How Well Do Monetary Fundamentals Forecast Exchange Rates?

For many years after the seminal work of the Meese and Rogoff (1983a), conventional wisdom held that exchange rates could not be forecast from monetary fundamentals. Monetary models of exchange rate determination were generally unable to beat even a naïve no-change model in out-of-sample forecasting. More recently, the use of sophisticated econometric techniques, panel data, and long spans of data has convinced some researchers (Mark and Sul, 2001) that monetary models can forecast a small, but statistically significant part of the variation in exchange rates. Others remain sceptical, however (Rapach and Wohar, 2001b; Faust, Rogers, and Wright, 2001). It remains a puzzle why even the most supportive studies find such a small predictable component to exchange rates. This article reviews the literature on forecasting exchange rates with monetary fundamentals and speculates as to why it remains so difficult. Christopher J. Neely is a Research Officer in the Research Department of the Federal Reserve Bank of St. Louis. Lucio Sarno is a Reader in Financial Economics at Warwick Business School, the University of Warwick, and a Research Affiliate of the Centre for Economic Policy Research (CEPR), London. The authors thank Nelson Mark and Jeremy Berkowitz for making programs and data available and Nelson Mark, David Rapach, and Mark Wohar for helpful comments on early drafts. Charles Hokayem provided research assistance.