Exchange rates and scapegoats

It is notoriously difficult to understand exchange rate fluctuations, especially in the shorter term. Predicting them is even worse, to the point that a random walk has consistently been shown to be the best predictor (with isolated exceptions). Consistently beating the random walk is seen as the holy grail in international finance.

Philippe Bacchetta and Eric Van Wincoop came up with an intriguing theory: there is no point in trying to relate exchange rate movements to observable fundamentals. Market participants react to rate changes by rationalizing them with some observed fundamental even when the true reason may be unobservable. And market participants keep changing the variables they look at. Everyone has made fun of press reports that explain that the dollar went up because of some event, and then the same event explains why the dollar went down the next day. This is what Bacchetta and Van Wincoop call scapegoating.

This is pretty much all theory. Marcel Fratzscher, Lucio Sarno and Gabriele Zinna have now found a way to test empirically this theory of scapegoats. The reason it took so long is that you need the right data: first a monthly survey of market participants on what they think is driving exchange rate movements, second the order flow data of a major market participant. The data supports remarkably well the scapegoat theory. When there is a large volume of orders, which are not public information and should thus be treated as unobservables that influence market outcomes, and one of the fundamentals moves more than usual in one way or the other, markets participants often link the latter to exchange movements. This is purely after the fact rationalizing, or as used in other contexts, superstition. How this is going to help us in forecasting exchange rate movements is not clear, though.