Macroeconomics Determinants of Exchange Rate Pass-Through: New Evidence from the Asia-Pacific Region

ABSTRACT

Some recent studies observe an increasing degree of exchange rate pass-through (ERPT) to domestic prices, which has raised questions about the nature of the incompleteness and decline in pass-through. This article reexamines the degree of ERPT to the import, producer, and consumer price indices in Australia, New Zealand, Japan, and Korea in the Asia-Pacific region using up-to-date data, with several important findings. First, we reveal that ERPT to domestic prices follows the distribution chain, in that exchange rate movements alter import prices in the first stage and then producer and consumer prices in the second stage. Second, we offer valid evidence of an increase in ERPT to import prices after the global financial crisis in Japan, Korea, and New Zealand and of a relatively stable ERPT in Australia. Third, the changes in ERPT elasticities are most affected by macroeconomic determinants such as inflation volatility, interest rates, and trade openness, but this varies considerably across the surveyed countries and the three price indices. All our findings make a significant contribution to the empirical literature on ERPT and have policy implications.

KEY WORDS:

• Asia-Pacific region
• exchange rate pass-through (ERPT)
• inflation

JEL CLASSIFICATION:

• E31
• F31
• F41

Notes

^1. In our robustness check, the output gap and the US PPI are used as proxies for domestic demand and foreign marginal cost. Additionally, exchange rate movements may indirectly affect import prices, and thus other domestic prices, via the impact on commodity prices, so we add the world oil price in equation (2)(2) $\mathrm{\Delta }ln\left(P_t^i\right)={\alpha }_0+\sum _1^m{\alpha }_m\mathrm{\Delta }ln\left(P_t^i\right)+\sum _0^n{\beta }_n.\mathrm{\Delta }ln\left(NEE{R}_t^i\right)+\sum _0^p{\delta }_p\mathrm{\Delta }ln\left(M{C}_t^i\right)+\sum _0^q{\gamma }_q\mathrm{\Delta }ln\left(GD{P}_t\right)+Dcrisis+Dquarterly+{\epsilon }_t$$\mathrm{\Delta }ln\left(P_t^i\right)={\alpha }_0+\sum _1^m{\alpha }_m\mathrm{\Delta }ln\left(P_t^i\right)+\sum _0^n{\beta }_n.\mathrm{\Delta }ln\left(NEE{R}_t^i\right)+\sum _0^p{\delta }_p\mathrm{\Delta }ln\left(M{C}_t^i\right)+\sum _0^q{\gamma }_q\mathrm{\Delta }ln\left(GD{P}_t\right)+Dcrisis+Dquarterly+{\epsilon }_t$(2) for a further check. The results are in Table A1.

^2. We re-estimate equation (2)(2) $\mathrm{\Delta }ln\left(P_t^i\right)={\alpha }_0+\sum _1^m{\alpha }_m\mathrm{\Delta }ln\left(P_t^i\right)+\sum _0^n{\beta }_n.\mathrm{\Delta }ln\left(NEE{R}_t^i\right)+\sum _0^p{\delta }_p\mathrm{\Delta }ln\left(M{C}_t^i\right)+\sum _0^q{\gamma }_q\mathrm{\Delta }ln\left(GD{P}_t\right)+Dcrisis+Dquarterly+{\epsilon }_t$$\mathrm{\Delta }ln\left(P_t^i\right)={\alpha }_0+\sum _1^m{\alpha }_m\mathrm{\Delta }ln\left(P_t^i\right)+\sum _0^n{\beta }_n.\mathrm{\Delta }ln\left(NEE{R}_t^i\right)+\sum _0^p{\delta }_p\mathrm{\Delta }ln\left(M{C}_t^i\right)+\sum _0^q{\gamma }_q\mathrm{\Delta }ln\left(GD{P}_t\right)+Dcrisis+Dquarterly+{\epsilon }_t$(2) with one lag as a robustness check. The results are presented in Table A1.