Migration's Contribution to Trade: State-Level Evidence on the Importance of Destination Geographic Proximity

ABSTRACT

I empirically analyze the importance of differing levels of geographic proximity, providing the first results of the migrant-trade nexus at the state level for both places of destination and origin. Relying on a unique dataset allowing the mapping of Mexican-born migrants' US states of residence to Mexican states of origin, I ensure a precise measurement of migrant networks and other potential determinants of international trade, including the distance and mass variables fundamental to any gravity model. Furthermore, I unmask distinct levels of geographic proximity that a single migration estimate disguises, estimating statistically significant elasticities of exports to in-state, and neighboring-state migration. These figures are not only qualitatitvely but also quantitatively important, corresponding to partial contributions of $1984 (in-state) and $538 (neighboring-state) to annual exports between respective US and Mexican states associated with each average additional migrant.

KEYWORDS:

• International migration
• international trade
• pro-trade effect of migration
• geographic proximity

Acknowledgements

I thank Cem Karayalcin and participants in the Florida International University Economics Departmental Seminar for useful discussion.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Michael Gove is an Assistant Professor of Economics in the Economics & Finance Department at the University of North Georgia.

Notes

1 For example, see Girma and Yu (2002), Peri and Requena (2010), White (2007), and Ehrlich and Canavire Bacarreza (2006), respectively.

2 In order to maintain consistency, I always refer to the geographic unit in terms of destination–origin throughout the paper. For example, I classify a study examining the connection between migrants living in the United States and their connection with exports from the US (entire country) to countries of migrants' origins as ‘country–country’.

3 For example, Yilmazkuday (2014) finds that the estimated coefficient for distance in the gravity-type models of determinants of international trade suffers from greater bias if not considering production location within countries (at the state or local level).

4 The matrícula consular is an identification card made available by the Mexican government to citizens residing abroad starting in 1871. The card must be renewed every five years, giving the holder access to opening a bank account, obtaining a driver's license and other services, depending on the specific country and state of residence. I define state of origin as the last state of permanent residence before migration occurred from Mexico to the US.

5 Card and Lewis (2007) examine the choice of US states of destination for Mexican migrants, analyzing changing trends during the 1990s. Hatzigeorgiou and Lodefalk (2011) also highlight the importance of new migrants in updating information.

6 See Felbermayr and Jung (2009) and Aleksynska and Peri (2011) for these respective emphases.

7 There is a limited amount of heterogeneity for these potential variables, for example with the presence of a number of languages in Mexico, however in the sample at hand this differentiation is so minimal that it does not justify inclusion in the regression equation as an additional control variable.

8 I discuss in Section 5 how the use of cross-sectional data limits direct treatment of the questions of potential endogeneity and direction of causality in the migration-trade context.

9 While imports data would clearly provide for useful comparison, unfortunately state–state data are not available presently.

10 Section 5 details modifications to the trade value calculation used to check for sensitivity of the results to these changes.

11 As of final revisions, data from 2010 are the most recent available.

12 This differs from the standard measure used by similar studies, that of great circle distance, due to the fact that the trade data (and a majority of Mexico–US migration) is by land. However, if great circle distance is indeed employed as the measure of distance, results change only minimally, with a slight increase in the magnitude of the distance coefficient.

13 A natural assumption may be that education level is associated with legal migration status, thereby making it more likely for individuals to obtain a matrícula consular if the state of origin corresponds to a low-education Mexican state on average. However, this assumption does not appear to be correct; see below for a related brief discussion.

14 This fraction with an upper limit of 1 is then multiplied by 100, resulting in the numbers expressed on the y-axis of Figure 1.

15 Felbermayr and Jung (2009) signal the extra importance of highly educated migrants for the pro-trade effect, while Aleksynska and Peri (2011) focus on migrants' participation in business networks, not exclusively on education level of migrants.

16 See Figure 2 for a scatter plot of this correlated data.

17 Given that adjacency is a binary variable and the dependent variable in Equation (1(1) ) is log-transformed, the coefficient of 1.2308 results in a 242% change in exports when changing from a non-adjacent to an adjacent state. The calculation, in percentage terms is: .

18 This figure can be alternatively calculated by multiplying the elasticity of exports to migration by the ratio of average state–state exports to average migrant stock.

19 Trade diversion could be a simple explanation for this negative effect, however I intentionally do not explore this further given the small magnitude and lack of significance of the estimate.

20 The calculation is as follows: .

21 Felbermayr and Toubal (2012) emphasize that while cross-country (state) wage differentials may be affected by aggregate trade, it is not completely clear how bilateral trade actually affects these wage differentials.

22 Aleksynska and Peri (2011) point out that the migration measure is a stock accumulated over years, making it likely to be determined before trade flows. Forwarding exports provides extra assurance that I am indeed completely addressing the potential of joint determination.

23 Gove (2016b), as indicated by the slight difference in title, is a partner study to this paper, focusing on the importance of geographic proximity of migrant sources in its contribution to international trade.

Additional information

Funding

This work was supported by Florida International University's dissertation fellowship.