Intro and literature Empirical approach Results Conclusion Overcoming non-tariff barriers to trade: the role of migrant networks Jos´ e L. Groizard and Joan Mart´ ın-Montaner Universitat de les Illes Balears – Universitat Jaume I & IEI Castell´ on, September 26, 2019 Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Results Conclusion Motivation The relationship between migration and trade has been studied for so long Migrants’ contribution to exchange involve different causal mechanisms that remain unclear Pro-trade effect is likely motivated by the way ethnic networks solve information problems Causal inference requires convincing identification strategies to sort out endogeneity trade relationships and migration decisions are potentially explained by common ‘unobserved’ shocks, and reverse causality Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Results Conclusion Idea We study the causal effect of a large migration shock into Spain exploiting the distribution of foreign born population per country of birth across provinces on export extensive margin by local firms Fixed cost-per product and fixed cost per transaction are common trade barriers We test if ethnic networks reduce the burden of regulations on the extensive margin of trade Identification Interaction between origin-specific push factors and destination-specific pull factors generates a quasi-random variation in the allocation of migrants across Spanish provinces (Burchardi et al, 2018) We use migration inflows over two decades 1991-2011 to 50 provinces and from 179 foreign countries and transaction level export data from province to each foreign country in 2015 Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Results Conclusion Literature Affinity between regions (social ties, trust, common ancestry, etc) correlate strongly with aggregate outcomes such as trade flows (Gould, 1994), FDI (Guiso, Sapienza and Zingales, 2009) and international assets flows (Portes and Rey, 2005) Several papers have attempted to assess the causal impact of migrations on trade and FDI using instrumental variables: Javorcik et al. (2011) use cost of acquiring a passport for the migration-FDI relationship; Cohen et al. (2015) use the location of Japanese internment camps during the WWII to infer trade flows between locations within the US and Japan; Parsons and Vezina (2016) use the placement of Vietnamese refugees after the Vietnam War to infer trade flows with Vietnam Nevertheless, in the last two papers instrumenting for migration from one country makes impossible to control for destination fixed effects Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Results Conclusion Literature (cont’) Peri and Requena-Silvente (2010) estimate a migration-trade causal effect for Spanish provinces relying on a Card (2001) instrument that might not be suitable in this context. Indicate strong extensive-margin effect of ethnic network Burchardi, Chaney and Hassan (2018) present a novel approach yielding a specific instrument for migrations from each origin to each destination at each point in time and allowing to control simultaneously for both origin and destination fixed effects Hornok and Koren (2015) and in Kropf and Saur´ e (2014) present theory and evidence of how administrative barriers to trade affect disproportionally the extensive margins of trade Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Results Conclusion Theory Trade frictions are large especially for poor countries (Anderson and van Wincoop, 2004). Migrants may have a positive effect on trade costs because its contribution to equalize preferences in consumption across countries they provide social collateral for contract enforcement, replacing weak institutions they reduce information frictions related to a country’s language, regulations, market opportunities and informal institutions Similar mechanisms are established for FDI or international business Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Indentification Results Data Conclusion Model Our goal is to estimate an equation like the following: T o , d = α o + α d + δ M 2011 + X ′ o , d + ǫ o , d (1) o , d where T o , d is a trade variable (i.e., exports) measuring either the number of transactions (shipments) or the number of products exported to the origin country, o from the destination region, d of migrants in the year 2015. M o , d is a measure of the extent of the ethnic network residing in region d with origin in country o in 2011, measured as the log of 1 plus units of foreign born resident. X ′ o , d is a vector of control variables normally included in the gravity-type of models, including the geographic distance between o and d , and difference in latitude between o and d Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Indentification Results Data Conclusion Extended model We include a new interaction variable to the model (1): T o , d = α o + α d + δ M 2011 + γ M 2011 × barrier o + X ′ o , d + ǫ o , d (2) o , d o , d where barrier o is an indicator variable taking value 1 when country o trade barrier is above world average γ coefficient tests whether migration effect of trade is higher at the most regulated economies Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Indentification Results Data Conclusion Identification The instrument is based on a dynamic equation of the form: M t − 1 � � I t o , d M t o , d = a t + a o , t + a d , t + b t M t − 1 o , d + I t d + η t + d t c t (3) o o , d M t − 1 I t o where the constant terms, a t , a o , t , and a d , t control for residual forces, such as demographics, which may vary over time, over space, and between different ethnic groups. The term b t M t − 1 o , d reflects the strength of the migration stock to attract new migrants with the same origin and destination being b t the parameter that modulates this intensity. I o measures the push factor common to all migrants with the same origin entering into Spain between t and t − 1. I t c t I t measures the the degree to which province o is particularly appealing to d migrants. The fraction of migrants from country o who already lives in M t − 1 province d , d t o , d , measures the strength migrant networks settle near other M t − 1 o countrymen. Coefficients, c t and d t control for the relative importance of the pull and recursive factors. Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Indentification Results Data Conclusion Measurement issues Pull factor, I t I t where I t = ∑ 50 d = 1 I t d : measured as the fraction of migrants 1 d coming from anywhere in the world who settle in d at between 2001 and 1991, excluding migrants from the same continent as o . Push factor, I t o : measured as the total number of residents living in Spain 2 that were born in o excluding migrants from o who settle in the same province as d between 2001 and 1991. We compute the interaction of these pull and push factors going back to 3 I t 1991 as our instrument: I t I t . d o Groizard & Mart´ ın-Montaner Migration and trade
Intro and literature Empirical approach Indentification Results Data Conclusion Data on migration Migration data are constructed from the population censuses of 1991, 2001 and 2011 We count as immigrants the residents of the province d born in a foreign country o , M t od We calculate the net flows of immigration in the intercensal period 1991-2001, I o , d , as the difference between the stock of immigrants Groizard & Mart´ ın-Montaner Migration and trade
Figure: Immigration by entry year in Census 2011
Intro and literature Empirical approach Indentification Results Data Conclusion Migration main figures Unlike migrations to the United States that date back to the 19th century, emigration to Spain is quantitatively a recent phenomenon Average foreign born residents at province-country pair (50–179) level went from 87 in 1991, to 653 in 2011 The wave of mass immigration to Spain has been a generalized shock in all provinces and from all origins Groizard & Mart´ ın-Montaner Migration and trade
Figure: Immigration flows by province, 2001 to 2011 (darker colors denote higher migration inflows).
(a) Immigration flows by continent (b) Immigration flows by country (top 10) Figure: Origins of immigrants to Spain, 2001-2011.
Recommend
More recommend
