Inverse Balassa–Samuelson effect in Mexico: the role of the oil sector

The field in question is the Cantarell field in Mexico’s southeast. (Interestingly, Cantarell is located close to the Chicxulub crater where an asteroid hit the Earth some 66 million years ago, an event widely believed to have brought about the extinction of the dinosaurs.).

Tica and Družić (2006) point out that there is consensus in the literature that the tradable sector comprises the industries of agriculture, mining, and manufacturing, while the non-tradable sector includes the rest, such as services and construction.

The sum of the productivity changes in the tradable subsectors may differ from the total because the weights of the sectors change over time and the estimation of the productivity of each sector is done independently of the total.

The general expression of the real exchange rate remains if we maintain the continuum of goods in the nontraded goods. Its price index can be expressed as \({P}_{N}={\left[{\int }_{i\in {\mathrm{\Omega }}_{N}}^{}{p}_{N}^{1-\sigma }(i)di\right]}^{\frac{1}{1-\sigma }}={\left[{\int }_{i=0}^{x}{\left(\frac{{A}_{T}\left(j=O\right)}{{A}_{N}}{p}_{T}\left(j=O\right)\right)}^{1-\sigma }di\right]}^{\frac{1}{1-\sigma }}+{\left[{\int }_{x}^{1}{\left(\frac{{A}_{T}\left(j=R\right)}{{A}_{N}}{p}_{T}\left(j=R\right)\right)}^{1-\sigma }di\right]}^{\frac{1}{1-\sigma }},\)

where x is a measure of the proportion of oil goods in the economy.

Therefore, the productivity differential between the traded and nontraded sectors could be seen as the weighted average of the productivity differential of the tradable subsectors.

A binary variable, indicating the period when NAFTA was renegotiated (from the second quarter of 2017 to the second quarter of 2018) and replaced with USMCA was included as an exogenous variable to control for the ensuing volatility in the exchange rate.

Several series not had the quarterly frequency and had to be interpolated. In Appendix B, it is shown that our results hold independently of the chosen interpolation method.

The results obtained in this analysis do not change if the steady-state investment is calculated as the average of the GFCF for the whole of 1993 or for the average in the period 1993–1994.

The specification of the deterministic terms in the auxiliary regression of the unit-root test is relevant. We therefore provide further explanations of our approach in Appendix A.

The complete set of results of the model appears in Table 1 (coefficients of the short-run equations) and Table 5 (cointegration evidence, long-run coefficients as well as the test results on the residuals of the short-run equations).

We identify the shocks using the Cholesky identification strategy. The rationale behind the order of the cointegrated variables is the following. Considering that our interest variable is the real exchange rate and how it is affected by the different productivity differentials (in Mexico’s oil and non-oil tradable sectors and in the USA), the real exchange rate is sorted last. This ensures that the rest of the variables can affect the RER contemporaneously. Then we place the US productivity differential as we assume the US shocks are exogenous to the model and as we do not expect Mexico's business cycle to alter that of the USA. Then, we place Mexico’s real variables: the oil and the non-oil productivity differentials, as Mexico, as a small open economy, could be affected by the US business cycle. Lastly, we place the financial variable: the real interest rate differential. This assumes that monetary policy could affect contemporarily the level of the exchange rate. Although in practice central banks do consider changes in the real exchange rate in their reaction function, the directional causality of the effect of monetary policy rate on the exchange rate may be arguable and is not the focus of the study. In addition, we tested whether switching the order between these two variables changes the effect of the shock of the oil productivity differential on the RER and found that it does not. The results are available under request. The remaining IRFs are shown in Appendix D.