Expectations Driven Distortions in the Foreign Exchange Market

The aim of this paper is to explain the phenomenon of distortions in the foreign exchange market. Distortions in the sense of (lasting) deviations of the exchange rate from its fundamental value are a sign of market inefficiency. One well known example for such a phenomenon is the bubble path of the US Dollar in the eighties. The starting point for our investigation is a chartists-fundamentalists model. Motivated by empirical observations a model is developed where boundedly rational market participants choose at the beginning of each new trading period between a technical and a fundamental trading rule to determine their speculative investment positions. If one subscribes to the strong assumption that the agents are able to figure out the true fundamental value of the exchange rate, then the exchange rate fluctuates in a complex way around its fundamental value and the foreign exchange market is more or less efficient. However, the contribution of this paper is to model the perception process of the fundamental exchange rate more realistically on the grounds of psychological evidence. Within this framework forces which influence the distortion are analysed: While the agents follow the news arrival process closely, mistakes in the information processing occur. These mistakes are propagated over time since the agents tend to stick to their previously perceived fundamental value (anchor heuristic). Moreover, if the agents belief that the spot exchange rate itself contains relevant information and thus incorporate it into their "anchor", the exchange rate becomes even more disconnected with its true fundamental. Nevertheless, in the long run the agents react to macroeconomic imbalances and thus adjust their perception. By this learning procedure, the distortion eventually is corrected and some long-term mean-reversion sets in. Note that the results are not the outcome of curious exchange rate fluctuations but that the generated time series shares some basic stylized facts with the empirical data. For instance, the model possesses the property of a non stable, fat-tailed distribution of the returns.