The openness–inflation puzzle revisited

Abstract

Dynamic panel estimates show the negative relation between trade openness and inflation found by Romer (Quarterly Journal of Economics, (VIII, 869–903, 1993) but questioned by Terra (Quarterly Journal of Economics, (XIII, 641–48, 1998) became more robust in the 1990s, both among high income OECD and developing countries. Trade openness was also associated with less variable inflation during the 1990s and had a stronger disinflation effect in economies with floating exchange rates.

Notes

¹ D. Romer himself is now among the sceptics, see his Advanced Macroeconomics (2000, 2nd edn, p. 492).

² Lane (1997), however, finds that after controlling for country size, inflation is negatively correlated with openness among OECD countries. Similar results are reported in Table 3 below.

³ These results are robust to the addition of structural variables such as per capita income, latitude, total PPP GDP (size) and regional dummies see Lane (1997) and Gruben and McLeod (2001). However, many of the variables used in these cross-country regressions are not available in time series or for the 1990s (central bank independence measures for example.)

⁴ Using the World Bank WDI 2002 imports of goods and services over its PPP GDP estimates yields similar results, but the World Bank WDI only includes PPP GDP estimates from 1975 on, so using the PWT 6.1 openness measures provides a larger sample.

⁵ Between the late 1980s and the late 1990s, the weighted average import share for the 118 countries in the sample rose from 19% to 24% of GDP while the weighted average inflation rate fell from 70% to 5%.

⁶ The coefficient of variation is the standard deviation of the log of one plus the inflation rate divided by mean inflation for 1986–1990, 1991–1995 and 1996–2000.