Necessary evil: water treaties and international trade

Abstract

Water is a limited resource but it lies at the heart of economies and life on Earth. As it ignores borders, its exploitation can affect water resource availability and quality even in geographically distant areas. As a solution, there is a growing interest in international agreements to protect water resources. However, these agreements can influence other aspects of the economy. This article focuses on the effect of water treaties (WTs) with environmental provisions on international trade. We argue that, as a key input in all production activities, protecting water resources can generate additional costs that can be detrimental to international trade. Empirical investigations are conducted using a structural gravity equation with exporter-year, importer-year and bilateral fixed effects. We find that WTs with environmental provisions decrease bilateral trade. Sectoral analyses show that these WTs reduce particularly mining trade and countries’ income levels and climatic conditions are important characteristics to consider when setting up these agreements. We also find that regional trade agreements (RTAs) with environmental provisions reduce mining and agricultural trade. So, while beneficial for sustainable development, policymakers should consider, when signing treaties with environmental provisions, that they constrain international trade in order to propose mitigation provisions.

Appendices

Appendix A: Robustness check—Effect of water treaties on trade with different lags

Table 7 below presents the results of the WTs’ effect on trade with different lags. We consider not only the contemporaneous effect of WTs, but also include two lags preceding the year of signing [column (1)] and thirteen lags preceding the year of signing [column (2)]. We can observe that, the coefficient on the 2-year lag is always significant. We also compute and present the cumulative effect of the WT variables (i.e. including all the WTs lags) at the bottom of the table. These cumulative effects are obtained by summing up the coefficients on the WT dummies. Calculations of point estimates, standard errors and significance levels, are based on the ‘delta method’.⁴⁰ The cumulative effect of WTs is negative and statistically significant at 5% in column (1). It is again negative in column (2) and statistically significant at only 10%.

Table 7 WA effect of trade—Results with more lags

Appendix B: The strict exogeneity test

As proposed by Wooldridge (2002), we check the possibility that trade and the WT variables can be co-determined by testing whether there is an anticipation behaviour in trade when a future treaty is going to be set up. The strict exogeneity test consists of verifying the absence of a potential anticipation effect. We thus estimate our specification by using the future level of WTs. Columns (1), (2) and (3) in Table 8 present the results with two, four and six years forward WT variables, respectively (i.e. $WT_{od,t + 2}$, $WT_{od,t + 4}$ and $WT_{od,t + 6}$).

Table 8 Strict exogeneity test of water treaties

The results in Table 8 below show no significant effect of $WT_{od,t + 2}$, $WT_{od,t + 4}$ and $WT_{od,t + 6}$ on trade. This confirms that there is no anticipation effect of trade with regards to future WTs.

Appendix C: WTs with environmental provisions and trade among dyads sharing river basins

We have replicated here the results regarding the effect of WTs with environmental provisions (presented in Table 2), on a restricted sample of dyads that shared at least a river basin. The new results are presented in Table 9 below. The coefficient on WTs with environmental provisions is negative and statistically significant [see Table 9, column (1)]. Results in Table 2, columns (2) concern the effect WTs without environmental provisions. The related coefficient is not statistically significant.

Table 9 The effect of WTs with environmental provisions on trade among countries sharing river basins