Despite doubts, there’s enough proof to conclude that foreign trade is good for poor

Areas where workers faced foreign competition tended to have greater reduction in poverty rates.

By Devashish Mitra. He is professor, economics, Maxwell School of Citizenship and Public Affairs, Syracuse University.

"In the backdrop to India’s recent decision to not join the Regional Comprehensive Economic Partnership (RCEP), the question of trade liberalisation’s effects on poverty has once again come to the fore. In their new book, Good Economics for Hard Times, Economics Nobel laureates Abhijit V Banerjee and Esther Duflo provide an extensive discussion on the impact of international trade on poverty.

Specifically, they write about work on India by their former PhD student, currently at the International Monetary Fund (IMF), Petia Topalova, ‘The national poverty rate dropped rapidly in the 1990s and 2000s, from about 35% in 1991 to 15% in 2012.

But, against this rosy backdrop, greater exposure to trade liberalisation clearly slowed poverty reduction — contrary to what the Stolper-Samuelson theory would tell us, the more exposed a particular district was to trade, the slower poverty reduction was in that district.’

Not a Poor Conclusion
Based on this statement, a reader could be forgiven for concluding that had India not undertaken trade liberalisation, it would have experienced faster poverty reduction than it did. There are, however, several reasons why such a conclusion would be incorrect.

First, Topalova’s results, based on analysing trade-poverty linkages using the district as the unit of analysis, are quite different from those of a paper I co-authored with Rana Hasan and Beyza P Ural (bit.do/fipeF). Working with states and regions (a region being a collection of districts within a state with similar agro-climatic features) as units of analysis, we found that areas where workers were, on average, faced with greater increases in exposure to foreign competition tended to have experienced greater reductions in rural, urban and overall poverty rates (and poverty gaps).

While working with districts can have some advantages from an econometric perspective, it also has some serious disadvantages. For example, the then-sampling methodology of the National Sample Survey Organisation’s (NSSO) household expenditure surveys used to estimate poverty did not allow for straightforward computation of representative poverty rates at the district level. Further, district boundaries change over time.

An individual district, relative to a state or region, is also more likely to be specialised in its production.

Districts specialised in importable goods will lack the product diversification that can generate offsetting real income enhancing effects of expanding export sectors. Also, irrespective of whether this makes qualitative differences to the results, Topalova emphasised as her main determinant of poverty the overall employment-weighted average tariff, including non-tradable industry tariffs, which she sets to zero.

In our study, we restricted attention to the weighted average for only the economy’s tradable part, since non-tradable employment itself responds to tradable tariffs. Also, a good’s non-tradability arises precisely from its prohibitively high (not zero) trade cost (inclusive of tariff). Our analysis is for tariffs and non-tariff barriers, using several alternative poverty measures.

Next, one should not conclude that trade liberalisation has harmed the cause of poverty reduction. As Banerjee himself has correctly pointed out, based on the analytical methods used in both studies, one can only say which states (or districts) reduced poverty faster or slower due to trade liberalisation, not what this liberalisation did to poverty at the national level.

Now, the national poverty rate certainly declined faster during India’s post reform period. To be sure, the 1991 economic reforms entailed more than just trade liberalisation. But being a key element of the reforms, it is likely that trade liberalisation contributed to higher economic growth and reduced poverty. That all boats were lifted — though some more than others — is consistent with both our study and Topalova’s.

Certainly, these results don’t warrant Banerjee and Duflo’s conclusion that “greater exposure to trade liberalisation clearly slowed poverty reduction”, which runs the risk of serious misinterpretation by lay readers. This claim is especially surprising since Topalova’s urban poverty results are mostly statistically insignificant — that is, lack precision.

Importantly, in some of our analysis, we allowed the transmission of tariffs to change with distance from ports and the quality and density of the transportation infrastructure that vary across states. Hasan and I later extended our analysis in a paper co-authored with Jewelwayne Cain, where we added a later round of NSSO data. All our earlier results survived, and we also found that financial development helps with this trade and poverty reduction channel.

Interestingly, a common finding across both Topalova’s study and ours concerns the role of labour market regulations. While in our study, labour market flexibility was found to enhance the relative speed of poverty reduction in open states, in Topalova’s, it was found to reduce the relative slowness of poverty reduction in open districts. Overall, given the evidence available today, it seems impossible to unambiguously conclude that trade is bad for the poor.  In fact, in India’s case, exactly the opposite applies."