Old shocks cast long shadows over the exchange rate

We propose a new exchange rate model using IRD time series as the input, and we fit the new model with empirical data for calibration. We assume that exchange rate modeling cannot be based on the response to a single shock but must instead be based on the response to a series of shocks, as previous shocks could still be playing out and affecting the overall response. We extend the Dornbusch overshooting model and make adjustments to account for empirical findings. The new model is substantiated by empirical data from several currency areas and can explain the so-called exchange rate "puzzles". Based on the model, we derive a relationship that explains when no interest rate differential (IRD) will suffice to support a stable exchange rate, which also suggests when policy-makers could be tempted to widen the IRD continually.