Journal of Interdisciplinary History, LII:4 (Spring, 2022), 537–564.

Blanca Sánchez-Alonso and Carlos Santiago-Caballero
Spain’s Loss of Human Capital after the Civil War:
Spanish Refugees in Mexico Forced migrations, exiles,
and genocides adversely affect millions of people. The literature
stresses the obvious differences between forced and voluntary
migration. Refugees do not choose their country of destination
or the time when they move; refugees feel the “push factors” in
their countries of origin more than they do the economic “pull
factors” in the destination countries. They are assumed to be eco-
nomically motivated to a much lesser degree than are economic
migrants. Refugees typically arrive in a host country with less
locally applicable human capital, including language and job skills,
than do economic migrants.

This article focuses on the Spanish exile to Mexico that
followed the Civil War. This exile differs substantially from the gen-
eral picture regarding refugees. Many of the Spanish refugees landed
in Mexico because other possible options were blocked, risky, or
undesirable (for instance, remaining in France or re-emigrating to
another country) and, particularly, because Mexico opened the door
to Spanish Republicans. They spoke the same language and presum-
ably had professional skills above average, compared to the Mexican

Blanca Sánchez-Alonso is Professor of Economic History, Universidad San Pablo-CEU. She is
the author of “The Age of Mass Migration in Latin America,” Economic History Review, LXXII
(2019), 3–31; with Leticia Arroyo Abad, “A City of Trades: Spanish and Italian Immigrants in
Late-Nineteenth-Century Buenos Aires, Argentina,” Cliometrica, XII (2018), 343–376.

Carlos Santiago-Caballero is Associate Professor of Economic History, Universidad
Carlos III. He is the author of “Domestic Migrations in Spain during its First Industrialization,
1840s–1870s,” Cliometrica, XV (2021), 535–564; “Intergenerational Occupational Mobility in
Nineteenth Century Spain ( Valencia), 1841–1870,” Revista de Historia Económica—Journal of
Iberian and Latin American Economic History (New Series), XXXIX (2021), 219–264.

The authors thank an anonymous referee for the improvements and the editors of the
journal for their help and suggestions. They are also grateful for ﬁnancial support from Uni-
versidad San Pablo-CEU (Project MCP20V03). Carlos Santiago-Caballero thanks the the
members of the Economic History Department at the London School of Economics for their
support writing this article.

This work has been supported by the Madrid Government (Comunidad de Madrid-
Spain) under the Multiannual Agreement with UC3M in the line of Excellence of University
Professors (EPUC3M05), and in the context of the V PRICIT (Regional Programme of
Research and Technological Innovation).

© 2022 by the Massachusetts Institute of Technology and The Journal of Interdisciplinary
History, Inc. Published under a Creative Commons Attribution 4.0 International (CC BY
4.0) license., https://doi.org/10.1162/jinh_a_01766

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

538
population. In contrast to the situation for today’s refugees, Mexican
citizenship was granted immediately, allowing the exiles to look for
employment without any legal or administrative barriers.1

Although the demoralizing effects of exile are dramatic and
direct, the secondary consequences, such as the loss of a signiﬁcant
stock of human capital in the country of origin, are usually under-
valued. Recent scholarship emphasizes the importance of the drain
caused by forced migrations and genocides, of which the Republican
exile of post–Civil War Spain is an excellent case. The literature
related to the issue is abundant in both quantity and quality, ranging
from the personal experiences of the exiles to the number of people
involved. The traditional view suggests that the importance of the
Spanish exile lies not only in its sheer volume but also in its break-
down of labor—farmers, merchants, and blue-collar workers along
with university professors, teachers, engineers, and liberal profes-
sionals. The literature treats this exile as a brain drain for Spain.
Whereas research paints a consistent picture of contemporary refu-
gees as disadvantaged both socially and economically relative to other
immigrants at arrival, the literature about the Spanish exile points to a
ﬂow of highly qualiﬁed immigrants not at all economically disadvan-
taged compared to the Mexican population.2

Little has been said, however, about the quantiﬁcation of
Spain’s loss of human capital, or the measurement of its quality,
after the Civil War. Unfortunately, we cannot calculate the total
loss of human capital due to exile at that time, only that relating to
Mexico, which is considered to be at the highest end. We do not
have data sufﬁcient to estimate the human capital of Spanish ref-
ugees in France (the large majority) or in other destinations, such

1 Timothy J. Hatton, “Asylum Migration to the Developed World: Persecution, Incentives,
and Policy,” Journal of Economic Perspectives, XXXIV (2020), 75–93; Courtney Brell et al., “The
Labor Market Integration of Refugee Migrants in High-Income Countries,” ibid., XXXXIV
(2020), 94–121. In the short run, many immigrants refused to become Mexican citizens and
retained their passports, anticipating a return to Spain when Francisco Franco fell from power.
The only condition that the Mexican government established for Spanish exiles was refraining
from involvement in Mexican politics.
2 Daron Acemoglu, Tarek A. Hassan, and James A. Robinson, “Social Structure and Devel-
opment: A Legacy of the Holocaust in Russia,” Quarterly Journal of Economics, CXXVI (2011),
895–946; María del Rosario Ruiz Franco and Sergio Riesco Roche, “Veinte años de produc-
ción histórica sobre la Guerra Civil Española (1975–1995): Una aproximación bibliométrica,”
Revista española de documentación cientíﬁca, XXII (1999), 174–197; Brell et al., “Labor Market”;
Gerhard Towes and Pierre-Louis Vézina, “Enemies of the People,” Working Paper 2020-20,
Quantitative Political Economy Research Group (King’s College London, 2020).

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 539

as the French colonies in North Africa, the Soviet Union, or other
Latin American countries.3

This article places its estimation of the quality of human capital
that left Spain for Mexico during the exile into a broader context,
comparing it with the years that preceded and followed it. The ini-
tial results show that, depending on the estimator used, the level of
accomplishment of those exiled to Mexico almost doubles that of
other more economically motivated migrants at the same time. This
article’s further estimation of the female human capital lost to Spain
in the Mexican exile is an important contribution given the tradi-
tional invisibility of women in recorded economic history. Female
human capital, in fact, represented a signiﬁcant proportion of the
human capital lost by the Republican ﬂight to Mexico.

THE SPANISH EXILE From January to March 1939, around 400,000
Spanish refugees crossed the French border before the end of the
Spanish Civil War. Most of them crossed the Pyrenees from the
neighboring regions of Catalonia, Aragon, Navarre, and the Bas-
que Country or from other more distant areas such as Valencia.
Many returned in the following months; by December 1939, only
140,000 Spaniards remained in France. Facing the problem of
Spanish refugees, the French government began negotiations with
Latin American countries to relocate the Spanish refugees, but
only Mexico, Uruguay, and Chile responded positively. The
Mexican government of President Lázaro Cárdenas agreed to open
the borders to Spanish Republican exiles with no limits. As a
result, Mexico (with a modest history of receiving Spanish emi-
grants) accounted for 12 percent of all exiles, thus becoming the
second-largest recipient behind only France.4

3
Information about the occupational composition of Spanish refugees in France ( July 1939)
reveals that a higher concentration of them were in the secondary sector (45%) than were in
the traditional migrant population in France. See Javier Rubio, La emigración española a Francia
(Barcelona, 1974), 230 (Table 51).
4 For the number of exiles who crossed into France, see Rubio, La emigración española a
Francia (Barcelona, 1974). See also Dolores Pla Brugat, “La presencia española en México,
1930–1990: Caracterización e historiografía,” Migraciones y Exilios, II (2001), 157–188. Initially,
the French government forced repatriations until May 1939 after which such repatriations to
Spain were banned because of protests in France against Francisco Franco’s repression. For an
estimation of a similar percentage, around 12%, see José Ortega and Javier Silvestre, “Las con-
secuencias demográﬁcas,” in Pablo Martín Aceña and Elena Martínez Ruíz (eds.), La economía
de la Guerra Civil (Madrid, 2006).

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

540

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

An estimated total volume of permanent emigration pro-
duced by the Spanish Civil War is about 190,000 people, mostly
to France. Ortega and Silvestre offer a lower ﬁgure (162,000) for
the net emigration from Spain between 1936 and 1944. The mag-
nitude is small when compared to traditional Spanish emigration as
well as to other political emigrations—for example, Russia’s after
the Revolution and Civil War from 1917 to 1920 or Europe’s in
general after World War II.5

Table 1 shows the basic numbers of Spanish emigration to
Mexico for our period of analysis. Not all these migrants were refu-
gees, but most of them were certainly exiles from the Civil War,
particularly from 1939 to 1942. According to Gallardo, the average
number of Spanish economic immigrants who arrived in Mexico
between 1909 and 1927 was 4,000 per year, but the ﬂow reduced sub-
stantially during the 1930s (1,400 per year between 1930 and 1936).6
Spanish historiography has attached extraordinary importance to
the emigration of Republican refugees to Mexico for several reasons:
First, unlike recent asylum seekers in the European Union (1997–
2014), who generally requested a destination country with other
immigrants of the same national origin, the Spanish exiles went not
to Argentina, Uruguay, or Cuba—where Spanish immigration had a
strong foundation—but to Mexico, solely at the political invitation of
President Cárdenas. Second, contrary to the portrait of traditional
Spanish migration to America (including Mexico), the literature
about the Spanish exile depicts a highly accomplished group who
represented a considerable loss of human capital for Spain. Pla Brugat
observes that the forced migration of Republican exiles to Mexico
included many elite intellectuals, even though it was a much more
heterogeneous movement than initially thought.7

The extent to which Mexico is a good proxy for the Repub-
lican exile is an important question. We considered including the
French exile ﬁgures in our calculations of lost human capital, but

5 Rubio, La emigración española a Francia (Barcelona, 1974), 228. Between 1910 and 1913, the
total volume of Spanish emigration reached nearly 900,000 people; in the 1920s, the peak was
650,000 from 1920 to 1923. See Sánchez Alonso, Las causas de la emigración española (Madrid,
1995). Ortega and Silvestre, “Las consecuencias demográﬁcas.”
6 César Yáñez Gallardo, La emigración española a América (siglos XIX y XX): Dimensión y car-
acterísticas cuantitativas (Colombres, 1994).
7 For twenty-ﬁrst century refugees, see Hatton, “Refugees, Asylum Seekers, and Policy in
OECD countries,” American Economic Review, CVI (2016), 441–445; Jose Luis Abellán García
González (ed.), El exilio español de 1939 (Madrid, 1977); Pla Brugat, “La presencia española.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 541

Table 1

Spanish Immigrants Arriving in Mexico, 1937–1948

MALE

FEMALE

CHILDREN

TOTAL

1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948

91
45
3,884
1,034
898
1,492
124
262
305
836
1,408
498

Total

10,932

96
72
2,352
712
713
1,042
160
248
282
592
1,035
522

7,826

36
33
1,161
309
306
521
53
123
122
195
349
157

3,335

223
150
7,397
2,055
1,917
3,055
307
633
709
1,618
2,852
1,177

22,093

SOURCE Calculated from Lida and Pacheco Zamudio, “El perﬁl de una inmigración: 1821–
1939,” in Lida (ed.), Una inmigración privilegiada: comerciantes, empresarios y profesionales en México
en los siglos XIX y XX (Madrid, 1994), 34.

the obstacles were too great: Although hundreds of thousands of
Spaniards crossed the frontier toward France, many of them
returned to Spain, often because of appalling living conditions in
France. Moreover, many Republican exiles reported that their
welcome in France was far from hospitable.

As still happens today, the Spanish refugees were housed in ref-
ugee camps that did not comply with the most basic living standards.
The poor sanitary conditions and the hard winter of 1939 produced
numerous casualties, and the French authorities initially did not allow
Spanish medical personnel to give medical care to their countrymen.
The options open to these Republican exiles in France—from direct
repatriation to impressment into the French Foreign Legion—were
equally dire. Franco’s regime sent emissaries to the refugee camps to
encourage people to return, reinforcing repatriation to Spain.8

Pla Brugat estimates that around one-half of the Republicans
relocated to France in June 1939 were employed in the secondary
(manufacturing) sector, one-third in the primary (agricultural)

8 For health conditions, see María Fernanda Mancebo, La España de los exilios ( Valencia,
2008), 96; for the situation in the camps, Juan Carlos Pérez Guerrero, La identidad del exilio
republicano en México (Madrid, 2008), 77; Marie-Cleude Rafaneau-Boj, Los campos de concentra-
ción de los refugiados españoles en Francia (1939–1945) (Barcelona, 1995), 149.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

542
sector, and nearly one-ﬁfth in the tertiary sector (services). But this
estimation, which might serve as a crude proxy of human capital,
changed during the following months when many of the exiles
returned to Spain. Although we can speculate about who went
back, the difﬁculties involved in identifying the proﬁle of those
who did, make the direct incorporation of French exiles into
the calculation of human capital difﬁcult.9

Was the proﬁle of the Mexican Republican exile representative
of the Spanish population? Pla Brugat used a sample of 4,660 exiles
who arrived in three ships (the Sinaia, Ipanema, and Mexique) during
the summer of 1939 to show that nearly 50 percent of them were
employed in the tertiary sector, 15 percent in the liberal professions,
and 13 percent in education as teachers, university professors, and
intellectuals. The Spanish population census of 1930 placed a total
of 27 percent of the population in the tertiary sector. Although these
data from the early arrivals to Mexico may not be representative of
the total ﬂow, the typical Mexican exiles were supposedly not a
good mirror of Spanish society. According to this interpretation,
they were comprised primarily of the most cultivated people within
the social spectrum. The numerous studies analyzing this movement
across the Atlantic after the Civil War, however, contain no deep
discussion about how the departure of scientists, artists, or politicians
affected the home country. Their loss surely would have dealt a
severe blow to the groups in which they had been leading partici-
pants. Political repression in Spanish universities and scientiﬁc cen-
ters during the 1940s did the rest of the dirty work, destroying
scientiﬁc and academic networks throughout Spain.10

9 Pla Brugat, Els exiliats catalans a Mexic: Un estudi de la immigració republicana (Afers, 2000).
Idem, “Características del exilio en México en 1939,” in Clara Eugenia Lida (ed.), Una
10
inmigración privilegiada: comerciantes, empresarios y profesionales en México en los siglos XIX y XX
(Madrid, 1994), 218–231; Juan Bautista Vilar, La España del exilio: Las emigraciones políticas espa-
ñolas en los siglos XIX y XX (Madrid, 2006), 360; Pedro L. Angosta, La República en México: Con
plomo en las alas (1939–1945) (Salamanca, 2009). Blas Cabrera Felipe, a leading physicist who
communicated with Albert Einstein, Neils Bohr, and Marie Curie, among others, received an
invitation to the Solvay Conference, a meeting of the world’s most illustrious scientists. For
the effects on the scientiﬁc community, see Josep Lluís Barona, “Los cientíﬁcos españoles exi-
liados en México,” in Los refugiados españoles y la cultura mexicana: Actas de las primeras jornadas
(Madrid, 1998), 109; Francisco Giral, Ciencia Española en el Exilio (1939–1989): El Exilio de los
Cientíﬁcos Españoles (Madrid, 1994). For the effects of repression on the research community in
Spain in the years following the Civil War, see López García, “La investigación cientíﬁca y
técnica antes y después de la guerra civil,” in Gómez Mendoza (ed.), Economía y sociedad en la
España moderna y contemporánea (Madrid, 1996), 256–276.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 543

Yet this emphasis on intellectuals and scientists is not consistent
with what the data show. The analysis of the aggregate data suggests
that intellectuals did not dominate the Mexican exile. Given the
outsized motivational importance that noneconomic factors have
for refugees, such populations are likely to include individuals from
a wide range of occupations more suited to their country of origin
than to their destination, as well as demographic types unlikely to
migrate primarily for economic reasons. Our data support this view.
Our aim is to go a step beyond the traditional representation and to
clarify both the quantity and the quality of the human capital that
left Spain and went to Mexico after the end of the Civil War.11

SOURCES AND DATA The main source used in this article is the
National Registry of Foreigners in Mexico—created in 1926 by
the Mexican government’s General Direction of Migratory
Services—which contains information about the Spanish immi-
grants in the country. It also documents any foreigner of other
nationality who had entered Mexico before May 1926. All the
migrants who lived in the country received a card with their per-
sonal information, issued as proof of their legal status. The registry
has information about several generations of immigrants to Mex-
ico starting in the mid-nineteenth century; from 1929, all new
immigrants were registered with a similar document.

The information in the registry provided by the expatriates
was highly detailed. The standard document had two pictures—
one facing to the front and one to the side—date of entry; infor-
mation about eyebrows, eyes, mouth, or facial hair; height; date
and place of birth; occupation; religion; and number of languages
spoken (see Figure 1). The cards are in Mexico’s General Archive
of the Nation. The Spanish Ministry of Culture digitalized them
for its online archival system, thus permitting us to consult and
transcribe the information from a sample of more than 26,000
cards—half the number of the Spanish migrants in The National
Registry. We extracted each immigrant’s given name and sur-
name, year entering Mexico, year of birth, Spanish province of
birth, gender, age, occupation, status as a political refugee (or
not), last reported residence, transportation taken to enter Mexico
(ships identiﬁed by name), height, foreign languages spoken, and

11 Pla Brugat, Els exiliats catalans; Brell et al., “Labor Market.”

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

544

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

Fig. 1 Example of Card Issued to Immigrants

SOURCE PARES (https://pares.culturaydeporte.gob.es/inicio.html).

religion. To compare exiles and traditional migrants, our sample
contains records only for individuals older than eighteen years.12
The information obtained enables a proﬁle of immigrants in
Mexico before 1936 and a proﬁle of Republican exiles. In addition
to proﬁling different sorts of migrations, this article also estimates
the human capital of the people who arrived in Mexico between
the mid-1920s and the late 1940s. The comparison of exiles with
traditional economic migrants situates the human capital of the
exiles into a broader context and isolates the effect of belonging
to an exiled group, an exercise apparently never before undertaken.
The ﬁrst step is to deﬁne human capital and to explicate the estima-
tors selected to measure it. Unfortunately, the immigration records of
the Mexican authorities do not offer information about levels of formal
education, but occupation can serve as a proxy to measure degrees of
human capital. Everyone entering the country had to provide detailed
information about his/her working experience; immigrants already
living in Mexico gave details of their current occupations.

We used the HISCAM scale proposed by Lambert et al. to adapt
the information from the occupations into a quantitative index in
which higher values represent more advantaged occupational

12 Apart from its large size, our sample is random; we double-checked its randomness by
comparing it with smaller samples extracted from the same source that used a systematic
random-selection procedure.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 545

positions. This proxy for human capital, however, is not free of
methodological problems; the extent to which a declared occupa-
tion was related (or not) to an individual’s actual stock of human
capital is a factor beyond our control. Another problem pertains
to the diligence of the Mexican agents who interviewed the immi-
grants. In some cases, the civil servants grouped occupations into
main categories, surmising, for example, that members of an
orchestra/band and composers were all musicians. Another prob-
lem is female migrants’ lack of heterogeneity; the fact that housewife
was the most frequent occupation of women hindered our ability to
distinguish differences in women’s level of human capital.13

We also observe a certain temporal inconsistency in the level
of detail offered by immigrants in relation to their occupations.
After the mid-1930s, the description of occupations became
detailed, providing not just a general description (say, engineer)
but one more ﬁne-grained (agricultural engineer or industrial
engineer). The job descriptions in the sources before the 1930s
are not so generous. In many cases, immigrants declared them-
selves only as employees, traders, or merchants.

The records about the immigrants who entered Mexico
before 1926 and were already established when the exiles arrived
are the least detailed. Because the assessment of human capital for
immigrants who arrived before 1930 is less reliable, we searched
for alternative estimators of human capital. One of them is the
number of languages other than Spanish that an immigrant could
speak. That trait allows us to identify, for example, differences
between women recorded as housewives much better than occu-
pation listings do. A housewife from a wealthy family probably
received a better education in foreign languages than one born
to a poor family. The use of foreign languages also solves the
potential problem of having to rely on general occupational terms,
such as employee or merchant, for social capital before 1930.14

We excluded from this proxy those languages that qualify as
mother tongues rather than foreign languages—Catalan, in the

13 Paul S. Lambert et al., “The Construction of HISCAM: A Stratiﬁcation Scale Based on
Social Interactions for Historical Research,” Historical Methods, XLVI (2013), 77–89; Vicente
Llorens, El exilio español de 1939: La emigración republicana de 1939 (Madrid, 1976).
14
the quality of education received in Spain.

Speaking a language other than Spanish has no relevance to living in Mexico but shows

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

546

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

case of migrants born in Catalonia, or Basque, for those born in
the Basque Country. The use of this proxy likely captures the
attainment of an elevated formal education. For instance, a highly
qualiﬁed worker, such as a master carpenter, who does not speak a
foreign language would be classiﬁed as low-skilled. The capacity
to learn foreign languages, however, presumably requires natural
skills that signify human capital beyond formal education or occu-
pation per se. In fact, the average number of languages spoken by
immigrants every year tends to correlate closely with the estima-
tors derived from occupational categories such as HISCAM. We
further transcribed the social classes of individuals according to
the HISCLASS scheme that divides occupations into ﬁve social
classes—(1) Elite (higher managers and higher professionals), (2)
lower middle class (lower managers, professionals, clerical and sales
personnel, and foremen), (3) self-employed farmers and ﬁshermen,
(4) medium-skilled and lower-skilled workers, and (5) unskilled
workers and farm workers.15

Finally, we also take migrants’ stature as a proxy of human
capital, given the literature’s suggestion of a high correlation
between both dimensions. Camara et al. showed that from 1855
to 1960—a period similar to the one covered in this article—
illiterate Spanish conscripts were generally shorter than those
who could write and read and even shorter than those who were
students. Similarly, Huang shows that the stature of Dutch con-
scripts in the mid-twentieth century also aligned with their level
of education. Silventoinen and Lahelma found the same correla-
tion in Finland and Sweden between 1920 and 1969, as did
Heineck in Germany between 1952 and 1981.

The correlation between human capital and height is not
restricted to historical studies. Meyer and Selmer’s study proved
that during the late twentieth century, educated men and women
tended to be taller. This correlation is hardly surprising, since adult
stature depends on nutrients and a child’s physical conditions dur-
ing the growing period; both variables are highly correlated with
the socioeconomic status of their families of origin. In a society
where parents transmitted their socioeconomic status to their

15 The correlation between languages and HISCAM variables is 0.66 for the whole period and
0.76 between 1900 and 1950. See Marco H. D. van Leeuwen and Ineke Maas, HISCLASS: A
Historical International Social Class Scheme (Leuven, 2011).

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 547

children, height is also a reasonable indirect proxy for both socio-
economic status and human capital.16

As in the case of foreign languages, one of the strengths of height
is its avoidance of the problems that attach to general occupations
such as housewife, employee, or merchant. Yet Spain also had impor-
tant regional differences in height that could have been natural rather
than social in origin. Quiroga discovered signiﬁcant regional differ-
ences in the height of Spanish conscripts between 1893 and 1954;
the Canary Islands consistently ranked at the top whereas the regions
of the interior ranked at the bottom. To this regional disparity, which
might have been partly genetic, we observe that trends in the average
height in Spain’s different regions over time were not homogeneous,
as Quiroga and Martinez Carrión explained. Therefore, the inclusion
of regional controls is necessary when comparing the differences in
height between individuals from different provinces.

Because stature tends to diminish after the age of ﬁfty, due to
the compression of the discs between the vertebrae, adjustments
were also necessary to make the height of individuals measured
at different ages comparable. Fernihough and McGovern esti-
mated an annual reduction in height between 0.08 and 0.1 percent
for males and 0.12 and 0.14 percent for females. We used the aver-
age of these ranges to adjust heights for migrants older than ﬁfty.17
Our ﬁrst approach to capture the potential differences between
exiles and traditional migrants was based on a descriptive analysis

16 Antonio D. Cámara et al., “Height and Inequality in Spain: A Long-Term Perspective,”
Revista de Historia Económica / Journal of Iberian and Latin American Economic History, XXXVII
(2019), 205–238; Ying Huang et al., “Differences in Height by Education among 371,105
Dutch Military Conscripts,” Economics & Human Biology, XVII (2015), 202–207; Karri
Silventoinen et al., “Body Height, Birth Cohort and Social Background in Finland and
Sweden,” European Journal of Public Health, XI (2001), 124–129; Guido Heineck, “Height
and Weight in Germany, Evidence from the German Socio-Economic Panel, 2002,” Econom-
ics & Human Biology, IV (2006), 359–382; Haakon E. Meyer and Randi Selmer, “Income,
Educational Level and Body Height,” Annales of Human Biology, XXVI (1999), 219–227.
For the transmission of status, see Santiago-Caballero, “Intergenerational Occupational
Mobility in Nineteenth Century Spain ( Valencia), 1841–1870,” Revista de Historia Económica /
Journal of Iberian and Latin American Economic History, XXXIX (2021), 219–264.
17 Gloria Quiroga, “Estatura, diferencias regionales y sociales y niveles de vida en España
(1893–1954),” Revista de Historia Económica, XIX (2001), 175–200; José Miguel Martinez
Carrión, “Estatura, salud y bienestar en las primeras etapas del crecimiento económico español:
Una perspectiva comparada de los niveles de vida,” Documento de trabajo N° 0102 2001
(Asociación Española de Historia Económica, 2001); Alan Fernihough and Mark E.
McGovern, “Physical Stature Decline and the Health Status of the Elderly Population in
England,” Economics & Human Biology, XVI (2015), 30–44.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

548
comparing the proﬁle of the exiles with that of the traditional
migrants. We later analyzed the evolution of our three human-capital
proxies over time to check not only whether their values increased
after the arrival of the exiles but also whether the changes are unex-
pected or in line with the time trends discovered for traditional
migrants.

COMPARING EXILES WITH TRADITIONAL MIGRANTS To evaluate
Spain’s loss of human capital because of the Mexican exile, we
must place it into a broader context. A proper understanding of
the exile’s effect requires comparing it with migrations to Mexico
before, during, and after the arrival of the Republican refugees.
Were refugees more skilled than traditional immigrants? If so, by
how much? What about those who moved to Mexico at the same
time and in later periods but did not register as political refugees?
Traditional immigration and Republican exile represented distinct
modes of migration, each with its own features.

In the age of mass migration, Mexico—unlike Argentina,
Brazil, Uruguay, and Cuba—was not a preferred destination for
Spanish emigrants. The peak in arrivals before 1939 happened dur-
ing the 1920s (Figure 2). As in the rest of the Americas, traditional
Spanish immigration to Mexico had virtually stopped after 1930,
when the number of Spaniards entering the country declined sub-
stantially. It remained low until the ﬁrst wave of Republican exiles
arrived right after the beginning of the Spanish Civil War in 1936,
followed by still small but increasing numbers in 1937 and 1938. Not
until 1939 did the highest number of exiles, both in absolute and
relative terms, arrive in Mexico. After the civil war ended, the ships
Mexique, Sinaia, and Ipanema brought 4,660 expatriates. Exiles con-
tinued to arrive in signiﬁcant numbers during the three following
years before declining precipitously in 1943; in 1947, another, much
smaller, wave of exiles entered Mexico. In relative terms, compared
to traditional migration, exiles represented the lion’s share of total
Spanish immigrants between 1939 and 1942. In 1944, the arrival
of conventional immigrants to Mexico resumed, returning to a tra-
ditional migration pattern well documented in the literature.18

18 Pla Brugat, Els exiliats Catalans; Lida and Pacheco Zamudio, “El perﬁl de una inmigra-
ción: 1821–1939,” in Lida (ed.), Una inmigración privilegiada: comerciantes, empresarios y profesio-
nales en México en los siglos XIX y XX (Madrid, 1994), 25–50.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 549

Fig. 2

Spanish Emigration of Traditional and Exiled Migrants to
Mexico, 1870–1950

SOURCE Computed with data from the Registro Nacional de Extranjeros.

Spanish economic immigrants in Mexico were distinct from
those going to Argentina, Uruguay, or Cuba. Lida and Pacheco
Zamudio deﬁned them as a “privileged” because of their success
as traders, entrepreneurs, and professionals. The average traditional
immigrant was a Roman Catholic (“José” being the typical male
name) who spoke only Spanish, was 168 cm tall, and lived in the
province of Asturias. Around his thirty-ninth birthday, José
decided to sail from Spain to the port of Veracruz in Mexico.
From there he moved to the interior of the country, Mexico’s
Distrito Federal (DF), where he lived and worked as a trader.
Maria, José’s female counterpart, also a thirty-nine-year-old Cath-
olic, spoke only Spanish, was 159 cm tall, and lived in Asturias
until she embarked on a ship from Spain to Veracruz and then
went to the Mexican capital where she became a housewife.19

Although he may have shared some traits with his Catholic
counterpart, the typical Republican exile lived instead in Barce-
lona where he worked in trade-related activities, spoke French,
and was 169 cm tall. At the age of thirty-eight, he was forced to

19 These results agree with those estimated by Lida and Pacheco Zamudio, “El perﬁl.” For
a detailed study of the question, see Lida, “Los españoles en el México independiente:
1821–1950. Un estado de la cuestión,” Historia Mexicana, LVI (2006), 613–650.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

550

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

leave Spain, and after passing through some other country, he too
disembarked at Veracruz, moving later to Mexico DF. His Roman
Catholic wife was three years younger, spoke only a Spanish dialect,
and was 159 cm tall. At the end of the Civil War, she followed her
husband from Barcelona to live with him as a housewife in Mexico’s
capital. Extracted from more than 26,000 individual records, these
proﬁles show some of the most important differences between the
two forms of migration. The ﬁrst one, voluntary emigration vs.
forced exile, is obvious. The second concerns their respective levels
of human capital; on average, almost 50 percent of the exiles spoke
at least one foreign language, whereas 7.5 percent of them spoke
two and 2.2 percent spoke three or more. Only around 11 percent
of the traditional immigrants spoke a foreign language, a surprisingly
high ﬁgure given Spain’s illiteracy rates of about 32 percent in 1930
and 23 percent in 1940.20

Table 2 presents the differences between traditional migrants
and exiles in several dimensions. Of the two columns for traditional
migrants, the ﬁrst uses information for the whole period, and the
second is conﬁned to the same entry years as the Republican exiles.
For those registered before 1929, the records include their occu-
pation in Mexico, whereas for those in the second column, the
occupations are the ones that they had in Spain. The most signif-
icant difference between both samples of traditional migrants lies
in the higher HISCAM score and the number of spoken foreign lan-
guages of those who entered Mexico between 1936 and 1950.
Traditional migrants in the later years were also younger when
they started and more apt to come from the primary and second-
ary sectors than were their predecessors. The proportion of men
was considerably smaller and that of minors larger. Exiles bore
similarities with the traditional migrants who moved with them,
although they spoke more foreign languages and had a greater
share of occupations in the secondary sector. The proportion of
exiles who declared themselves as Catholics was also much smaller
than that of traditional migrants in either of the two samples.21

20 Pla Brugat, Els exiliats Catalan, 171. The number of foreign languages spoken was com-
puted with data from the Registro Nacional de Extranjeros.
21 Between 1930 and 1948, a period of generally low emigration from Spain to Latin Amer-
ica, the importance of professionals emigrating increased, to the detriment of rural and even
industrial laborers. See César Yañez Gallardo, La emigración española a América (siglos XIX y
XX): Dimensión y características cuantitativas (Colombres, 1994), 183–200.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 551

Table 2 Descriptive Statistics of Indicators, Traditional Migrants vs. Exiles

TRADITIONAL MIGRANTS

INDICATOR

HISCAM
Foreign languages
Height
HISCLASS
Age
Primary sector
Secondary sector
Tertiary sector
Share men
Share minors
Share Catholics

ALL

60.50
0.15
165.70
2.20
38.80
8.30
7.60
84.10
0.74
3.13
96.10

1936–1950

64.20
0.29
165.50
2.30
36.80
13.60
12.60
73.80
0.57
7.20
96.20

EXILES

64.90
0.55
164.80
2.50
36.70
9.70
19.40
70.90
0.58
5.10
63.60

SOURCE Computed with data from the Registro Nacional

Figure 3 presents the evolution of the average HISCAM of all
migrants, including the Republican exile, between 1886 and
1950. The trends show a steady decline in value between 1886
and the early 1920s, followed by an intense recovery until 1935
and high levels persisting until 1943 when the average HISCAM
score decreased but still remained at high historical levels. Long-
term trends were similar for both genders, though average female
HISCAM scores were lower in general. A quick look at the average
number of foreign languages spoken by all migrants shows relative
stability between 1886 and the early 1920s when, as in the case of
HISCAM, the series witnessed an intense increase in 1936 coinciding
with the arrival of the ﬁrst Republican exiles (Figure 4). This
increase peaked in 1942, only to reverse quickly when the number
of exiles decreased. It was high in the late 1940s, although its values
resemble those expected if the period from 1922 to 1934, which
represents the traditional migrants’ trend, is extrapolated. As in the
case of HISCAM, long-term trends were similar in both genders, also
showing a high correlation in long-term changes.22

The evolution of average height also shows a general
improvement in the long term, although short-term changes and
minimum and maximum levels differed from those presented by
HISCAM and languages (Figure 5). The lowest values occurred at the

22 Women recorded as housewives are not part of the estimate, since this status is not
included on the scale.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

552
Fig. 3 Average HISCAM of Migrants, 1886–1950 (Five-Year Moving

Average)

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SOURCE Computed with data from the Registro Nacional de Extranjeros.

beginning of the 1880s period and the maximum levels in the
1940s. The short-term changes were a deterioration between the
ﬁnal years of the nineteenth century and the early 1910s and then a
steady increase that continued until the end of the period for males
and until the early 1940s for females. In the case of height, espe-
cially with regard to women, a short-term acceleration of the pre-
vious trends coincided with the arrival of the ﬁrst exiles in 1936. It
continued to rise until 1942 when their numbers dwindled. Our
descriptive analysis of the data suggests that the Republican exiles
shared some similarities with traditional migrants, especially with
those who moved in the same year, despite some stark differences.

ECONOMETRIC ANALYSIS To highlight the effect of the exile, we
carried out an econometric analysis comparing, at individual level,
the potential determinants of the differences in human capital

SPA IN ’S LOS S OF H UM AN CA PITAL
Fig. 4 Foreign Languages per Migrant, 1886–1950 (Five-Year Moving

| 553

Average)

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SOURCE Computed with data from the Registro Nacional de Extranjeros.

between refugees and traditional migrants, using ordinary least
squares (OLS) pooled regressions with the following speciﬁcations:

Human Capitali

¼ ∝ þ β

1 Exilei þ Di þ Pi þ εi

Human Capitali

¼ ∝ þ β

1 Exilei þ β

2Malei þ Di þ Pi þ εi

Human Capitali

¼ ∝ þ β

1 Exilei þ β
þ Di þ Pi þ εi;

2Malei þ β

2HISCLASSi

(1)

(2)

(3)

where Human Capitali represents each one of the three proxies
(HISCAM, foreign languages, and height) for individual i and Exile
represents a dummy variable that takes value 1 if a migrant claims
to be an exile from the Spanish Civil War. The existence of a gen-
der gap in the three human-capital dimensions, especially in the
case of height, when genetic differences are clear, implies that
we should also control for gender with the dummy variable Male,

554

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

Fig. 5 Average Migrant Height, 1886-1950 (Five-Year Moving

Average)

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SOURCE Computed with data from the Registro Nacional de Extranjeros.

which takes value 1 if the individual is male. We should also con-
sider that the average values of the three proxies of human capital
could change over a period as lengthy as the one in this article.
Hence, we introduce the factor variable D in the models to con-
trol for a migrant’s decade of birth. In a similar way, to take
account of possible signiﬁcant regional differences in the average
levels of human capital considered in our three dimensions over
time, we introduce the factor variable P to control for an indi-
vidual’s province of birth.

Finally, we include HISCLASS as a factor variable with ﬁve pos-
sible values, ranging from 1 (elites) to 5 (unskilled workers and
farm workers) as described above. This variable is not included
when HISCAM is used as proxy because of the endogeneity that
exists between both variables. We expect that this variable will
be highly correlated with our three dimensions of human capital;
with its inclusion, we attempt to ﬁnd signiﬁcant differences

SPA IN ’S LOS S OF H UM AN CA PITAL

| 555

between exiles and traditional migrants who belong to the same
social class. In other words, our objective is to estimate whether
exiles were not just at the top of the human-capital hierarchy
within the whole sample but also within their respective social
classes.23

Tables 3 to 5 present the four speciﬁcations for the three
dimensions of human capital—HISCAM, foreign languages, and
height. For HISCAM, our estimates show that when gender, decade,
and province of birth are controlled, individuals who declared
themselves as exiled have a HISCAM value 4 points higher than
do traditional migrants. Males also present a higher HISCAM than
females. In Model III, which restricts the sample to those migrants
who entered Mexico from 1929 and to their occupation in Spain,
the value of the exile dummy is lower, although still signiﬁcant. As
reference, with a HISCAM around 60 (average of the sample), are
occupations like bread maker, singer, or butcher; an improvement
of 4 points would putatively reach occupations like telephone or
telegraph operator, ofﬁce clerk, or lithographer.24

Table 4 presents the same three models using foreign lan-
guages as human-capital proxy. The results are like those for HIS-
CAM; the exile dummy presents values that are stable in all the
speciﬁcations. On average, exiles speak around 0.3 more foreign
languages per head than do traditional migrants, a value that
increased once social classes are introduced as controls. Women
suffer a penalty even when social classes are considered, where
as expected, the gradient from the top two classes to the rest is
pronounced.

By contrast, the analysis using height, presented in Table 5,
shows that without gender controls, exiles were 0.7 cm shorter
than were traditional migrants, an outcome that changes once
gender is introduced. These results are explained by the share of
women in the exile group being higher than in the sample of tra-
ditional migrants and by the genetic gender gap in statures in
Models II–III, males being between 9 and 10 cm taller than
females. Contrary to the case of HISCAM, exiles did not seem to

23 For reference, the models including HISCAM and HISCLASS are presented in Table A2 of the
appendix.
24 More examples of occupations in HISCAM can be found in Table A1 in the online
appendix.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

556

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

Table 3 Correlates of HISCAM

DEP. VARIABLE: HISCAM

MODEL I

MODEL II

MODEL III

Exiled
Male
Decade of birth control
Province of origin control
Constant

Obs.
R2
F-test

3.6*** (0.33)

Yes
Yes

61.00

16,630
0.05
0.00

4.0*** (0.34)
3.4**** (0.43)
Yes
Yes

0.9** (0.38)
3.9*** (0.49)
Yes
Yes

57.60

16,630
0.06
0.00

67.60

6,805
0.07
0.00

*Signiﬁcance at 10 percent level.
**Signiﬁcance at 5 percent level.
***Signiﬁcance at 1 percent level.
NOTE Robust standard errors in parentheses.

enjoy higher statures than traditional migrants in any of the spec-
iﬁcations. Individuals in class 1 were the tallest, although the gra-
dient from the lowest social class (class 5) to the highest is not as
clear as in the case of HISCAM.

Table 4 Correlates of Languages

DEP. VARIABLE: LANGUAGES

MODEL I

MODEL III

Exiled
Male
HISCLASS (CLASS 1—ELITE BASELINE)

0.29*** (0.01) 0.30*** (0.01)

0.13*** (0.007)

Class 2—lower middle
Class 3—self-employed
farmers and ﬁshermen
Class 4—skilled workers
Class 5—unskilled workers

0.33*** (0.02)
0.07*** (0.02)

−0.32*** (0.02)
−0.46*** (0.02)

−0.43*** (0.02)
−0.42*** (0.03)

and farm workers

Decade of birth control
Province of origin control
Constant

Obs.
R2
F-test

Yes
Yes
0.33

21,727
0.13
0.00

Yes
Yes
0.33

21,727
0.15
0.00

Yes
Yes
1.17

15,902
0.20
0.00

*Signiﬁcance at 10 percent level.
**Signiﬁcance at 5 percent level.
***Signiﬁcance at 1 percent level.
NOTE Robust standard errors in parentheses.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 557

Table 5 Correlates of Heights

DEP. VARIABLE: HEIGHT (CM)

MODEL I

MODEL II

MODEL III

Exiled
Male
HISCLASS (CLASS 1—ELITE BASELINE)

−0.7*** (0.15)

Class 2—lower middle
Class 3—self-employed
farmers and ﬁshermen
Class 4—skilled workers
Class 5—unskilled workers

0.04 (0.16)
8.9*** (0.10)

0.08 (0.16)
10.0*** (0.22)

−1.4*** (0.18)
−1.4*** (0.23)

−1.7*** (0.22)
−1.3*** (0.46)

and farm workers

Decade of birth control
Province of origin control
Constant

Yes
Yes

162.2

155.3

158.0

Obs.
R2
F-test

20,505
0.03
0.00

20,505
0.30
0.00

15,236
0.15
0.00

*Signiﬁcance at 10 percent level.
**Signiﬁcance at 5 percent level.
***Signiﬁcance at 1 percent level.
NOTE Robust standard errors in parentheses.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

DISCUSSION Our results show that HISCAM and foreign languages
as proxies for human capital both reached maximum historical
levels with the arrival of the Republican exile in Mexico, though
with important differences. The increase in HISCAM started in the
early 1920s, well before the arrival of exiles, whereas the exponen-
tial growth in spoken foreign languages clearly correlates with the
arrival of the ﬁrst political refugees. Height, however, shows a
deterioration from the late nineteenth to the early twentieth cen-
tury, recovering later, with no signiﬁcant changes during the
arrival of the exiles.

The econometric analysis of the three proxies for human cap-
ital supports this descriptive analysis. The ﬁrst main conclusion
combining both approaches is that exile status does not seem to
correlate with taller stature. The possible reasons for this result
are not necessarily incompatible with those obtained for the other
two dimensions of human capital. First, although height might be
expected to correlate with human capital, it is a more tangential
proxy than are HISCAM and foreign languages. Second, congruent

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

558
with the ﬁrst reason, being tall requires less of an investment than
does achieving an elevated social position or learning a foreign lan-
guage. In other words, an average family that could afford to
nourish their children and regulate their energy to promote their
growth would not necessarily have been able to afford the educa-
tion required to guarantee them a good occupation or a facility
with foreign languages. The signiﬁcance of the exile dummy for
HISCAM and foreign languages when social classes are considered
suggests as much; the exile dummy is signiﬁcant for HISCAM and
foreign languages because exiles were at the top of their respective
social classes and so more able to invest in education.

As Lida and Pacheco Zamudio explain, traditional migrants
were considerably taller than the average Spaniard who did not
migrate. A recent study by Cámara et al. estimates that the average
height of Spanish conscripts was 162 cm for the cohorts born in
1840 and 165 cm for those born in 1920; the ﬁgures for the same
cohorts of Spanish migrants to Mexico were 166 and 169 cm. A
comparison with the heights of the elites in Brazil and Colombia
reinforces the tall statures of the Spanish migrants to Mexico. Elite
males reached heights like those of their Colombian and Brazilian
counterparts in the early twentieth century (around 168 cm).
Female Spanish migrants in Mexico were, on average, 1 cm taller
than elite Colombian women, around 159 vs. 158 cm.25

The results in Tables 3 and 4 indicate that exiles presented
higher values for both HISCAM and languages than did traditional
migrants. Exiles had an average HISCAM around 4 points higher
than did traditional emigrants at the same time, who averaged a
HISCAM of 66. The gap between both groups regarding languages
was more substantial; exiles spoke around 0.3 more languages per
head than did traditional migrants (0.34 languages per head) who
moved to Mexico at the same time. Why was the connection of
exile status with foreign languages much stronger than that with
HISCAM? Many of the exiles who claimed to speak French might

25 Lida and Pacheco Zamudio, “El perﬁl”; Cámara et al., “Height and Inequality.” For the
heights of Latin-American elites, see Adolfo Meisel and Margarita Vega, “The Biological
Standard of Living (and Its Convergence) in Colombia, 1870–2003: A Tropical Success
Story,” Economics & Human Biology, V (2007), 100–122; Daniel Franken, “Anthropometric
History of Brazil, 1850–1950: Insights from Military and Passport Records,” Revista de Historia
Económica / Journal of Iberian and Latin American Economic History, XXXVII (2019), 377–408.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 559

have learned it not from formal education but from a long stay in
France before sailing to Mexico. Many Spanish exiles moved to
southern France during the war hoping to return after the conﬂict.
When the war ended with the Republican defeat, a considerable
number of them sailed to Mexico. Although the notion that they
learned to speak French while living in France is pure speculation,
Table 6 repeats the most complete model with four new speciﬁ-
cations and several robustness checks of our results.

Model V’s dependent variable is the number of foreign lan-
guages other than French that an individual claimed to speak.
Model VI excludes from the sample all exiles who arrived from
Catalonia or Aragon, since the bulk of exiles who escaped to
France and potentially traveled later to Mexico came from these
two regions bordering France. Model VII excludes all the exiles
who arrived on the ships that the Republican government in exile
hired to transport some of the refugees living in France. Finally,
Model VIII is the most restrictive, excluding all exiles who
claimed that French was their only language—clearly a lower
bound, since the model purges a large number of exiles regardless
of where they learned French, the most common language spoken
in the sample. The results in all the models are consistent with
those obtained in the previous models, suggesting that the exile
dummy in foreign languages is not a consequence of a stay in
France. The values of the coefﬁcients for Models V and VIII
are, as expected, lower. In the case of Model V, since the average
number of languages other than French was just 0.13, the relative
impact of the exile dummy was also relevant. In the case of Model
VIII, the average number of languages spoken in the reduced sam-
ple was 0.18, also suggesting a relative impact of the exile dummy
similar to that in the original models.26

Quantitative analysis of the exile as it relates to women is
nonexistent, although qualitative research about the exiled female
Republicans is available. Our data, however, permit a quantitative
analysis of the human capital of Spanish women in the exile. As
mentioned before, the female occupational category of housewife
is not conducive to any conclusions about their real qualiﬁcations.

26 The ships sent by the Republican government in exile were Flandre, Guinea, Ipanema,
Manuel Arnus, Mexique, Nyassa, Orinoco, Santo Domingo, Siboney, and Sinai.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

I
I
I

L
E
D
O
M

I
I

L
E
D
O
M

S
E
G
A
U
G
N
A
L

:
E
L
B
A
R
A
V

.
P
E
D

e
c
n
a
r
F

s
e
l
i
x
E

r
o
f

d
e
t
s
u
j
d
A

s
e
g
a
u
g
n
a
L

f
o

s
e
t
a
l
e
r
r
o
C

e
l
b
a
T

)
2
0
.
0
(

*
*
*
6
1
.
0

)
2
0
.
0
(

*
*
*
1
.
0

)
2
0
.
0
(

3
3
0
.
0
(

*
*
*
4
3
.
0
−

*
*
*
5
4
.
0
−

)
3
0
.
0
(

*
*
*
5
4
.
0
−

*
*
*
3
4
.
0
−

s
e
Y

7
1
.
1

7
6
9
,
4
1

3
1
.
0

0
0
.
0

)
2
0
.
0
(

*
*
*
7
3
.
0

*
*
*
1
1
.
0

)
2
0
.
0
(

*
*
*
9
2
.
0
−

*
*
*
6
3
.
0
−

)
2
0
.
0
(

)
3
0
.
0
(

*
*
*
8
3
.
0
−

s
e
Y

1
1
.
1

6
9
8
,
4
1

9
1
.
0

0
0
.
0

)
2
0
.
0
(

*
*
*
3
3
.
0

*
*
*
5
0
.
0

)
2
0
.
0
(

*
*
*
0
3
.
0
−

*
*
*
1
4
.
0
−

)
2
0
.
0
(

)
3
0
.
0
(

*
*
*
8
3
.
0
−

s
e
Y

2
7
.
1

4
4
1
,
4
1

8
1
.
0

0
0
.
0

)
2
0
.
0
(

*
*
*
9
0
.
0

)
2
0
.
0
(

*
*
8
0
.
0

)
2
0
.
0
(

*
*
*
2
2
.
0
−

*
*
*
9
2
.
0
−

)
2
0
.
0
(

)
3
0
.
0
(

*
*
*
0
3
.
0
−

*
*
*
9
2
.
0
−

s
e
Y

0
1
.
1

2
0
9
,
5
1

7
0
.
0

0
0
.
0

)
E
N
I
L
E
S
A
B

E
T
I
L
E

e
l
d
d
i
m

—

r
e
w
o
—

S
S
A
L
C
(

d
e
y
o
p
m
e
–
f
l
e
s

n
e
m
r
e
h
s
ﬁ

—

s
s
a
l
C

d
n
a

s
r
e
m

r
a
f

d
e
l
i
x
E

e
l
a

S
S
A
L
C
S
I
H

s
r
e
k
r
o
w
d
e
l
l
i
k
s
n
u
—

s
r
e
k
r
o
w
d
e
l
l
i

k
s
—

s
s
a
l
C

s
r
e
k
r
o
w
m

r
a
f

d
n
a

o
r
t
n
o
c

n
i
g
i
r
o

f
o

e
c
n
i
v
o
r
P

t
n
a
t
s
n
o
C

o
r
t
n
o
c

h
t
r
i
b

f
o

e
d
a
c
e
D

.
s
b
O

2
R

t
s
e
t
–
F

.
l
e
v
e
l

t
n
e
c
r
e
p

t
a

.
l
e
v
e
l

t
n
e
c
r
e
p

0
1

t
a

.
l
e
v
e
l

t
n
e
c
r
e
p

t
a

e
c
n
a
c
ﬁ
n
g
i
S
*
*

e
c
n
a
c
ﬁ
n
g
i
S
*
*
*

e
c
n
a
c
ﬁ
n
g
i
S
*

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

.
s
e
s
e
h
t
n
e
r
a
p

s
r
o
r
r
e

d
r
a
d
n
a
t
s

t
s
u
b
o
R

E
T
O
N

SPA IN ’S LOS S OF H UM AN CA PITAL

| 561

Furthermore, occupations such as seamstress, common for immi-
grant women, may have been only temporary occupations held by
exiled women during their early years in Mexico, not correspond-
ing to their pre-exile activity. The resort to languages spoken
other than Spanish helps to overcome these obstacles in assessing
women’s human capital. Our results show that women were cru-
cial in the loss of human capital in the Mexican ﬂow; according to
our estimations, more than one-third of all the foreign languages
spoken by the exiles were spoken by women, representing a large
share of human capital.27

Women represented 42 percent of the Republican exiles to
Mexico. The difference between their share of population and
their share of human capital reﬂects the relative social standing of
men and women. Yet even though they attained lower human
capital levels than men did, women comprised a substantial per-
centage of the total brain trust. The Mexican exile highlights the
long-neglected importance of women. Figures 3–5 reveal the rel-
evance of this issue, revealing parallel trends in the evolution of
male and female migrants’ proﬁles over time. The synchronicity
of the long-term waves in the three human-capital proxies for both
genders provides new and key evidence for the study of the effects
of the brain drain in the age of mass migrations, usually focused on
men due to limitations imposed by the historical sources.

The study of the Spanish Civil War and its consequences has gen-
erated intense and controversial academic debates, including the
role of the Republican exiles and the loss of human capital that
they represented. Refugees are assumed not to be motivated by
economic factors to the same degree as other migrants are; for
them, push factors at home are stronger than economic pull factors
abroad. The Spanish exile to Mexico after the Civil War is a good
case study of a highly selected migratory ﬂow by people reputed to
possess skills above the average of the Mexican population. The
literature argues that the importance of the Spanish exile does
not rely so much on the number of people who left the country
as on their characteristics. The traditional view presents the exile as

27 For an example of the study of female exiles, see Josebe Martinez, Exiliadas: Escritoras,
guerra civil y memoria (Barcelona, 2007).

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

562

| S Á N C H E Z – A LO N S O A N D S ANTIAGO-CABA LLERO

a loss of human capital and a brain drain for Spain. This article
quantiﬁes the loss and places it in a larger context, venturing
beyond the traditional view of the Spanish exile focused almost
exclusively on intellectuals, writers, and highly qualiﬁed profes-
sionals. Overall, the Spanish refugees in Mexico represented a res-
ervoir of considerable human capital compared with the traditional
population of economic immigrants.

We created a proﬁle of the immigrants in Mexico before 1936
and a proﬁle of the Republican exiles to compare traditional eco-
nomic immigrants with refugees. To quantify the human capital of
the two groups, we use three different indicators—HISCAM for
occupations, number of foreign languages spoken, and stature.
Our data allow us to create an index of human capital based pri-
marily on the average number of foreign languages spoken that
captures the highest level of human capital for Spanish migrants
during the period considered. When using both HISCAM and for-
eign languages as proxies of human capital, exiles presented higher
values for both variables than did traditional migrants, though exile
status did not correlate with taller height. In fact, traditional immi-
grants in Mexico were considerably taller than the average Span-
iard who did not migrate.

Given languages as a marker for the highest levels of accom-
plishment, followed by HISCAM, Spanish exiles indeed appear to
have been highly qualiﬁed. Our data show that, on average, almost
50 percent of the exiles spoke at least one foreign language, com-
pared to the 11 percent or so of traditional immigrants. The high
selectivity of the exiles is also reﬂected in the signiﬁcance of the
exile dummy when social-class controls were introduced in our
models, suggesting that exiles were among the most skilled mem-
bers of each of the classes.

The ﬁndings herein regarding women are pioneering; this
article’s proxies measure human capital for women much better
than do occupations. Heretofore, the economic-history literature
has largely omitted any attempt to assess women’s human capital
because of a lack of recorded data about their occupations.
Depending on the proxy used, around one-quarter of the human
capital lost with the Mexican expatriation belonged to women. An
important lesson of this result highlights the importance of quan-
tifying the relevance of a group that is usually invisible to eco-
nomic history.

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

SPA IN ’S LOS S OF H UM AN CA PITAL

| 563

APPENDIX: HISCAM SCORES AND CORRELATES

Table A1 Examples of Occupations with HISCAM Scores around 60 and 64

OCCUPATIONS WITH HISCAM AROUND 60

58240 Private police guard
36000 Transport conductors
77620 Bread baker
17300 Actors and stage directors
18000 Athletes, sportsmen, and related workers
17145 Singer
75990 Other spinners, weavers, knitters, dyers, and related workers
77310 Butcher, general

OCCUPATIONS WITH HISCAM AROUND 64
38000 Telephone and telegraph operators
84175 Machinery erector and installer
92400 Printing engravers (except photo engravers)
55120 Concierge (apartment house)
92415 Lithographer
39310 Ofﬁce clerk, general
41000 Working proprietors (wholesale and retail trade)
21340 Sales manager (retail trade)

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

0
5
9
1
–
9
2
9
1

Y
R
T
N
E

E
L
P
M
A
S

L
L
U
F

L
E
D
O
M

I
I
I

L
E
D
O
M

I
I

L
E
D
O
M

M
A
C
S
I
H

:
E
L
B
A
R
A
V

.
P
E
D

M
A
C
S
I
H

f
o

s
e
t
a
l
e
r
r
o
C

2
A

e
l
b
a
T

)
5
3
.
0
(

)
4
3
.
0
(

*
*
*
6
.
5
2
−

*
*
*
7
.
5
3
−

)
6
3
.
0
(

)
1
5
.
0
(

*
*
*
2
.
3
3
−

*
*
*
5
.
1
4
−

)
6
2
.
0
(

)
5
2
.
0
(

*
*
*
3
.
8
2
−

*
*
*
3
.
6
3
−

)
8
2
.
0
(

)
5
3
.
0
(

*
*
*
7
.
3
3
−

*
*
*
1
.
2
4
−

)
8
2
.
0
(

*
*
*
2
.
1

)
1
2
.
0
(

2
0
.
0

)
9
4
.
0
(

*
*
*
9
.
3

)
8
3
.
0
(

*
*
9
.
0

)
0
2
.
0
(

)
5
2
.
0
(

*
*
*
2
.
1

*
*
*
8
.
0

)
3
4
.
0
(

*
*
*
*
4
.
3

)
4
3
.
0
(

*
*
*
0
.
4

)
3
3
.
0
(

*
*
*
6
.
3

s
e
Y

1
.
8
8

2
0
8
,
6

0
7
.
0

0
0
.
0

s
e
Y

5
0
8
,
6

7
0
.
0

0
0
.
0

6
.
7
6

s
e
Y

1
.
5
8

0
3
6
,
6
1

2
6
.
0

0
0
.
0

s
e
Y

0
3
6
,
6
1

6
0
.
0

0
0
.
0

6
.
7
5

s
e
Y

0
3
6
,
6
1

5
0
.
0

0
0
.
0

0
.
1
6

s
r
e
m

r
a
f

d
e
y
o
p
m
e
–
f
l
e
s

)
E
N
I
L
E
S
A
B

E
T
I
L
E

e
l
d
d
i
m

—

r
e
w
o
—

S
S
A
L
C
(

—

d
n
a

s
r
e
k
r
o
w
d
e
l
l
i

n
e
m
r
e
h
s
ﬁ

d
n
a

k
s
n
u
—

k
s
—

s
s
a
l
C

d
e
l
i
x
E

e
l
a

S
S
A
L
C
S
I
H

s
s
a
l
C

o
r
t
n
o
c

n
i
g
i
r
o

f
o

e
c
n
i
v
o
r
P

t
n
a
t
s
n
o
C

o
r
t
n
o
c

h
t
r
i
b

f
o

e
d
a
c
e
D

s
r
e
k
r
o
w
m

r
a
f

.
s
b
O

2
R

t
s
e
t
–
F

.
l
e
v
e
l

t
n
e
c
r
e
p

t
a

.
l
e
v
e
l

t
n
e
c
r
e
p

0
1

t
a

.
l
e
v
e
l

t
n
e
c
r
e
p

t
a

e
c
n
a
c
ﬁ
n
g
i
S
*
*

e
c
n
a
c
ﬁ
n
g
i
S
*
*
*

e
c
n
a
c
ﬁ
n
g
i
S
*

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

:
/
/

d
i
r
e
c
t
.

i
t
.

e
d
u

/
j
i

n
h
a
r
t
i
c
e
–
p
d

f
/

5
2
4
5
3
7
2
0
6
5
2
8
9

/
j
i

n
h
_
a
_
0
1
7
6
6
p
d

b
y
g
u
e
s
t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

.
s
e
s
e
h
t
n
e
r
a
p

s
r
o
r
r
e

d
r
a
d
n
a
t
s

t
s
u
b
o
R

E
T
O
N Journal of Interdisciplinary History, LII:4 (Spring, 2022), 537–564. image

Download pdf