The `news paradigm' that has dominated recent thinking about the foreign exchange market requires that `news' occur whenever the exchange rate changes, so market observers now search for news to explain otherwise `inexplicable' movements. In Discussion Paper No. 466 Research Fellow Paul de Grauwe and Hans Dewachter use a monetary model of the exchange rate that incorporates speculative behaviour. `Chartists' and `fundamentalists' interact; the weight of each class of speculators changes depending on market circumstances. `Chartists' simply extrapolate recent exchange rate changes using moving averages, while `fundamentalists' base their expectations on calculations of purchasing power parity.

De Grauwe and Dewachter find that applying these simple rules may produce very complex and essentially unpredictable behaviour or `chaos' even though the underlying model is deterministic. News of the money stock usually has strong effects, but sometimes not, and turbulence in the exchange market often occurs when there is no news. The interaction of speculators using different pieces of information can generate complex and not wholly explicable dynamics, so that `news' remains important, but need not be invoked to explain every observed movement of the exchange rate. De Grauwe and Dewachter's model therefore provides a synthesis view of the `news' model that has dominated academic thinking and the popular view that exchange rates are driven by speculative dynamics.

Their model also implies that the initial conditions are important, particularly for evaluating the effects of monetary disturbances. Moreover, the theory explains why out-of-sample forecasts of exchange rates using structural econometric models are typically poor. Very small measurement errors in the estimation of the underlying model can seriously change its exchange rate dynamics, making it quite unsuitable for predictive purposes. Their model takes no account of rational expectations, since agents have no incentive to invest time and effort in seeking knowledge of the underlying structural model. To be useful for predictive purposes, such knowledge would need a degree of precision that is not currently attainable in the social sciences. Agents therefore do not use sophisticated structural models for predictive purposes. Instead, they use simpler `rules of thumb' to compute the fundamental rate.

A Chaotic Monetary Model of the Exchange Rate
Paul De Grauwe and Hans Dewachter

Discussion Paper No. 466, October 1990 (IM)