Openness and Inflation Based on David Romer s Paper Openness and Inflation: Theory and Evidence ECON 5341 Vinko Kaurin
Introduction Link between openness and inflation explored Basic OLS model: y = β 0 + β 1 x 1 + u Dependent variable (y): Average Inflation (lninfl) Independent variable (x 1 ): Openness, represented by average share of imports in GDP (import) Inverse relationship between inflation and openness expected
Theory Domestic expansion increases output at home relative to output abroad thus reduces the relative price of domestic goods. More open an economy is (more it imports), higher the cost of this depreciation of domestic products relative to products abroad. Real depreciation means, by definition, that the prices of foreign goods in domestic currency units rise faster than prices of domestic goods.
Theory (ctd.) So, for a given impact of output on the prices of domestic goods, the impact on inflation (measured by CPI) increases as the fraction of imports increases. This is because the rise in consumer price index (CPI) causes wages to increase, which increases costs of production for domestic firms. As a result, increased openness causes a monetary expansion to lead to a larger increase in domestic prices for a given increase in output.
Theory (ctd.) Therefore, under discretionary policy, since greater openness reduces the benefits of increase in output above its natural rate, policymakers have lower incentive to expand in more open economies. Not undertaking expansion because of higher openness, results in lower equilibrium inflation, hence the inverse relationship between openness and inflation.
Findings Displayed in 4 tables Table 1: Basic model, Simple Linear Regression Dummies added (data and regional) Real income per capita added Table 2: Model expanded by including (separately) political instability bank dependency Table 3: Subgroups Table 4: Instrumental Variable
Table 1 Basic OLS model: y = β 0 + β 1 x 1 + u Dependent variable: log average inflation (lninfl) Independent variable: openness, represented by average share of imports in GDP (import) Additional RHS variables included: Log real income per capita (lnincome) Data dummies (dinfl_cpi, dimport_imf) Regional dummies (dsouth_america, dcentral_am_carib, dafrica, doecd_member, doil_prod)
Table 1 (ctd.) Constant and openness significant Column 1: t-stat = -14.8 and -3.79 Basic model robust to inclusion of additional variables With regional dummies, openness is 80% of the original value Statistically significant (t-stat = -3.1)
Table 2 Index of political instability measured in revolutions and coups per year Two observations missing, reducing sample to 112 observations Index of central bank dependence represented by Cukierman, Webb, and Neyapti s [1992] overall index of central bank dependence for the 1980s Multiple observations missing, reducing sample to 62 observations
Table 2 (ctd.) Higher political instability, higher inflation on average Statistically significant More dependent central bank, higher inflation on average Statistically significant
Table 3 Robustness across various subsamples Many coefficients on openness are statistically insignificant 18 highly developed economies 18 highly developed countries Regression Excluded from regression Openness insignificant Openness significant
Table 3 (ctd.)
Table 4 Instrumental variable Endogeneity of openness? Find an instrument that is Correlated with openness (import) Uncorrelated with inflation Land area as an instrument (lnland) Current policies are not correlated with land area Explains 44.8% of variation in avg. share of imports in GDP Highly statistically significant Residuals insignificant Dependent Variable: LNINFL Method: Least Squares Date: 11/06/15 Time: 21:16 Sample: 1 114 Included observations: 114 Variable Coefficient Std. Error t-statistic Prob. C -1.785986 0.701904-2.544487 0.0124 RESID44_TEST_ENDO... 0.289746 0.546849 0.529846 0.5974 IMPORT -1.000449 0.424259-2.358108 0.0203 LNINCOME -0.016377 0.104969-0.156015 0.8763 DINFL_CPI 0.078120 0.176042 0.443757 0.6581 DIMPORT_IMF 0.312285 0.246508 1.266837 0.2081 DSOUTH_AMERICA 1.099486 0.243910 4.507751 0.0000 DCENTRAL_AM_CARIB 0.246627 0.220657 1.117694 0.2663 DAFRICA 0.105725 0.180911 0.584405 0.5602 DOECD_MEMBER -0.146754 0.260559-0.563229 0.5745 DOIL_PROD 0.014740 0.311229 0.047362 0.9623 R-squared 0.370597 Mean dependent var -2.103226 Adjusted R-squared 0.309490 S.D. dependent var 0.709202 S.E. of regression 0.589325 Akaike info criterion 1.871835 Sum squared resid 35.77231 Schwarz criterion 2.135854 Log likelihood -95.69461 Hannan-Quinn criter. 1.978986 F-statistic 6.064709 Durbin-Watson stat 1.917844 Prob(F-statistic) 0.000000
Table 4 (ctd.) No evidence of endogeneity of imports being the source of negative correlation between inflation and openness Coefficients on openness larger than in OLS Residuals insignificant
Extension of the Original Paper Data from 1989 to 2014 included Regression with data from 1989 to 2014: Results very similar to the original Same inference Regression using data for all years (from 1973 to 2014) Almost exactly the same results as in the original sample Same inference; inflation and openness have an inverse relationship
Extension of the Original Paper (ctd.)
Conclusion Average rates of inflation lower in smaller, more open economies Exception: highly developed countries Low inflation, not correlated with openness Relationship between inflation and openness stronger for countries that are: more politically unstable, or have less independent central bank. Inference remains the same when data is extended until 2014
Thank You