Immigration and the Trade of Provinces
| Author | Keith Head,Don Wagner,John Ries |
| Published date | 01 November 2002 |
| DOI | http://doi.org/10.1111/1467-9485.00245 |
| Date | 01 November 2002 |
IMMIGRATION AND THE TRADE OF
PROVINCES
Don Wagner*, Keith Head **and John Ries **
ABSTRACT
A link between immigration, imports, and exports has been found by a number of
papers that have used the gravity equation to analyze bilateral trade patterns. We
discuss what this research implies about the mechanisms through which immigrants
expand trade and identify strengths and weakness of the various approaches. This
paper also contributes to this literature by estimating immigrant effects for Canada
using cross-province variation in international trade and immigration patterns. We
derive an alternative functional form capturing the relationship between immigra-
tion and trade based on the proposition that immigrants use their connections and
superior ‘market intelligence’ to exploit trade opportunities that non-immigrants do
not access. We find that the average new immigrant expands exports to his=her
native country by $312 and expands imports by $944.
II
NTRODUCTION
The gravity model of international trade has consistently revealed a strong
association between immigration and trade. Not only have different studies
revealed a robust relationship for different samples and specifications, the
strength of the immigration effect varies in sensible ways for different trading
partners, products, and types of immigrants. The estimated magnitude of the
immigration effect, however, differs greatly across studies. The analysis of Head
and Ries (1998), Dunlevy and Hutchinson (1999, 2001), Rauch and Trindade
(2002), Girma and Yu (2002) and Combes et al. (2002) based on cross-sectional
information find large effects whereas Gould’s (1994) estimation based on times
series variation indicates smaller effects. The discrepancy from an econometric
standpoint is easy to explain— cross-sectional estimates may be upwardly biased
due to unobserved characteristics of trading relationships whereas fixed effect
estimates may have the opposite bias due to the magnification of measurement
error caused by this technique. Alternative explanations for the discrepancy are
differences in specifications and samples.
Scottish Journal of Political Economy, Vol.49, No. 5, November 2002
#Scottish Economic Society 2002,Publ ishedby Blackwell Publishers Ltd, 108 Cowley Road, Oxford OX4 1JF, UK and
350 Main Street, Malden, MA 02148, USA
507
*University of Prince Edward Island
**University of British Columbia
This paper reviews studies of the immigration-trade link to establish what we
know so far about the mechanisms by which immigrants increase trade and the
magnitude of the effects. We identify the consistent findings in the literature as
well as conflicting ones. We also evaluate the different approaches employed in
existing studies and discuss their strengths and weaknesses. Finally, we introduce
new estimates based on the trade of Canadian provinces. These results are
derived from specifications explicitly linked to theories underlying the
immigration-trade link. The estimates exploit cross-sectional information on
trade and immigration across Canadian provinces and control for fixed effects
between Canada and its trading partners. We also report how controls for
language commonality between a Canadian province and a foreign trading
partner influence the immigration effect on trade.
In our review of the literature, we discuss a number of methodological issues.
The issues addressed in our commentary extend beyond immigration-trade
studies to the large number of recent papers using the gravity model to evaluate
the factors influencing trade such as a common currency, common language,
and foreign aid on trade levels.1One important issue is specification. Gravity
models posit a log-linear relationship between trade volumes, source and
recipient country GDPs, and trading distances. Generally, researchers simply
insert measures of other factors into a gravity model without regard to
theoretical considerations. A second issue is regression technique. Estimates may
alternatively be based on cross-sectional (country) variation or time-series
variation. A third issue concerns endogeneity. Is the association between trade
and immigration causal or driven by unmeasured common factors?
The paper has the following structure. Section II reviews the literature and
identifies issues warranting further study. We argue that our study based on
Canadian provincial trade addresses some of these issues. Our first specification
is a standard one that employs country-fixed effects and introduces a detailed
language variable measuring the overlap between languages spoken in Canada
and its trading partners. Our ensuing specifications examine how the results
change when we employ specifications based on theories predicting how
immigrants expand trade. These results are reported and discussed in
Section III. We conclude by summarizing our current knowledge of the trade
and immigration link and suggesting areas for future research.
II THE LITERATURE
The papers we review are Gould, Head and Ries, Dunlevy and Hutchinson,
Rauch and Trindade, Girma and Yu, and Combes et al. Gould studies US trade
with 47 trading partners over the 1970– 1986 period, whereas Head and Ries
consider Canadian trade with 136 countries from 1980 to 1992. Dunlevy and
Hutchinson evaluate US imports from 1870– 1910 and they analyze US exports
ova the same period in their 2002 manuscript. Rauch and Trindade do not
1Wagner (forthcoming) considers the effects of overseas development assistance; Frankel and
Rose (2002) a common currency; and Helliwell (1999) common language.
508 DON WAGNER, KEITH HEAD AND JOHN RIES
#Scottish Economic Society 2002
To continue reading
Request your trial