Shock-Dependent Exchange Rate Pass-Through : Evidence Based on a Narrative Sign Approach

This paper studies shock-dependent exchange rate pass-through for Japan with a Bayesian structural vector autoregression model. We identify the shocks by complementing the traditional sign and zero restrictions broadly following Forbes(2018) with the narrative sign restrictions related to the Plaza Accord. We find that the narrative sign restrictions are highly informative, and substantially sharpen and even change the inferences of the structural vector autoregression model originally identified with the traditional sign and zero restrictions. Besides, we confirm that there is a significant variation in the exchange rate pass-through across different shocks. Nevertheless, the exogenous exchange rate shock remains the most important driver of exchange rate fluctuations. Finally, we apply our model to "forecast'' the extent of pass-through conditional on certain foreign exchange interventions in 2018. We show with a novel structural scenario analysis that our model can improve the capability of the Japanese government to set policies more appropriately