Classifying Exchange Rate Regimes by Regression Methods

A new and easily implemented regression method is proposed for distinguishing floating from pegged regimes, whilst simultaneously identifying anchors of pegged currencies. The method can distinguish pegs with occasional devaluations from floats, and can be used to generate annual regime classifications. The method largely confirms the accuracy of the IMF’s de facto classification, but also shows that a significant minority of managed floats is close to being US dollar pegs. Even flexible managed floats have a strong tendency to track the US dollar.