Are exchange rate models better today?

 

Is there hope for exchange rate forecasting? A recent paper suggests that exchange rate models, the traditional ones that have been used for decades, are actually getting better at doing their job. The relationships between exogenous variables and exchange rates are more well-behaved, or put differently, the economic relationships are stronger than what they have been in the past. See "Exchange Rate Models are Better than You Think, and Why They Didn't Work in the Old Days".

Specifically, the global risk and liquidity do a good job of predicting exchange rates in the 21st century. Both home and foreign economic variables do a good job and are significant in empirical models using monthly changes in log exchange rates. The change in fit is found through looking at 20 years rolling samples. 

So, what is the cause for this change in the quality of empirical models? The paper suggests that central bank creditability associated with inflation targeting and the Taylor principle is the cause. The strengthening central bank independence is also associate with the improvement. The exchange rate models did poorly because the monetary variables were insignificant and the wrong sign. Exchange rates have been more closely linked with Taylor Rule modeling and thus with inflation surprises which has translated to better empirical models. Stay calm and follow empirical exchange rate models and you will be rewarded with better forecasts.