Somatic distance, trust and trade

• Author: Jacques Melitz,Farid Toubal
• Date: 01 August 2019
• Published date: 01 August 2019
• DOI: http://doi.org/10.1111/roie.12397

786

|

© 2019 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/roie Rev Int Econ. 2019;27:786–802.

1

|

INTRODUCTION

In sociology, homophily is taken as an established fact: individuals tend to associate and bond with

similar others, as in the proverb “birds of a feather flock together.” Also, “Homophily in race and

ethnicity creates the strongest divides in our personal environments” (McPherson, Smith‐Lovin, &

James Cook, 2001, p. 415). The authors of this oft‐cited article go on: “Age, religion, education, occu-

pation, and gender [follow] in roughly that order.” Indeed, discrimination based on race and ethnicity

is readily apparent in economics too, especially in labor studies (see Bertrand & Mullainathan, 2004,

and Edo, Jacquemet, & Yannelis, 2017, concerning job applications; Lang & Lehmann, 2012, and

Borowczyk‐Martins, Bradley, & Tarasonis, 2017, concerning wage earnings and employment). Yet in

studies of bilateral trade between countries, homophily is mostly absent from the list of determinants.

It emerged only recently in a highly influential article by Guiso, Sapienza, and Zingales (2009) (here-

after GSZ). Since they wrote, two works have picked up on the theme: Spring and Grossmann (2016)

(in a critical spirit) and Yu et al. (2015). GSZ introduced somatic distance or difference in physical

appearance based on an Italian source: a work by Biasutti (1954, first edition) in four volumes, which

Received: 9 October 2018

|

Revised: 27 December 2018

|

Accepted: 4 January 2019

DOI: 10.1111/roie.12397

ORIGINAL ARTICLE

Somatic distance, trust and trade

JacquesMelitz1

|

FaridToubal2

1ENSAE and CEPII, Paris

2ENS de Paris‐Saclay, CREST and CEPII

Correspondence

Jacques Melitz, ENSAE and CEPII,

20 Avenue de Segur, Paris, 75007.

Email: j.melitz@hw.ac.uk

Abstract

Somatic distance, or differences in physical appearance,

proves to be extremely important in the gravity model of

bilateral trade in conformity with results in other areas of

economics and outside in the social sciences. This is also

true independently of survey evidence about bilateral trust.

These findings are obtained in a sample of the 15 members

of the European Economic Area in 1996. Robustness tests

also show that somatic distance, as well as co‐ancestry, has

a more reliable influence on bilateral trade than the other

customary cultural variables. The article finally discusses

the interpretation and breadth of application of these

results.

JEL CLASSIFICATION

F10, F40, Z10

|

787

MELITZ and TOUBaL

summarizes and extends a huge literature on racial differences in physical anthropology. However,

while making use of Biasutti, GSZ and the two aforementioned studies, subordinate the whole issue of

somatic distance by treating it as affecting bilateral trade essentially via trust: that is, as an instrument

for trust in an instrumental‐variable‐regression interpretation.1 In this contribution, we shall instead

introduce somatic distance as a direct influence on bilateral trade right from the start and show that it

belongs there.

In his path‐breaking work on The Economics of Discrimination, Becker (1957) modeled discrimi-

nation in the labor market as founded on the distaste of employers for interacting with certain groups

of workers. His view has undergone subsequent modification, with allowances for the significance of

beliefs about other people and some rational elements in labor market discrimination (for example,

Phelps, 1972; Arrow, 1973). But as the literature in the preceding paragraph is witness, his basic posi-

tion stands: racial prejudice as such retains a firm position in explanations of racial discrimination. It

is clear that such prejudice partly stems from and partly breeds distrust. But there is no precedent for

viewing the prejudice and, broadly, somatic distance in general, as impinging on international trade

strictly through the filter of trust. No one would argue that sex discrimination is only a matter of trust.

There should be little question either that discrimination based on physical appearance is not strictly

a matter of trust. In the specific case of international trade, it is standard, since Armington (1969), to

allow that national preferences for different trade partners may intervene in explaining bilateral trade.

Should somatic distance be one of the reasons for these national preferences, personal affinities could

well be the source, independently of trust. The mechanism could take several forms. For example, an

increase in the number of foreign markets to which national firms export raises their fixed costs and

necessitates choices. In making these choices, “animal spirits” may operate and thus explain why

somatic distance enters. As another example, exporters of consumption goods might find that their

wares have more appeal to foreigners who resemble them, partly because of similar tastes, but partly

also because of a preference for associating with them in commerce.

Once we admit that somatic distance has a direct place in a gravity equation for bilateral trade, the

variable emerges as highly significant. It remains so in the presence of other cultural factors, reflecting

language, religion, law, co‐ancestry, and the history of wars, as well as sample evidence from ques-

tionnaires about trust. Indeed, it is more robust than the rest. Trust, based on questionnaire evidence,

and the history of wars never matter. Since immigrants are particularly important in studying cultural

interactions, we introduce them. This has a seriously damaging effect on two important cultural vari-

ables, same legal origin and common religion. Adding a population‐weighted measure of physical

distance in the presence of immigrants notably reduces the significance of Common native language

too. Somatic distance is hardly affected throughout. Somatic distance even outperforms co‐ancestry

though this last variable also holds up well in a new guise.

All these results occur in a European sample close to GSZ’s. There are two strong reasons for stick-

ing close to this sample. The more important one is that we want to control for trust and we know no

reasonable alternative to GSZ’s measure. But secondly, even if we were to drop GSZ’s trust variable,

we could not extend the analysis very far, only to the rest of Europe outside the European Economic

Area (EEA) in 1996, since Biasutti’s data for somatic distance permits going no further. This would

essentially add Eastern Europe and Switzerland.

As indicated, both somatic distance and co‐ancestry, or two different aspects of genetic distance,

appear important. The two variables matter singly and jointly. Yet our emphasis will be solely on so-

matic distance, essentially because of its wrongful neglect and its confusion with trust.

The next section offers the test evidence, the following one provides robustness tests concerning

the significance of somatic distance, and the last one engages in general discussion and interpretation.