Foreign Languages and Trade
Jan Fidrmuc
Jarko Fidrmuc
Brunel University
University of Munich
Introduction

Impact of economic/monetary
integration well explored
Gravity models of trade
 Free-trade areas, customs unions and
monetary unions increase trade


Sharing the same language also
increases trade

Rose (2000): common language
increases trade by 50%; cf. common
currency increases trade by factor of 3
Introduction (cont’d)
Common language facilitates
communication and lowers costs
 Gravity models typically account only
for official languages
 Effect of languages measured with
dummy variables


E.g. French official language in
Canada
Introduction (cont’d)

Mélitz (2000): indigenous languages
based on Ethnologue database
 Open-circuit communication: if official
or spoken by more than 20% 
dummy variable
 Direct communication: if spoken by
more than 4%  sum of products of
percentages (capped at 1)
Our Contribution
Effect of native and foreign (learned)
languages alike
 Unique recent data set on language
proficiency in the EU

Native and up to 3 foreign languages
 Self-assessed proficiency level
 All major and all EU languages
included

Data
Special Eurobaromenter 255:
Europeans and their Languages,
November - December 2005
 Nationally representative surveys;
only EU nationals included
 Mother’s tongue and up to 3 other
languages that they speak well
enough to have a conversation
 Self-assessed proficiency: basic,
good, very good

30.0%
20.0%
10.0%
0.0%
TURKIYE
MAGYARORSZAG
ROMANIA
LIETUVA
LATVIA
PORTUGAL
ESPANA
BALGARIJA
CESKA REP.
SLOVENSKA
POLSKA
FRANCE
ITALIA
EESTI
HRVATSKA
SUOMI
ELLADA
EU27
DEUTSCHLAND
LUXEMBOURG
SLOVENIJA
BELGIQUE
ÖSTERREICH
KYPROS
DANMARK
SVERIGE
MALTA
NEDERLAND
UNITED KINGDOM
IRELAND
English: Native and Foreign Language
(good/very good skills)
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
0.0%
MALTA
PORTUGAL
ESPANA
TURKIYE
KYPROS
IRELAND
UNITED KINGDOM
ROMANIA
LATVIA
ITALIA
LIETUVA
FRANCE
SUOMI
BALGARIJA
ELLADA
EESTI
MAGYARORSZAG
POLSKA
SVERIGE
BELGIQUE
HRVATSKA
SLOVENSKA REP.
CESKA REP.
SLOVENIJA
EU27
DANMARK
NEDERLAND
LUXEMBOURG
DEUTSCHLAND
ÖSTERREICH
German: Native and Foreign Language
(good/very good skills)
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
EESTI
LATVIA
MAGYARORSZAG
TURKIYE
LIETUVA
SLOVENSKA REP.
SUOMI
HRVATSKA
POLSKA
CESKA REP.
SLOVENIJA
DANMARK
SVERIGE
BALGARIJA
MALTA
ELLADA
KYPROS
ESPANA
ÖSTERREICH
DEUTSCHLAND
IRELAND
UNITED KINGDOM
PORTUGAL
ROMANIA
ITALIA
NEDERLAND
EU27
BELGIQUE
LUXEMBOURG
FRANCE
French: Native and Foreign Language
(good/very good skills)
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
MALTA
PORTUGAL
FRANCE
BELGIQUE
NEDERLAND
TURKIYE
LUXEMBOURG
ITALIA
UNITED KINGDOM
ESPANA
SVERIGE
DANMARK
SLOVENIJA
IRELAND
ÖSTERREICH
HRVATSKA
SUOMI
MAGYARORSZAG
KYPROS
ROMANIA
ELLADA
EU27
DEUTSCHLAND
POLSKA
CESKA REP.
SLOVENSKA REP.
BALGARIJA
EESTI
LIETUVA
LATVIA
Russian: Native and Foreign Language
(good/very good skills)
Data (cont’d)

Probability that two random
individuals from two different EU27
countries speak the same language
EU27
English
German
French
Spanish
Italian
Mean
17
5
3
0
0
Min
1
0
0
0
0
Max
97
89
98
7
34
Data: Example
Communication probability
NL-S
NL-PL
English
52
14
German
7
5
French
1
0
Spanish
0
0
Italian
0
0
Gravity Model
Tijt   ij   t   1  y it  y jt    2 d ij   3 bij    DL d ,i    FL f ,i   ijt
D
F
d
f
Trade between i and j determined by
 Economic mass of i and j, yit and yjt
 Distance between i and j, dij
 Common border, bij
 Common language dummies, DLdi and
communication probabilities, DLfi
Results: EU 15, 1995-03
(1)
(2)
(3)
(4)
(5)
(6)
(7)
-20.993
***
-21.313
***
-21.445
***
15.138
***
13.833
***
-29.521
***
-32.022
***
0.897
***
0.895
***
0.896
***
0.260
***
0.278
***
1.046
***
1.086
***
-0.739
***
-0.694
***
-0.691
***
-0.733
***
-0.691
***
-0.730
***
-0.690
***
0.489
***
0.643
***
0.533
***
0.504
***
0.300
***
0.509
***
0.306
***
English
0.929
***
0.488
***
0.499
***
0.607
***
0.741
***
0.607
***
0.745
***
German
-0.009
0.216
*
0.679
***
0.668
***
French
0.012
0.628
**
-0.260
Swedish
0.544
***
0.563
***
0.293
Dutch
0.695
***
0.782
***
-0.082
0.610
***
Intercept
GDP (US$)
Distance
Border
Official Languages:
-0.185
***
0.290
***
-0.081
Proficiency:
English
0.549
***
1.040
***
0.970
***
**
French
-0.661
1.115
German
-0.174
-0.012
Spanish
6.581
Italian
-2.316
1.043
***
0.960
***
1.018
*
-0.042
***
9.931
***
10.094
***
*
9.682
***
9.714
***
TD
yes
yes
yes
yes
yes
no
no
CD
no
no
no
yes
yes
no
no
CTD
no
no
no
no
no
yes
yes
N
1800
1800
1800
1800
1800
1800
1800
R2bar
0.917
0.918
0.921
0.969
0.972
0.971
0.975
Results: EU15, 1995-03

Common official language raises trade


Especially English
English proficiency raises trade

Accounting for proficiency in English
lowers common-language effect
French/ German: weak/mixed results
 Spanish/Italian: seemingly strong effect


Most country pairs’ values close to zero
Results: EU15, 1995-03
Example 1

Consider column (7)
UK-IRL trade increased 2.1 times
because English is official language
 2.5 times because of English proficiency
 Overall effect of English on UK-IRL
trade: 5.3 fold increase
 NL-S trade is increased 1.6 times and
NL-UK trade is doubled
 Average effect in EU15: 25% increase
due to English proficiency

Results: EU15, 1995-03
Example 2

English proficiency increased by 10% in
all EU15 countries (except UK & IRL)
Average increase in trade by 12%
 Range: +8% (FR & PT) and +18%
(NL); UK & IRL trade +14%

Results: EU15, 1995-03
Example 3

English proficiency increased to NL
level in EU15 (except UK & IRL)
Average increase in trade by 61%
 Range +39% (NL) to +75% (PT)

Results: EU 15, Quantile Regressions
Q10
Q25
Q50
Q75
Q90
0.962
***
0.931
***
0.874
***
0.836
***
0.795
***
-0.464
***
-0.695
***
-0.709
***
-0.787
***
-0.852
***
Border
0.673
***
0.483
***
0.687
***
0.591
***
0.319
***
English official language
1.088
***
0.890
***
0.433
**
0.426
***
0.400
***
English proficiency
0.304
0.340
***
0.697
***
0.426
***
0.272
***
-23.557
***
-20.109
***
-17.209
***
-14.193
***
GDP (US$)
Distance
Intercept
-27.083
***
N
1800
1800
1800
1800
1800
Pseudo R2
0.738
0.735
0.722
0.716
0.714
Results: EU15, Quantile
Regression
Effect of English proficiency highest for
50th percentile  outlier-free
regression
 ???

Conclusions
Language has a strong effect on trade
 Countries with common official
language trade more with each other
 Proficiency in foreign languages also
raises trade
 English plays leading role
 On average, trade in EU15 is higher
by ¼ because of average proficiency
in English

Conclusions (cont’d)
Universal proficiency in English would
raise trade 2.5 times
 Rose: estimated effect of monetary
unions  2-3 fold increase in trade


Common currency costly (OCA theory)
Improving English proficiency does
not require abandoning national
languages
 Large gains possible at little cost

Descargar

Foreign Languages and Trade