RDP 9407: Explaining Import Price Inflation: A Recent History of Second Stage Pass-through Appendix 3: Tests of Exogeneity

This appendix addresses the issue of exogeneity in the estimation of first-stage pass-through. For the single-equation framework to be valid, it is necessary that the contemporaneous import-weighted exchange rate and the world price are exogenous to changes in the import price. In particular, the exchange rate and the world price have to be weakly exogenous, so that the error-correction term enters only the import price equation.[47]

To ensure that this is the case, tests of Granger causality were carried out in a vector autoregressive (VAR) model. This is actually a stronger test of exogeneity than required; Granger non-causality implies strong exogeneity.

A fourth order VAR on the differences was estimated and the results of Granger causality tests are presented in the table below. Sufficient lags were included to eliminate autocorrelation.

Table A4.1: Granger Causality Results
Null hypothesis
Test statistic
F (4,73) (significance)
Changes in the import price
do not Granger-cause changes
in the exchange rate
1.969 (0.108) Exchange rate is
strongly exogenous
Changes in the import price
do not Granger-cause changes
in the world price
0.767 (0.550) World price is
strongly exogenous


For a discussion of the different concepts of exogeneity, see Ericsson (1992). [47]