Osler, Anna Pavlova, Ilias Tsiakas, Bent Vale, Giorgio Valente, Eric van Wincoop, Paolo Vitale, to other

References (60)

  • EvansM.D.D. et al.

    Do currency markets absorb news quickly?

    J. Int. Money Financ.

    (2005)
  • EvansM.D.D. et al.

    How is macro news transmitted to exchange rates?

    J. Financ. Econ.

    (2008)
  • FamaE.F.

    Forward and spot exchange rates

    J. Monet. Econ.

    (1984)
  • FaustJ. et al.

    Exchange rate forecasting: the errors we've really made

    J. Int. Econ.

    (2003)
  • GlostenL.R. et al.

    Bid, ask, and transaction prices in a specialist market with heterogeneously informed traders

    J. Financ. Econ.

    (1985)
  • KilleenW.P. et al.

    Fixed versus flexible: lessons from EMS order flow

    J. Int. Money Financ.

    (2006)
  • MeeseR.A. et al.

    Empirical exchange rate models of the seventies: do they fit out of sample?

    J. Int. Econ.

    (1983)
  • MolodtsovaT. et al.

    Out-of-sample exchange rate predictability with Taylor rule fundamentals

    J. Int. Econ.

    (2009)
  • PayneR.

    Informed trade in spot foreign exchange markets: an empirical investigation

    J. Int. Econ.

    (2003)
  • WestK.D. et al.

    A utility based comparison of some models for exchange rate volatility

    J. Int. Econ.

    (1993)
  • AckermannC. et al.

    The performance of hedge funds: risk, return, and incentives

    J. Finance

    (1999)
  • AdmatiA.R. et al.

    A theory of intraday patterns: volume and price variability

    Rev. Financ. Stud.

    (1988)
  • AndersenT.G. et al.

    Micro effects of macro announcements: real-time price discovery in foreign exchange

    Am. Econ. Rev.

    (2003)
  • BacchettaP. et al.

    A scapegoat model of exchange-rate fluctuations

    Am. Econ. Rev. Pap. Proc.

    (2004)
  • BacchettaP. et al.

    Can information heterogeneity explain the exchange rate determination puzzle?

    Am. Econ. Rev.

    (2006)
  • Bank for International Settlements
  • Berger, D.W., Chaboud, A.P., Hjalmarsson, E., 2009. What drives volatility persistence in the foreign exchange markets?...
  • BjønnesG.H. et al.

    Liquidity provision in the overnight foreign exchange market

    J. Int. Money Financ.

    (2005)
  • Breedon, F., Vitale, P., 2005. An empirical study of liquidity and information effects of order flow on exchange rates....
  • BrownS.J. et al.

    Offshore hedge funds: Survival and performance, 1989–95

    J. Bus.

    (1999)
  • Cited by (171)

    • A new LASSO-BiLSTM-based ensemble learning approach for exchange rate forecasting

      2024, Engineering Applications of Artificial Intelligence
    • Price discovery and triangular arbitrage in currency markets

      2023, Journal of International Money and Finance
    • Currency exchange rate predictability: The new power of Bitcoin prices

      2023, Journal of International Money and Finance
    View all citing articles on Scopus
    View full text