Long-run pro-trade effects of diasporas: evidence on Italian regions

ABSTRACT

This paper assesses the diaspora effects of the Italian mass emigration of the late 19th-early 20th centuries on exports of Italian regions in the 2000s. We find statistically significant elasticities of exports to overseas economies on average higher than those estimated for European countries. The variability of the results across categories of exports provides indirect evidence on the presence of a preference effect of migrants to Argentina, while the information effect is more significant for migrants in the United States. Overall, the results suggest the new idea of searching for pro-trade effects of migration in the long run.

KEYWORDS:

• migration
• trade
• diaspora effects
• Italian regions

JEL:

• F10
• F14
• F22
• R1

ACKNOWLEDGEMENTS

The authors are grateful to Roberto Basile, Bianca Biagi, Pasquale Commendatore, Paul Elhorst, Ingrid Kubin, Enrico Pugliese, Domenico Scalera, two anonymous referees and the participants at the AISRE (Bolzano, Italy, 2018), ESCOS (Naples, Italy, 2018) and RSA (Santiago de Compostela, Spain, 2019) conferences for useful comments, and to Ettore Savoia for his great support in migration data extrapolation. Special thanks are due to Francesca Di Iorio who reviewed a fraction of the paper in some detail.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1 According to Gabaccia (1998), it was not unusual at that time for Italians in New York to raise chickens, goats and sometimes a pig in the house or basement, as well as to grow vegetables, especially tomatoes, in kitchen gardens.

2 In the United States, immigrant Italian families used to purchase olive oil at US$4 a gallon and pecorino cheese at US$1.25 a pound, even when they were economically unable to feed themselves sufficiently or to buy fuel for heating (Cinotto, 2001). In Canada, immigrants paid up to five times more in order not to be without original Italian products (Harney, 1984).

3 Bratti et al. (2014, p. 561) provide a useful summary of the results of a sample of relevant contributions to the literature in terms of the estimated elasticity of trade (both import and export) to immigration. Several studies at the regional level estimate an elasticity of export between 0.08 and 0.24: a 1% increase in the stock of immigrants raises exports from the country of origin of immigrants by 0.08–0.24.

4 As noted by Hatton and Williamson (1998), historical sources do not include comprehensive estimates of return migrants, thus preventing us from measuring net migration flows in the considered period. Officials began to record return migration at ports of entry in 1902, but even these figures are incomplete, not including migrants returning from European destinations.

5 Among specific regional patterns of migration, according Hatton and Williamson (1998) the extraordinary surge in emigration flows towards Brazil in the early 1890s is largely accounted for by Veneto. Furthermore, as it is documented by the peak in the bottom panel of Figure 4, Veneto also registered the highest migration rates to European destinations among northern regions: 31.8 (per thousands) as compared with, for instance, 19 for Piedmont and 15.5 for Lombardy.

6 This data adjustment presumably does not significantly impact on the results given that Valle d’Aosta and Molise are the smallest Italian regions as regards both population and territory, thus only marginally impacting on historical migration of the times. Current and historical borders essentially overlap for Trentino-Alto Adige. The same does not hold for Friuli-Venezia Giulia and its historical predecessor, Venetia Giulia: the former is larger in territory.

7 We have deflated nominal values using an industrial producer price index delivered by ISTAT (2018) for Italian firms active on foreign markets (reference year 2015).

8 Low-technology industries are: (1) Food products, beverages and tobacco; (2) Textiles, textile products, leather and footwear; (3) Wood, pulp, paper, paper products, printing and publishing; (4) Coke, refined petroleum products and nuclear fuel; (5) Rubber and plastics products and other non-metallic mineral products; and (6) Basic metals and fabricated metal products. High-technology industries are: (1) Chemicals excluding pharmaceuticals; (2) Pharmaceuticals; (3) Electrical machinery and apparatus, n.e.c.; (4) Office, accounting and computing machinery; (5) Machinery and equipment, n.e.c.; and (6) Motor vehicles, trailers and semi-trailers.

9 Since we run a regression for each country, the lack of variation might explain why distance is not statistically significant in the other cases. As a robustness check, we have pooled the data along the different countries and the results (available from the authors upon request) confirm that distance is negative and statistically significant.

10 These data are based on linear interpolation of regional figures available for 1881, 1901 and 1911. Annual regional GDP estimates for historical periods in the past are not available.

11 In the absence of data for 1901–04, those of 1905 were used for the following regions: Liguria, Umbria, Marche, Lazio, Basilicata and Sardinia.

12 The Mortara index is available only for the two periods, 1900–02 and 1910–12. Data for the remaining years are obtained by applying a linear interpolation between the two periods.