Policy Research Working Paper                         9912




 Long-Term Effects of the 1923 Mass Refugee
    Inflow on Social Cohesion in Greece
                                      Elie Murard




Social Sustainability and Inclusion Global Practice
January 2022
Policy Research Working Paper 9912


  Abstract
 After the 1919–1922 Greco-Turkish conflict, 1.2 million                            exhibit higher political and civic participation. At the com-
 Greek Orthodox were forcibly displaced from Turkey to                              munity level, places with a higher share of refugees in 1928
 Greece, increasing the host population by 20 percent within                        are more likely to have at least one sport association 80 years
 a few months. Refugees were pro-vided with farmland, new                           later. There is no impact on political fragmentation nor on
 houses and schools, and were granted the Greek citizenship.                        crime. The historical refugees’ integration starkly contrasts
 This paper analyses the long-term social integration of refu-                      with the social marginalization of recent Albanian immi-
 gees and the effect of their resettlement on social cohesion.                      grants who, unlike the former, neither spoke Greek nor had
 Combining historical and modern population censuses and                            the same religion as locals upon arrival. These results suggest
 surveys, this paper finds that, by the 2000s, refugees display                     that early investments in inclusion policies can be effective
 a high rate of intermarriage with Greek natives, report levels                     at fostering refugees’ assimilation, at least when newcomers
 of trust in others and in institutions similar to natives, and                     and locals have similar cultural profiles.




 This paper is a product of the Social Sustainability and Inclusion Global Practice. It is part of a larger effort by the World
 Bank to provide open access to its research and make a contribution to development policy discussions around the world.
 Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The author may be
 contacted at elie.murard@ua.es.




         The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
         issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
         names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
         of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
         its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                       Produced by the Research Support Team
              Long-term effects of the 1923 mass refugee inflow
                                     on social cohesion in Greece ∗
                                                          Elie Murard †

                                         University of Alicante, NOVA SBE, IZA, LEAP




          JEL Classification Codes: O15, J62 ,N34, N44
         Keywords: Refugees, social cohesion, integration, historical persistence, immigration




   ∗
       This paper was commissioned by the World Bank Social Sustainability and Inclusion Global Practice as part of the activity “Pre-
venting Social Conflict and Promoting Social Cohesion in Forced Displacement Contexts.” The activity is task managed by Audrey Sacks
and Susan Wong with assistance from Stephen Winkler. This work is part of the program “Building the Evidence on Protracted Forced
Displacement: A Multi-Stakeholder Partnership”. The program is funded by UK aid from the United Kingdom’s Foreign, Common-
wealth and Development Office (FCDO), it is managed by the World Bank Group (WBG) and was established in partnership with the
United Nations High Commissioner for Refugees (UNHCR). The scope of the program is to expand the global knowledge on forced
displacement by funding quality research and disseminating results for the use of practitioners and policy makers. This work does not
necessarily reflect the views of FCDO, the WBG or UNHCR.
   †
     I thank Sara Tonini for her very helpful suggestions and Chrysovalantis Vasilakis for sharing the crime data.
1    Introduction
Mass displacement of people is common throughout history. The 20th century alone experienced
the uprooting of millions of Jews, Hindus, and countless others. More recently, civil wars and con-
flicts displaced millions of Sudanese, Syrians, and Rohingyas. While significant attention has been
given to the short-term costs that mass refugee inflows may cause for hosting countries in the short-
term, less is known about the long-term legacy of such events. In particular, little evidence exists
on how large-scale population resettlements affect the social cohesion of receiving communities in
the long-run.1 This question is important because the short-term and long-term effects could be
very different.
    This paper aims to fill this gap by exploiting the 1923 Greek population resettlement to examine
the long-run consequences of mass refugee inflow on the social cohesion of the receiving society. In
the aftermath of the Greco-Turkish war of 1919–1922, escaping religious persecutions, the Greek
Orthodox population of the Ottoman Empire fled from their homeland in Turkey they inhabited
since antiquity. In total, 1.2 million Greek Orthodox were forcibly resettled to Greece, increasing
the host population by more than 20 percent within a few months. Confronted with such a sudden
influx of refugees, Greece – an agrarian country economically drained by several years of war
– called for the help of the League of Nations, which not only provided immediate relief (food
and shelter) but also implemented an ambitious resettlement program. Former marshlands were
reclaimed and refugees were provided with arable land, livestock and other farm inputs. New
houses were build and new schools opened to accommodate the newcomers. Furthermore, given
that refugees could not return to their homeland and had to remain in Greece permanently, they
were granted the Greek citizenship.          2

    Despite these substantial investments, refugees – initially in a state of “utter destitution”3 –
were often met with hostility by the locals, who felt their way of life under threat. Serious clashes
occurred over the distribution of farmland, which mostly benefited refugees at the disadvantage of
local cultivators.4 Furthermore, although refugees and natives shared the same religion and often
the same language, they were often met with prejudice and rejection, especially as “competition for

    1
      For a review of the literature see Becker and Ferrara (2019); Verme and Schuettler (2019); Ruiz and Vargas-Silva
(2013).
    2
      Some of the terminology (“refugees”, “resettlement”) used in this paper does not align with contemporary legal
definitions or customary uses in international refugee protection regimes. By “refugees”, I designate Greek Orthodox
population who fled Turkey due to religious persecutions, even if they cannot be considered as refugees under the 1951
Genena Convention as they were already Greek nationals upon they entry on Greek territory. Likewise, by “resettlement”,
I mean the displacement of population from an origin to a receiving country (and not the transfer of refugees from an
asylum country to another State).
    3
      Kontogiorgi (2006), p. 88.
    4
      Large tracts of arable land were also left vacant by the forced expulsion of all Muslim inhabitants of Greece (about
400,000 people), who resettled in Turkey 1928. The League of Nations split the large estates of the expelled Muslims
into smallholdings and distributed them to the Greek refugees, while virtually none were redistributed to the native
peasantry.


                                                            1
land and livelihood lowered the standard of living for everyone”.5 The refugees’ distinct identity
was further cemented by clear allegiances to opposing political parties during the interwar period,
with refugees actively supporting the liberals and natives standing behind the conservatives.

   Has the refugees’ initial economic hardship upon arrival turned into persistent social marginal-
ization? Has the historical strife between refugees and natives cemented into deep-seated mistrust
and oppositional identities? Has social cohesion – understood as “a sense of shared purpose and
trust [...] and willingness to engage and cooperate” among members of a same society (Kim, Sheely
and Schmidt, 2020) – been eroded by major inter-group cleavages? Or, to the contrary, have the
League of Nations’ resettlement efforts proved sufficient to prevent refugee’s exclusion and foster
their integration into the Greek society ?
   To address these questions, I use Murard and Sakalli (2018)’s geocoded dataset of refugee set-
tlements in 1928, which I combine with several indicators of social cohesion at the local level in
the 2000s and 2010s. I also use various Greek population censuses from 1928 to 2001 to exam-
ine refugees’ social integration, and the European Social Survey to look at the second-generation
refugees’ political values, trust, and civic engagement.
   The first important result is that refugees experienced a successful social integration in the
host society. By 2001 refugees display a high rate of intermarriage, with close to 50 percent of
refugee women being married with a Greek native. Second generation refugees also report levels
of trust in others and in institutions similar to natives, do not perceive to be discriminated against,
and are more likely than natives to engage in political activism and to participate in voluntary
associations. I also find no persistent inter-group inequalities between refugees and natives in
terms of educational attainment or occupation. Furthermore, the civic engagement of refugees
appears to have contributed to higher involvement in voluntary associations at the community
level. I find that, among small communities (bellow 1,100 inhabitants), places with a higher share
of refugees in 1928 are more likely to have at least one sport association in the early 2000s. I
find no evidence that refugees increased political fragmentation – which is consistent with the fact
that voting patterns of refugees and natives are very similar – nor that refugees increased crime in
places of resettlement.

   These results have important policy implications. They suggest that early investments in re-
settlement efforts (e.g., provision of farmland) and inclusion policy (e.g., citizenship) can have
significant returns in the long-run by fostering refugees’s assimilation and sustaining social cohe-
sion. Under the threat of profound social disruption caused by the influx of 1.2 millions destitute
refugees, the Greek state sought to promote national unity by including refugees in the national
community. This was accomplished not only by providing refugees with the economic tools for their

   5
       Kontogiorgi (2006), p. 166, and Kritikos (2005)


                                                         2
integration (i.e., the material and productive assets lost during displacement), but also by granting
them the same civil and political rights as natives. Furthermore, new schools were opened to teach
refugee children the same education and civic values taught to Greek natives in order to promote a
cohesive national identity.6 Consistent with this historical account, I find evidence that the supply
of new schools played an important role in the refugee children’s progress in literacy and in the
catching up with natives in the interwar period.
    These results may yet be of limited external validity for current inflows of refugees with totally
different ethnic and cultural backgrounds compared to the local population. The successful inte-
gration of the 1923 refugees stands in stark contrast with the social marginalization of more recent
Albanian immigrants who, unlike past refugees, neither spoke Greek nor had the same religion as
locals upon arrival (coming from a Muslim-majority country). Albanian immigrants have not expe-
rienced the same social assimilation as refugees, neither in terms of education, intermarriage, nor
civic participation.7 The discrepancy in the socio-economic trajectories of the two groups of new-
comers highlights the specific challenges of providing integration policies for migrants who have
profoundly different linguistic, religious and cultural profiles relative to those of the locals.

    The remainder of the paper is organized as follows. Section 2 provides some elements of
historical context. Sections 3 reviews the related literature and presents the paper’s contribution.
Section 4 describes the data and empirical strategy. Section 5 presents the main results, while
section 6 concludes.


2       Context

2.1     The Exodus

Since it gained its independence from the Ottoman Empire in 1832, Greece pursued the goal to
unite the ethnic Greek population scattered across the Ottoman territories. After the First World
War, backed by the Allies, the Greek Army occupied territories in western Turkey that were heavily
populated by ethnic Greeks. However, while the Greek advance stalled as the military support of
the Allies faded, the Turkish Army’s counter-offensive at the end of August 1922 was successful.
The persecution of Greek Orthodox communities by the advancing Turkish Army, in particular the
fire in the Christian neighborhoods of Smyrna in September 13, 1922, triggered a mass exodus of
population to Greek islands and mainland. Almost one million of Greek Orthodox, whose ancestors
have been living in Asia Minor for millennia, left Turkey within a few months’ time.
    6
     Kontogiorgi (2006), p. 241.
    7
     As I cannot provide evidence on the second generation of Albanian immigrants, I examine the integration outcomes
of first generation migrants who arrived in Greece at different age in order to measure intergenerational trends. Evidence
from both the 2001 census and the ESS suggest that Albanian immigrants have not yet experienced any progress in
integration outcomes between younger and older immigrants of the first generation.


                                                           3
    The Lausanne Peace Treaty, signed in July 1923, ended the Greco-Turkish War. As a part of
the treaty, Greece and Turkey signed the “Convention for the Exchange of Populations”, which
devised and implemented the principle of a compulsory exchange of population for the first time
in modern history. The convention dictated that all Greek Orthodox of Turkey and all Muslims of
Greece were to be denaturalized, leave the country they reside in, and acquire the citizenship of
their destination country. As a result of the population exchange, Greece, a country with less than
five million inhabitants, received more than one million refugees from Turkey. In return, about
400,000 Muslims left Greece


2.2     Resettlement of refugees

At the beginning of the twentieth century, Greece was primarily a rural agrarian country. In 1920,
the urbanization rate was only 23 percent, and more than half of the population was employed
in agriculture. The relief and resettlement of such a large-scale refugee inflow were far beyond
the limited resources of the Greek State, which had almost gone bankrupt after a long and costly
war. As a result, the League of Nations intervened and founded the Refugee Settlement Commis-
sion (RSC) in September 1923. The RSC had full authority over the distribution of funds and
resettlement of refugees.
Agricultural Resettlement. The mandate of the RSC was to establish refugees in agricultural
work and assist them to become self-supporting. To this end, the RSC provided them with land
suitable for cultivation and with various farm inputs, such as livestock, seeds, and some machinery
(Kontogiorgi, 2006, pp. 78, 86, 90). The RSC concentrated its operation in Northern Greece where
large tracts of arable land were left vacant by the expelled Muslim population. The RSC split the
large estates of Muslim landlords (chiftliks) into smallholdings and distributed them to the Greek
refugees.8
Infrastructure Works. The RSC also expanded the cultivable land area by undertaking important
and costly drainage and reclamation works in the valleys of the Northern provinces, often water-
logged and prone to malaria. As a result, the area of cultivated land doubled in Northern Greece
(Kontogiorgi, 2006, pp. 277–290) and the drained marshlands hosted many new settlements of
refugees.9 In addition to distributing land and houses to refugees, the RSC also initiated a public
utility program: it constructed roads, bridges, dispensaries, and schools (Pentzopoulos 1962, p.
108; Kontogiorgi 2006, p. 297). The decennial 1922–1932 total budgetary expenditures for the re-
lief, housing and settlement of refugees amounted to a little more than 10 millions British pounds,
with 9.1 millions spent for agricultural settlement and 1.3 millions for urban settlement (Kritikos

    8
      By contrast, the refugees who settled in urban centers — about 50 percent of them — were given no means of
providing for themselves, apart from accommodation (Kontogiorgi, 2006, p. 103).
    9
      In total, the Greek regions of Macedonia, Thrace, Epirus, and Thessaly hosted more than 90 percent of the refugees’
agricultural settlements (uthors’ calculation using data from the 1930 Statistical Yearbook of Greece)


                                                           4
(2005); Pentzopoulos (1962)). This amount represented between 3 and 4% of Greek GDP.10 For
the agricultural resettlement, this translates into an expenditure of about 1500 US current dollars
per rural refugee.
School supply The refugee inflow displaced more than 185,000 between 6 and 12, which repre-
sented a 20% increase in the number of school-age children living in Greece. Facing the high risk
of over-crowded schools, the Greek State and the RSC responded by vastly increasing the supply
of schools. Between 1921 and 1927, 1,466 new schools were opened in order to accommodate
the newcomer children, i.e, almost 8 new schools were opened for 1000 refugee children. These
new schools were mostly build in provinces who receive a very large inflow of refugees (> 20% of
their population), where the increase in school supply (per children) was twice as high as in other
provinces (see Table A.3 in the appendix).
National interest The important investment in the integration of the 1923 refugees was also made
possible not only out of legal considerations (the 1923 Convention implementation) but primar-
ily because it served national interests. First, it contributed to mitigate the devasting social and
financial impact of the military defeat in the Greco-Turkish War. Second, the refugee resettlement
was linked to national security problems in key border areas (Macedonia, Epirus, Thrace) in the
proximity of Turkey and Bulgaria. The refugee resettlement served the purpose to consolidate
these border areas by making them more ethnically homogenous ahead of a potential new round
of armed conflict with neighboring nation-states.


2.3    Tensions with natives upon arrival

Conflict Over Land Distribution. The transfer of Muslim farmland to refugees by the RSC created
tensions with the native population. For a long time, the native landless peasants had nurtured
expectations of becoming owners of the land abandoned by the Turks, which they had been cul-
tivating for generations under a crop-sharing tenure system during the Ottoman Empire. Serious
clashes with refugees occurred when almost all of the former Turkish estates were transferred to
refugee families and virtually none were redistributed to the native peasantry (Kritikos, 2005).
Cultural differences Before the exchange of population, all Orthodox Christian population of
Turkey were grouped by the Ottoman authorities in the same confessional community, the Rum
millet, under the spiritual jurisdiction of the Ecumenical Patriarch in Constantinople, also known
as the Greek Orthodox Church. Religious affiliation was the criterion for identity. The expul-
sion of the refugees in 1923 was based on their adherence to the Greek-Orthodox church. While
the refugees and the local Greeks undoubtedly shared the same religion and had the same eth-
nic bonds (traditions from the Byzantine civilisation) they constituted two different worlds. The
historical evolution of communities in Asia Minor (Turkey), their culture, and their outlook were
  10
   At the end of the1920s, the Greek gross domestic product was about 14 billion in 1990 US dollars according to
Maddison (2010), or about 320 millions in 1930 pounds.


                                                       5
generally different to the local people’s cultural patterns (way of life, customs and mores, dress,
homes), since each group had developed as a separate entity under radically different historical and
social conditions. Furthermore, there were clear and significant differences between the refugees
depending on the area from which they had come. Although the majority of refugees shared the
same language with local Greeks, refugees from the Black Sea region (Pontics), who made up a
large percentage of the refugees, could not communicate with local people very easily, and there
were numerous Turkish-speaking groups among the refugees as well. This is why when locals and
refugees first came into contact and realized how very different they were, they suffered what
Mavrogordatos (1983) describes as “a traumatic cultural shock”. The ethnic boundaries separating
refugees (Prosphyges) and their hosts (the “Old Greeks”, Palioelladites) were sufficient to cause
strife and prejudice, particularly as “competition for land and livelihood lowered the standard of
living for everyone” (Kontogiorgi, 2006, p. 166). Turkish-speakers and Pontics faced more difficul-
ties, rejection, and discrimination, from the local population than Greek-speaking refugees, since,
culturally speaking, they stood out more: They were often derided for their customs and traditions
and natives referred to them as “Turkish seed” (Tourkosporoi) and often “regarded the newcomers
as a sickly lot and looked upon them with contempt and suspicion” (Kontogiorgi, 2006, p. 160).
Political cleavage The refugees’ distinct cultural identity was further cemented by clear political
allegiance in the 1930s. The Liberal Party led by E. Venizelos quickly recognized the importance
of refugees’ votes and worked for the integration of the refugees into the Greek political system.
The refugees became an integral part of the Venizelist camp (Pentzopoulos, 1962, p. 186). Partly
due to their exclusion from native patron-client networks, the refugees rapidly developed their own
political machines, headed by refugee or pro-refugee politicians and enjoying access to the top Venizelist
leaders (...) The refugees justifiably saw the Venizelist Republic as “their” regime, and the Liberal Party
particular as “their” party (Mavrogordatos, 1983, p. 202). The Venizelists’ embrace of the refugees
caused a conservative backlash against the newcomers, with anti-venizelists accusing the refugees
of depriving the native majority of its legitimate political power and demanding the exclusion
of refugees from the political process. As Mavrogordatos (1983) puts it, “the distinction between
refugees and natives provided the basis for the most salient cleavage in inter-war Greek society, which
truly dominated the politics of the period.”


2.4    Recent Albanian immigration

Albanians form the largest group of immigrants in present-day Greece. Differently from the 1923
refugees, most Albanian nationals share neither the same language nor the same religion with
Greek natives. They come from a country whose language has no similarities with Greek and
where the predominant religion is Islam (with some Christians that are typically Catholic and not
Orthodox).
   Most Albanian immigrants arrived in Greece after the collapse of the communist regime in

                                                    6
1991 and were largely unwanted by the Greek authorities. In fact, according to Baldwin-Edwards
(2004), the Greek immigration policy since 1991 has been (and continue to be) fundamentally
exclusionary, offering no realistic possibility of legal immigration for non-ethnic Greeks from the
Balkan countries. The illegal status of Albanian immigrants contributed to the negative perceptions
and even hostility of the host population. From their very first arrival in Greece, Albanian migrants
were stigmatised as “criminals” by the Greek media and became stereotyped as “dangerous Albani-
ans” (Karydis, 1992). Sociological studies report that, confronted with systematic discrimination,
Albanian immigrants have adopted tactics to conceal or de-emphasise their Albanian or Muslim
background, by assuming Greek names, being baptised, or pretending to be Christian in order to
avoid prejudices (Hatziprokopiou, 2003; Pratsinakis, 2005). All these factors combined (cultural
differences with natives, illegal status, negative stereotypes) may explain why Albanian immigrants
had not experienced the same successful integration as the 1923 refugees.


3     Related literature and contribution

3.1   Defining and Measuring Social Cohesion

Despite its ubiquity in the literature, social cohesion remains a concept with no consistent defini-
tion. Social cohesion is variously described as the absence of social conflict (Durkheim, 1897), the
“affective bond between citizens” (Chipkin and Ngqulunga, 2008) “local patterns of cooperation”
(Fearon, Humphreys and Weinstein, 2009) and “the glue that bonds society together, promoting
harmony, a sense of community, and a degree of commitment to promoting the common good”
(Colletta et al., 2001). A contemporary definition that I will adopt for the rest of the paper is given
by Kim, Sheely and Schmidt (2020): “a sense of shared purpose and trust among members of a
given group or locality and the willingness of those group members to engage and cooperate with
each other to survive and prosper”.
    To operationalize this definition, this paper focuses on the presence or absence of major inter-
group cleavages, which according to Chan, To and Chan (2006) and King, Samii and Snilstveit
(2010) represents a key dimension of social cohesion, as strong divisions (whether by income, eth-
nicity, political party) can polarize the society and strain social solidarity. In the first part of the
paper, I use different measures of refugees’ socio-economic integration and cultural assimilation to
test whether the refugee inflow has caused the emergence of new cleavages (native vs. newcomer)
within the Greek society as a whole. I notably use survey data on self-reported trust and self-
perceived discrimination to measure the extent of social exclusion of the refugee population. I also
use census data to measure educational and occupational inequalities between the two groups. In
the second part of the paper, I focus on community-level measures of social cohesion, understood
either as absence of societal conflict and division or, more positively, as civic participation. In partic-



                                                    7
ular, I use the presence of sport associations in a community to measure the degree of involvement
in voluntary associations, which is a conventional indicator of community ties and pro-social norms
in the literature on social capital.11 I also use municipal or regional level indicators of crime rates
and political fragmentation.


3.2     Refugees’ integration

While the literature on voluntary migration has a long tradition of studying the assimilation of
migrants (Abramitzky and Boustan, 2017; Verdier et al., 2012), a smaller but fast-growing body
of work has specifically focused on forced displacement and the integration of refugees in the host
society. Relative to migrants, refugees are more disadvantaged from the start for a number of rea-
sons: (i) having experienced persecutions and expulsions, they may bear lasting effects of physical
and psychological trauma, (ii) most of them have lost possessions and assets due to destruction
or hasty departure, (iii) as they often have limited control over their final destination, they may
resettle in suboptimal locations where they cannot rely on past family or ethnic networks, or where
their skills are not easily transferable. Consistent with these initial disadvantages, a number of
studies show that present-day refugees experience slower and more difficult economic integration
than other migrants and this gap persists over time up to 10-15 years after migration – see Fasani,
Frattini and Minale (2021b) and Bevelander (2020) for a review.
    On the contrary, taking a longer-term perspective by studying historical expulsions of population
in the wake of WWII, a number of papers show that forced displacement does not necessarily have
detrimental effects. Looking at the resettlement of Poles, Becker et al. (2020) shows that being
forcibly uprooted (and expropriated) has lasting positive effects on educational choices – over
three generations – by increasing the subjective value of investing in portable assets, in particular
in education. Forced migrants and their descendants tend to invest more in human capital also
because, in the case of refugee farmers, they may not have access to land and may thus need to
                                                                              aki, Uusitalo and
invest in skills outside agriculture (Bauer, Braun and Kvasnicka, 2013; Sarvim¨
 antti, 2019). More in general, while forced displacement may in itself leave deep scars, it compels
J¨
refugees to start from scratch and reinvent themselves, which may put them on an upward social
trajectory, possibly across generations.12 Furthermore, unlike voluntary migrants, refugees often
have nowhere to return to and may thus have a stronger incentive to invest in country-specific
human capital in order to assimilate in their new destination (Cortes, 2004).
    In the debate over the determinants of refugee integration in the host country, increasing atten-

   11
     See Putnam (2000, 2002) and Stolle (2003) among others.
   12
     Nakamura, Sigurdsson and Steinsson (2016) call this “the gift of moving”. The authors document the positive effect
of forced migration on education and lifetime earnings of families displaced by the eruption of a volcano off the coast
                                                                      aki, Uusitalo and J¨
of Iceland in 1973.They interpret their findings in a similar to Sarvim¨                  antti (2019): some people are
“stuck” in locations and habits that do not fully exploit their economic potential. When displaced, people are able to
benefit from opportunities available away from the origin.


                                                          8
tion has been given to the crucial role of asylum and resettlement policies, both in developed and
developing countries. In particular, previous studies have analysed the effectiveness of different
program aimed at removing barriers to refugees’ (formal) employment, with the idea that facilitat-
ing their participation in the labor market is instrumental in fostering integration (Clemens, Huang
and Graham, 2018; Fasani, Frattini and Minale, 2021a). Another strand of research has shown the
considerable benefits of giving greater rights to newcomers (Gathmann et al., 2020). In particular,
granting the citizenship dramatically improves the rate of economic and social integration of immi-
grants, and encourages greater investments in immigrant children’s human capital and education
(Felfe, Rainer and Saurer, 2020).
    This paper contributes to the existing literature in several ways. First, unlike most of studies
on refugees’ integration , I investigate medium and long-run outcomes across the first and second-
generation of refugees. I also consider a broader set of outcomes: apart from educational attain-
ment, I explore intermarriage with natives, political attitudes, voting, trust, and civic engagement
of second-generation refugees. Furthermore, the specific historical context of the Greek refugee in-
flow – in which refugees were established in agricultural work and granted the citizenship – allows
me to shed light on the effectiveness of resettlement policies in a poor agrarian economy, not unlike
many developing countries today.13 Finally I provide new, albeit suggestive, evidence on an often
overlooked factor of long-term integration: the access to primary schooling for refugee children.


3.3     Refugees’ effects on social cohesion

A large literature has explored how ethnic diversity affects social capital. Higher diversity is usu-
ally found to be associated with lower mutual trust and civic participation (Alesina and La Ferrara,
                                                         emet and Laitin, 2016), and lower wel-
2000, 2002; Putnam, 2007), higher social anomia (Algan, H´
                                                                                         no-Ort´
fare and public redistribution (Alesina, Baqir and Easterly, 1999; Desmet, Weber and Ortu˜     ın,
2009) – see Alesina and Ferrara (2005) for a review. With respect to the effects of (present-day)
refugees more specifically, a number of studies has explored how natives’ support for far-right pop-
ulist parties respond to refugee influx. Results on this topic are mixed: depending on the context,
certain studies find that local exposure to refugees increases far-right voting (Dustmann, Vasiljeva
and Piil Damm, 2019; Hangartner et al., 2019; Dinas et al., 2019), while others find that expo-
sure to refugees either reduces or has no impact on anti-refugee voting in hosting municipalities
(Steinmayr, 2020; Gehrsitz and Ungerer, 2017). These conflicting results can be reconciled by the
intergroup contact theory, which predicts that different types of exposure can generate different at-
titudinal reactions towards newcomers.14 According to this theory, the contact between an in-group

   13
      85 percent of the 70 million currently displaced people in the world are hosted in developing countries (UNHCR,
2018). Sub-Saharan Africa currently hosts the largest number of forcibly displaced people (26 millions), which are
mostly resettled in rural areas (Huang and Graham, 2018).
   14
      See Allport’s (1954); Pettigrew (1998); Pettigrew and Tropp (2006); Pettigrew et al. (2011).


                                                         9
(i.e., the native population) and an out-group (i.e., the refugees) reduces prejudice if the following
features characterise the contact situation: firstly, equal status of the groups in the situation, which
means that the two groups are equally engaged in the relationship and ideally have similar back-
grounds and characteristics. Secondly, the groups work together on common goals, which should
not be achieved in competition but rather in intergroup cooperation. Thirdly, personal interaction
with cross-group members. Lastly, support of authorities, law, or custom.
     This paper contributes to this extensive literature by highlighting the important role of reset-
tlement policies in creating favourable conditions for positive intergroup contact. By providing
refugees with farmland and making the self-sufficient, the League of Nations reduced the com-
petition with natives in the labor market. By opening new schools, constructing new roads and
dispensaries, the Leagues’ public utility program further helped mitigate the risk of congestion and
overcrowding caused by the sudden refugee inflow. Granting the citizenship also contributed to
equal status between the newcomers and the locals.


4        Research Design

4.1      Data

4.1.1     Historical data on the refugees

The primary data source to identify the spatial distribution of the refugee inflow is the 1928 pop-
ulation census.15 This dataset provides information on the number of refugees for reach of the
140 provinces and 5,042 municipalities that existed in Greece in 1928. While the main variable
of interest is the population share of refugees in 1928, I also draw on the 1923 refugee census at
the municipality level to check the robustness of the analysis to using the distribution of refugees
in 1923.16 The 1928 census provides a rich set of information on the socio-economic character-
istics, e.g., literacy rate by age and professions, of the total Greek population and of the refugees
separately. It also tabulates the characteristics of the population in 1920 for comparison. However,
these information are available only at the province level level.
     According to the refugee census, 778,000 refugees arrived in Greece in April 1923. After the
population exchange, their number rose up to 1.07 million in 1928, accounting for 16.6 percent
of the total population (out of 6.08 million). 53.9% of refugees settled in urban areas, while only
30% of Greek natives lived in cities in 1928. The three biggest cities in Greece, Athens, Piraeus,
and Thessaloniki, hosted about half of the urban refugees. I exclude these three cities from the


    15
     See Figure A.2 in the Appendix for the questionnaire (bulletin) and an example of historical table.
    16
     This census was taken in April 1923, after the mass inflow of refugees but before the Lausanne Peace Treaty and the
agreement on the forced population exchange between the Kingdom of Greece and Turkey. Figure A.1 in Supplementary
Appendix presents the timeline of events that are relevant for the settlement and enumeration of refugees.


                                                         10
analysis as they may have a different development trajectory and may thus introduce endogeneity.
Figure 1 maps the distribution of refugees in 1928 at the community level.


4.1.2     Contemporary social cohesion at the local level

I use three main indicators of social cohesion at the local level: participation in voluntary associa-
tions, political fragmentation, and crime.
Sports clubs To measure participation in voluntary associations, I use the register of sports clubs
inventoried by the General Secretariat of Sports (GGA) in the early 2000s. The dataset provides the
geolocation of more than 22,000 sport association across the country. The geolocation is precise
enough to identify the number of association within each of the 5775 communities that existed in
Greece in 2001.
    Political fragmentation To measure political fragmentation at the local level, I follow the po-
litical science literature (Gross and Sigelman, 1984) by using an index of fragmentation of votes
between different political parties:

                               Political fragmentation = 1 −                 Vote share2
                                                                                       i
                                                                   Party i


This index simply measures the probability that two people drawn at random in the population
have voted for the same party at the last election. This index is lower when people tend to vote
for the same party and higher when votes are scattered between different parties. Since the Greek
political landscape has many parties that sometimes hold relatively similar ideological position, I
also used a fragmentation index where parties are grouped into a leftist coalition and a rightist
coalition.The electoral data I use covers the legislative elections of June 2012 and January 2015
and provides the number of votes for each of the 325 municipalities of Greece at that time.17
Crime data I draw on crime data from Vasilakis (2018) who collected the data from the Greek
police. This data comprise the total number of robberies, drug arrests and burglaries in 2015 for
each of the 56 departments of Greece.
Matching places over time Out of the 5,042 municipalities that existed in 1928, 4,845 could be
geocoded and matched to the 5,775 communities that existed in Greece in 2001. I collapse the
dataset at the administrative borders of 1920 and treat observations that merged after 1920 as a
single unit. As a result, I obtain a panel dataset with 4,409 observations (cluster of municipalities)
covering the period between 1920 and 2001, achieving a match of 97.7 percent of the 1928 popu-
lation to the rest of the dataset.18 Similarly, out of the 325 municipalities of Greece in the 2010s, I

   17
      Between 1991 and 2010, Greece went through two major administrative reforms that sharply reduced the number
of municipalities from more than 5,000 to less than 400 (known as the 1997 Kapodistrias reform and the 2010 Kallikratis
Programme)
   18
      See Murard and Sakalli (2018) for details of the matching procedure.


                                                         11
could match 275 to the 1928 census.


4.1.3   Integration outcomes of refugees

I measure the integration of refugees by examining different generations of refugees: (i) those
born in turkey who were forcibly resettled in Greece at a different age (ii) those born in Greece to
refugee parents, i.e. the second-generation.
First-generation I measure the socio-economic integration of first-generation refugees by compar-
ing the educational attainment of refugees displaced at a different age, i.e., either when adults
(above 15) or when children (below 15). I rely on the 1928 census to obtain the literacy rate of
refugees at arrival, and focus on those older than 15 years in 1923. To measure the socio-economic
outcomes of refugees at a later period, I unfortunately cannot draw on population censuses as no
information on country of birth is available. The earliest census to provide this information is the
2001 census and I use the 10 percent extract provided by IPUMS.19 I identify as refugees the 3,486
sampled individuals born in Turkey between 1908 and 1922, who represent about 9% of the pop-
ulation born in the same period. Reassuringly, the share of individuals born in Turkey drops to
less than 2% among cohorts born after 1922, while the share of foreign-born (in other countries
than Turkey) remains stable across cohorts and start increasing only after the 1950s (see Figure
A.1 in the appendix). The 2001 census also allows to construct the rate of intermarriage between
refugees and natives. Among the 2,102 refugee women in the sample, 293 report to be married
with a currently co-residing husband – the other women being mostly widowed. When the hus-
band belongs to the surveyed household, I can identify mixed refugee-native couple as instances in
which the wife is refugee and the husband is born in Greece. Theoretically, although native-born,
the husband could still be the child of refugee parents, which would hardly correspond to a mixed
couple. This is almost never the case in practice as husbands are typically older than their wife,
and hence are born before refugees arrived in Greece for the first time.
Second generation I use the European Social Survey (ESS) that provides information on parents’
birthplace for a sample of individuals living in Greece in the 2000s. I identify second-generation
refugees as individuals born in Greece with at least one parent born in Turkey. I focus on the cohorts
born between 1923 and 1950 as they are the most likely to be the children of first-generation
refugees. Table A.2 in the appendix shows that the proportion of ESS respondents with parents
born in Turkey is the highest for people born in the 1920s, the 1930s, and the 1940s. It dramatically
drops for people born after 1950. In contrast, the share of individuals with parents born outside
of Turkey remains stable across birth cohorts, with a peak in the 1990s. This provides empirical
support that I am able to identify second-generation refugees. The ESS does not only provide data
on the socio-economic outcomes of second-generation refugees (such as education or occupation)
   19
    Minnesota Population Center. Integrated Public Use Microdata Series, International: Version 7.3 [dataset]. Min-
neapolis, MN: IPUMS, 2020. https://doi.org/10.18128/D020.V7.3


                                                        12
but also contains information on a wide range of socioeconomic and political attitudes, beliefs, and
values.


4.2     Empirical strategy

The first part of the empirical analysis is essentially descriptive as it documents the socio-economic
outcomes of refugees for different generations and relative to the native population. The second
part aims at examining the long-term consequences of the refugee resettlement on the social cohe-
sion of host communities.
    I estimate the long-term effects of refugees on social cohesion using the following regression at
the community level:


                                             Ycp = Refcp + Xcp α + δp + εcp                                                   (1)

where c indexes communities and p indexes provinces. Ycp is a measure of contemporary social
cohesion in the 2000s. Refmp is the share of refugees in the 1928 population of community c.
Importantly, the specification includes a set of province fixed-effects δp that controls for unobserved
heterogeneity at the province level. The vector of covariates Xcp , comprises population density in
1920; a dummy variable indicating whether the locality was a city in 1920; and some geograph-
ical features: a dummy variable indicating whether the municipality is on an island; distances to
the railway network in 1920, to the shoreline, and to the national border; altitude; average an-
nual temperature and precipitation; and suitability of crops for rain-fed agriculture. I correct the
standard errors for spatial correlation across communities by clustering the standard errors at the
province level.
Identification An important concern with the estimation equation (1) is the self-selection of refugees
into their final places of settlement. Refugees could have been attracted to places with better eco-
nomic opportunities and greater potential for long-run growth. This could result in biased OLS
estimates. I address this concern in two different ways. I first document in section B of the ap-
pendix that places of resettlement were neither more prosperous nor growing faster before the
refugee inflow. If anything, places of resettlement tended to be more agricultural ans less econom-
ically developed. This is consistent with historical accounts according to which the RSC assigned
refugees to agricultural colonies mostly based on vacant land suitable for cultivation. As a result,
refugees could not often choose their settlement location.20 Second, I test the robustness of the
estimates to using the distribution of refugees in 1923 instead of that in 1928. Self-selection on
   20
     The refugees were transported to the settlement location, where a government representative delivered the land
(Pentzopoulos, 1962, p. 108). The RSC planners determined the settlement locations based on “several quite patent
factors, of which the geographical was the controlling one. The capacity of the country in the absorption of refugees was
limited by its physical features” (Kontogiorgi, 2006, p. 255). Cases of individual settlement were very rare and the reset-
tlement process “left little if any scope for choice, on the part of the refugees, as regards the place of their final resettlement”
(Kontogiorgi, 2006, p. 143).


                                                               13
unobservables is arguably less problematic for the distribution of refugees reported in the census
of April 1923. At that time, refugees had been in Greece for less than 6 months (since the fire of
Smyrna in September 1922), and still hoped to return to their homeland in Turkey. Upon arrival,
the main worry of refugees, who were in a state of “utter destitution”, was to find relief and shel-
ter where they could and less so to seek places with better economic opportunities (Kontogiorgi,
2006, pp. 88). The first place of settlement in 1923 was often temporary, as more than one third
of refugees relocated to a different settlement between 1923 and 1928. As Figure A.5 shows, the
spatial distribution of refugees in 1923 is correlated with but far from identical to the one in 1928.


5        Results

5.1      Social integration of refugees

Closing the gap I first examine the educational outcomes of refugees. The Panel A of Table 1 shows
that, among adult refugee men in the 1928 census (15-65 years old in 1923), 66% are literate, while
it is the case for the 75% of native men.21 This initial negative gap disappears for refugees arrived
in Greece at a younger age, i.e., below 15. The 1,270 refugee men sampled in the 2001 census
and born between 1908 and 1922 are as likely as natives (born in the same period) to be literate
(93%) and to have completed primary school (70%).22 Differently from men, adult refugee women
in 1923 had a literacy rate similar to native women (around 35%). When displaced as children,
they became later in life more educated than native women. As for the second-generation refugees
born in Greece, the Panel C of Table 1 shows that, among adult men, descendants of refugees and
natives have similar educational attainment. Among women, refugees’ offspring are more educated
than natives’.
Factors of convergence What factors can account for the observed pattern of convergence? Con-
fronted with a mass inflow of refugee children that increased the school-age population by more
than 20% in a few months, the RSC and the Greek State managed to open enough new schools
to avoid overcrowding and accommodate the newcomers (see section 2). This public investment
likely contributed to the impressive progress in literacy among refugees who were young enough
to seat in the newly opened classrooms: the 1928 census reveals that the refugee-native gap in
literacy went from -12% to less than -1% among male cohorts born between 1910 in 1917. Figure
2 provides evidence of the important role of the school supply. In provinces where the number of
newly opened schools (per children, over the 1921-1917 period) was low (bottom tercile), male
    21
      This data is provided by the 1928 census who ask respondents whether they are able to read and write in their
mother tongue (not necessarily in the Greek language) - See Figure A.2.
   22
      The sample I use in the 2001 census are made of individuals older than 78. One concern might be it is a non-
random sample with respect to education as mortality rates are typically higher for less educated people. Yet there is no
reason to expect that the educational gradient of mortality rates is different between natives and refugees. Second, life
expectancy is typically above 80 years in Greece.


                                                          14
refugees did not catch up with natives and the male literacy gap only narrowed from -10 to -5
percentage points between the 1910-1917 cohorts. In contrast, in provinces where the increase in
school supply was higher, the literacy gap closed entirely by 1917, with a stark catching up from
-15 to +1 percentage point in just seven cohorts.

    To put the social integration of refugees into perspective, it is useful to examine another group of
immigrants. I focus on the largest group of immigrants in Greece (about two-third of the immigrant
population), the Albanians, most of whom arrived in the 1990s, after the the collapse of the Soviet
Union.


5.2     Refugees’ integration compared to Albanian immigrants

Education and occupation Table A.4 in the appendix presents regression estimates of the educa-
tional gap between Albanian immigrants and Greek natives, for different generation of migrants.
In terms of secondary schooling Albanian migrants who arrived in Greece when adults are 22 per-
centage points less likely to have completed secondary schooling relative to natives (71% of whom
hold secondary education). There is absolutely no catching up over generations: the educational
gap with natives remains almost unchanged for migrants who arrived at a younger age, i.e., below
12, and even below 5. The same is true in terms of tertiary education. Furthermore, the ESS data
allows to examine the occupations of second-generation refugees, either in their current or in their
last job.23 As Table A.5 in the appendix shows, refugees are less likely than natives to report be-
ing farmers and they have comparable household income (by decile). In contrast, first-generation
Albanian immigrants are more likely to work in low-skill occupations (in the non-primary sector)
and have a lower household income than natives.
Intermarriage Refugees are more socially assimilated than recent Albanian immigrants in terms
of intermarriage. Figure 3 reports the intermarriage rate of refugee women who were displaced
at a young age, below 15. By 2001, almost 50% of refugee women got married with a Greek
native. This represents a very high intermarriage rate in international comparison, with a typical
rate between 2% and 25% among immigrant women in Western Europe in the 2000s (Lucassen
and Laarman, 2009). In particular, the refugees’ high propensity to intermarry stands in contrast
with Albanian immigrant women, who, despite living in Greece since childhood (15 or younger),
have a less than 10% probability of getting married to a Greek native.




   23
     The occupation of first-generation refugee cannot be observed in the 2001 census since most refugees are retired
(being 79 or older).


                                                        15
5.3    Trust, Political preferences and civic engagement

Was the successful integration of refugees sufficient to lead to full cultural assimilation and the
disappearance of refugees’ distinctive identity and values? Or, to the contrary, has the historical
strife between refugees and natives (detailed in section 2) cemented into persistent mistrust, dis-
crimination, and polarized political attitudes?

Trust and self-perceived discrimination To address these questions, I first look at trust and self-
perceived discrimination, as elicited in the ESS survey. Table 2 shows that only 3% of second-
generation refugees report to have ever felt discriminated against, no differently from natives. This
stands in stark contrast with first-generation Albanian immigrant, 50% of whom declare to have
been discriminated against, as Table 3 shows. This difference is also related to the fact that only
54% of Albanians report to speak Greek at home, while 99% refugees do. Another crucial difference
is that only 12% Albanians report to have the Greek citizenship while all refugees have it. With
respect to self-declared trust in others or in institutions (legal system, police, politicians), I find no
significant differences between refugees and natives, with a gap of typically less than 5% of the
standard-deviation in the attitudinal scale between the two groups.
Voting I also examine the voting patterns of second-generation refugees and of Greek natives.
The ESS asks respondents the political party they have voted at the last national election. Ta-
ble 2 shows that, on average, the voting patterns of refugees and natives are virtually identical.
Self-reported political preferences measured with a 11-point left-right scale are also very similar
between refugees and natives, with a gap of less than 2% of the standard-deviation of this scale
between the two groups.
Political values Forced displacement, a life-changing experience of “uprootedness”, can have long-
lasting effect on individual preferences. An important body of evidence has documented how
exposure to violence, conflicts, or economic shocks, even temporary, can induces a permanent shift
in individual preferences, possibly persisting across generations.24 Table 2 shows that, relative to
Greek natives, second-generation refugees report more egalitarian preferences, and in particular
higher support for the principle of equality of opportunity. Refugees also display more altruistic
values and greater openness towards understanding different people. This is in line with the exist-
ing evidence that exposure to conflict and to economic shocks (e.g., loss of possessions in case of
displacement) increases individual support for redistribution and egalitarian values (Giuliano and
Spilimbergo, 2014; Bauer et al., 2014) . Furthermore, refugees report greater support for a strong
government that protects its citizen and ensures safety, as one could expect from people whose
family and community have been persecuted and forcibly displaced by conflict.
Civic engagement Importantly, refugees’ more egalitarian and prosocial values result into more

  24
    See, among others, Bauer et al. (2014); Voors et al. (2012); Cassar, Grosjean and Whitt (2013). On economic
shocks, c.f. Giuliano and Spilimbergo (2014) and Fisman, Jakiela and Kariv (2015).


                                                      16
civic behaviours. Refugees report to be more interested in politics than natives, and are more likely
to engage in political activism (boycott, petition, demonstration, etc.). They are also significantly
more likely to participate in voluntary organisations, such as sport or cultural clubs, trade unions,
or humanitarian organisations. Again, this pattern contrasts with Albanian immigrants, who are
much less likely to engage in political or civic activities relative to Greek natives (Table 3). These
findings are consistent with existing evidence (Blattman, 2009; Bellows and Miguel, 2009) that
exposure to violence and conflicts – such as the religious persecutions Greek refugees were victim
from in Turkey – can lead victims to higher political and civic mobilization and engagement (as
a post-traumatic coping mechanism).25 An important body of work even suggests that individual
exposure to war violence may increase social cooperation at the local level, including community
participation and prosocial behavior (for a review, see Bauer et al. (2016)).

    One question that remains is whether Albanians will gradually assimilated over time, and in
particular whether the second generation of Albanian immigrants will be equally integrated as the
1923 refugees. Although it is probably too early to say (as most Albanian arrived in Greece after
1991), the ESS provides some suggestive evidence that Albanians have not yet experienced any
progress in integration outcomes. Although the ESS sample contains (almost) no second generation
Albanians, I can examine the integration outcomes of first generation migrants who arrived in
Greece at different ages. If second generation migrants were to integrate better than the previous
generation we would expect to detect such a trend already between migrants who arrived at a
young and older age. Table A.6 in the appendix shows that it is not the case: migrants who arrived
at a young age (<24) are as likely as those who arrived later in life to report Greek as the first
language spoken at home, to feel discriminated, to have the Greek citizenship, or to participate
in civic organisation. This apparent absence of social assimilation is consistent with the lack of
integration in terms of educational outcomes among younger and older immigrants from the first
generation (see Table A.4).


5.4    Effects of the refugee inflow on host communities’ social cohesion

I now explore the effect of the refugee inflow on the social cohesion of the host communities in the
long-run, more than 80 years after the resettlement.
Sport associations Table 4 presents the OLS estimates of the relationship between the 1928 share
of refugees and the presence of at least one sport association in the early 2000s at the community
level. I estimate various specifications: starting from a regression controlling for geographical and
historical variables (see description in section 4.2 ), for the community’s population in 1991 and
   25
      A set of the literature explores positive responses to trauma, the so-called post-traumatic growth theory (Tedeschi
and Calhoun, 1996; Powell et al., 2003), that also includes changes in political action and beliefs. For example, Palestini-
ans who personally survived aerial attacks are more likely to engage in political activism (Punamaki, Qouta and Sarraj,
1997).


                                                            17
for some indicators of contemporaneous socio-economic development (column 1), I add to the re-
gression controls region fixed-effects (column 2), and province fixed-effects (column 3).26 I find
no significant effect of refugees on average. I then examine whether the effect is heterogeneous
depending on the population size of the community as the presence of a sport association greatly
varies with the population (almost all communities with more than 1,100 inhabitant have a asso-
ciation, while it is the case for only 58% of smaller communities). As columns 4 to 6 show, the
effect of refugees becomes significantly positive among small communities with less than 1,100 in-
habitants (which corresponds to the bottom two terciles of the distribution), while the effect is null
in larger communities.27 The estimates suggest that, among small communities, a 10 percentage
point increase in the share of refugees is associated with a 1.1 percentage point increase in the
probability of having at least one sport association.
    Importantly, this result is robust to using the distribution of refuges in 1923 (instead of 1928).
At that time, refugees had been in Greece for less than 6 months and self-selection on unobservables
is arguably less problematic as their principal worries were to find relief and shelter. Table A.7 in
the appendix shows that, among small communities, the share of refugees in 1923 is positively
correlated with the presence of a sport association.
    This positive relationship is fully consistent with refugees’ higher civic engagement relative to
natives, as evidenced in section 5.3. Interestingly, I find that Albanian immigrants – who are less
socially integrated than refugees and rarely engage in public life – have no significant effects on
presence of sport associations in the communities they live in (see Table A.8 in the appendix). This
suggests that refugees’ positive impact on social capital is likely driven by their active participation
in the civic and political life and by their efforts of assimilation in the Greek society more in general.

Political fragmentation Table 5 reports OLS estimates of the effects of refugees on political frag-
mentation across 325 municipalities of Greece in the 2010s. I find no evidence that refugees’
presence significantly affect political fragmentation, either in the 2012 or in the 2015 election. I
continue to find insignificant effects when I use an alternative fragmentation index in which parties
are grouped into a leftist and a rightist coalition (Table A.9), or when I estimate alternative spec-
ifications using a quadratic or a logarithm function of the refugees’ share (Table A.10). This null
result is consistent with the fact that refugees’ voting patterns are extremely similar to natives.
Crime Table 6 presents some suggestive evidence on the refugees’ impact on crime at a broad
geographical level, i.e., across 55 departments of Greece. I find that a higher share of refugees is
significantly associated with a lower number of crimes in 2015, with a 1% increase in the refugees’
share leading to a 2.4% reduction in the number of crimes.

   26
      More specifically, regression controls in column 1 include terciles of the 1991 population, the share of secondary
and tertiary educated individuals in the 20-65 population in 1991, the employment share of non-agricultural occupations
and of the manufacturing sector in 1991.
   27
      All regressions include a binary variable for small communities in the controls.


                                                         18
6        Policy and Program Implications
Confronted with the sudden influx of 1.2 million destitute refugees, Greece – a poor agrarian
country drained by several years of war – was under threat of serious social turmoil in 1923.28
The timely intervention of the League of Nations and the ambitious resettlement program that
ensued turned this historical challenge into a formidable opportunity in the long-run. As Murard
and Sakalli (2018) document, refugees contributed to higher economic growth in places of resettle-
ment, notably by bringing new complementary skills and knowledge conducive to industrialization.
     This paper shows that, while fostering economic prosperity, the refugee resettlement did not
undermine social cohesion in the long-term.            29   To the contrary, I find that refugees successfully in-
tegrated in the Greek society, both in social, political and cultural terms, and that their active civic
participation contributed to sustain social cohesion in places of resettlement. This was achieved
by an ambitious integration policy that combined several features: (i) the provision of productive
assets (farmland, livestock) that made refugees self-reliant (ii) an infrastructure program that pre-
vented the overcrowding of public services (dispensaries, schools) and accommodated the newcom-
ers (new houses) (iii) the granting of the citizenship that included the newcomers in the national
community and give them political rights. The combination of the above likely created favourable
(if not necessary) conditions for refugees’ assimilation and positive intergroup contact with the
native population under the umbrella of national unity promoted by the state.
     The historical setting was one in which refugees spoke the same language, had the same reli-
gion, and had more cultural similarities than differences with the local native population. These
findings are thus likely relevant for current refugee inflows that are hosted in neighboring countries
culturally similar to the refugees’ origin country, which is a frequent context of refugee resettlement
at the global level (Devictor, Do and Levchenko, 2021).30 The stark discrepancy in the integration
outcomes of past refugees and present-day Albanian immigrants coming from a Muslim-majority
and non-Greek speaking country (and who were typically denied legal residence permit in Greece)
suggests that the paper’s policy recommendations may not be applicable when newcomers have
profoundly different linguistic, religious and cultural profiles relative to those of the locals. For
refugees with a totally different background from the locals, integration services may need to be
tailored to their specific needs and profile so that they can benefit from the integration policies.
     As Clemens et al. (2018), puts its, “immigration is not inherently “good” or “bad”; its effects

    28
      Among which an epidemic outbreak, as typhus and cholera were ravaging the refugee camps.
    29
      This result contrasts with Tabellini (2020) who finds that, in the U.S. context, the economic benefits brought by
early 20th century immigrants came at the expense of political divisions, lower redistribution and lower social cohesion.
   30
      According to the authors, in the period 2012-2017, 77 percent of global refugees find themselves in a country that
shares a land border with their home country. It is there often the case that refugees are hosted in countries that are
culturally similar. For example, Bangladesh hosts current Muslim Rohingya refugees who fled religious persecution in
Myanmar; most of the Syrian refugees are hosted in neighboring Muslim countries some of which are Arabic speaking;
and Colombia hosts Venezuelan refugees who migrated after the economic collapse of the country.


                                                            19
depend on the context and the policy choices that shape it.” In the Greek setting, the notion that
“migration is what you make it” could not be more accurate, as only the critical resettlement efforts
of the League of Nations and the Greek state could have turned a potential refugee crisis into what
Pentzopoulos (1962) has termed as a “blessing in disguise”.




                                                 20
References
Abramitzky, Ran, and Leah Boustan. 2017. “Immigration in American economic history.” Journal
  of economic literature, 55(4): 1311–45.

Albarosa, Emanuele, and Ben Elsner. 2021. “Forced Migration, Social Cohesion and Conflict:
  The 2015 Refugee Inflow in Germany.” Unpublished Working paper. Commissioned as part of
  the “Preventing Social Conflict and Promoting Social Cohesion in Forced Displacement Contexts”
  Series. Washington, DC: World Bank Group.

Alesina, Alberto, and Eliana La Ferrara. 2000. “Participation in heterogeneous communities.” The
  quarterly journal of economics, 115(3): 847–904.

Alesina, Alberto, and Eliana La Ferrara. 2002. “Who trusts others?” Journal of public economics,
  85(2): 207–234.

Alesina, Alberto, and Eliana La Ferrara. 2005. “Ethnic diversity and economic performance.”
  Journal of economic literature, 43(3): 762–800.

Alesina, Alberto, Reza Baqir, and William Easterly. 1999. “Public goods and ethnic divisions.”
  The Quarterly Journal of Economics, 114(4): 1243–1284.

                      emet, and David D Laitin. 2016. “The social effects of ethnic diversity at
Algan, Yann, Camille H´
  the local level: A natural experiment with exogenous residential allocation.” Journal of Political
  Economy, 124(3): 696–733.

Allport’s, Katz I Gordon. 1954. “The nature of prejudice.” Cambr., Mass.

Baldwin-Edwards, Martin. 2004. “Albanian emigration and the Greek labour market: Economic
  symbiosis and social ambiguity.” SEER-South-East Europe Review for Labour and Social Affairs, ,
  (01): 51–65.

                                                a, and Joseph Henrich. 2014. “War’s enduring
Bauer, Michal, Alessandra Cassar, Julie Chytilov´
  effects on the development of egalitarian motivations and in-group biases.” Psychological science,
  25(1): 47–57.

                                                   a, Joseph Henrich, Edward Miguel, and
Bauer, Michal, Christopher Blattman, Julie Chytilov´
  Tamar Mitts. 2016. “Can war foster cooperation?” Journal of Economic Perspectives, 30(3): 249–
  74.

Bauer, Thomas K., Sebastian Braun, and Michael Kvasnicka. 2013. “The Economic Integration of
  Forced Migrants: Evidence for Post-War Germany.” The Economic Journal, 123(571): 998–1024.


                                                21
Becker, Sascha O, and Andreas Ferrara. 2019. “Consequences of forced migration: A survey of
  recent findings.” Labour Economics, 59: 1–16.

Becker, Sascha O, Irena Grosfeld, Pauline Grosjean, Nico Voigtlander, and Ekaterina Zhu-
  ravskaya. 2020. “Forced migration and human capital: Evidence from post-WWII population
  transfers.” American Economic Review, 110(5): 1430–63.

Bellows, John, and Edward Miguel. 2009. “War and local collective action in Sierra Leone.”
  Journal of public Economics, 93(11-12): 1144–1157.

Betts, Alexander, Maria Flinder Stierna, Naohiko Omata, and Olivier Sterck. 2021. “Social Co-
  hesion and Refugee-Host Interactions: Evidence from East Africa.” Unpublished Working paper.
  Commissioned as part of the “Preventing Social Conflict and Promoting Social Cohesion in Forced
  Displacement Contexts” Series. Washington, DC: World Bank Group.

Bevelander, Pieter. 2020. “Integrating refugees into labor markets.” iza World of Labor.

Blattman, Christopher. 2009. “From violence to voting: War and political participation in
  Uganda.” American political Science review, 231–247.

Cassar, Alessandra, Pauline Grosjean, and Sam Whitt. 2013. “Legacies of violence: trust and
  market development.” Journal of Economic Growth, 18(3): 285–318.

Chan, Joseph, Ho-Pong To, and Elaine Chan. 2006. “Reconsidering social cohesion: Develop-
  ing a definition and analytical framework for empirical research.” Social indicators research,
  75(2): 273–302.

Chipkin, Ivor, and Bongani Ngqulunga. 2008. “Friends and family: social cohesion in South
  Africa.” Journal of Southern African Studies, 34(1): 61–76.

Clemens, Michael, Cindy Huang, and Jimmy Graham. 2018. “The Economic and Fiscal Effects of
  Granting Refugees Formal Labor Market Access.” CGD Report No. 496. Washington DC: Center
  for Global Development.

Clemens, Michael, Cindy Huang, Jimmy Graham, and Kate Gough. 2018. “Migration Is What
  You Make It: Seven Policy Decisions that Turned Challenges into Opportunities.” Center for
  Global Development, May 2018 Note.

Colletta, Nat J, Teck Ghee Lim, Anita Kelles-Viitanen, et al. 2001. Social cohesion and conflict
  prevention in Asia: managing diversity through development. World Bank Publications.

Cortes, Kalena E. 2004. “Are refugees different from economic immigrants? Some empirical evi-
  dence on the heterogeneity of immigrant groups in the United States.” Review of Economics and
  Statistics, 86(2): 465–480.

                                                 22
                                             no-Ort´
Desmet, Klaus, Shlomo Weber, and Ignacio Ortu˜     ın. 2009. “Linguistic diversity and redis-
  tribution.” Journal of the European Economic Association, 7(6): 1291–1318.

Devictor, Xavier, Quy-Toan Do, and Andrei A Levchenko. 2021. “The globalization of refugee
  flows.” Journal of Development Economics, 150: 102605.

Dinas, Elias, Konstantinos Matakos, Dimitrios Xefteris, and Dominik Hangartner. 2019. “Wak-
  ing up the golden dawn: does exposure to the refugee crisis increase support for extreme-right
  parties?” Political analysis, 27(2): 244–254.

Durkheim, Emile. 1897. Le Suicide: Etude de Sociologie. Presses universitaires de France.

Dustmann, Christian, Kristine Vasiljeva, and Anna Piil Damm. 2019. “Refugee migration and
  electoral outcomes.” The Review of Economic Studies, 86(5): 2035–2091.

Fasani, Francesco, Tommaso Frattini, and Luigi Minale. 2021a. “Lift the Ban? Initial Employ-
  ment Restrictions and Refugee Labour Market Outcomes.” Journal of the European Economic As-
  sociation, forthcoming.

Fasani, Francesco, Tommaso Frattini, and Luigi Minale. 2021b. “(The struggle for) refugee in-
  tegration into the labour market: Evidence from Europe.” Journal of Economic Geography, forth-
  coming.

Fearon, James D, Macartan Humphreys, and Jeremy M Weinstein. 2009. “Can development aid
  contribute to social cohesion after civil war? Evidence from a field experiment in post-conflict
  Liberia.” American Economic Review, 99(2): 287–91.

Felfe, Christina, Helmut Rainer, and Judith Saurer. 2020. “Why birthright citizenship matters for
  immigrant children: Short-and long-run impacts on educational integration.” Journal of Labor
  Economics, 38(1): 143–182.

Fisman, Raymond, Pamela Jakiela, and Shachar Kariv. 2015. “How did distributional prefer-
  ences change during the great recession?” Journal of Public Economics, 128: 84–95.

Gathmann, Christina Monscheuer, et al. 2020. “Naturalization and citizenship: Who benefits?”
  IZA World of Labor.

Gehrsitz, Markus, and Martin Ungerer. 2017. “Jobs, crime, and votes: A short-run evaluation of
  the refugee crisis in Germany.” IZA DP No. 10494.

Giuliano, Paola, and Antonio Spilimbergo. 2014. “Growing up in a Recession.” Review of Eco-
  nomic Studies, 81(2): 787–817.



                                                  23
                            on-Ciliotta, and Steven Stillman. 2021. “Immigration, Labor Mar-
Groeger, Andre, Gianmarco Le´
                                                                     u.” Unpublished Working
  kets and Discrimination: Evidence from the Venezuelan Exodus in Per´
  paper. Commissioned as part of the “Preventing Social Conflict and Promoting Social Cohesion
  in Forced Displacement Contexts” Series. Washington, DC: World Bank Group.

Gross, Donald A, and Lee Sigelman. 1984. “Comparing party systems: A multidimensional ap-
  proach.” Comparative politics, 16(4): 463–479.

Hangartner, Dominik, Elias Dinas, Moritz Marbach, Konstantinos Matakos, and Dimitrios Xef-
  teris. 2019. “Does exposure to the refugee crisis make natives more hostile?” American Political
  Science Review, 113(2): 442–455.

Hatziprokopiou, Panos. 2003. “Albanian immigrants in Thessaloniki, Greece: processes of eco-
  nomic and social incorporation.” Journal of Ethnic and migration Studies, 29(6): 1033–1057.

Huang, C, and J Graham. 2018. “Are refugees located near urban job opportunities.” CGD Report.
  Washington DC: Center for Global Development.

Karydis, Vasilis. 1992. “The fear of crime in Athens and the construction of the “dangerous alba-
  nian stereotype”.” Chroniques, 5: 97–193.

Kim, J, R Sheely, and C Schmidt. 2020. “Social Capital and Social Cohesion Measurement Toolkit
  for Community-Driven Development Operations. Washington, DC: Mercy Corps and The World
  Bank Group.”

King, Elisabeth, Cyrus Samii, and Birte Snilstveit. 2010. “Interventions to promote social cohe-
  sion in sub-Saharan Africa.” Journal of development effectiveness, 2(3): 336–370.

Kontogiorgi, Elisabeth. 2006. Population exchange in Greek Macedonia the rural settlement of
  refugees 1922-1930. Oxford:Oxford University Press.

Kritikos, Georgios. 2005. “The agricultural settlement of refugees: A source of productive work
  and stability in Greece, 1923-1930.” Agricultural history, 321–346.

Lucassen, Leo, and Charlotte Laarman. 2009. “Immigration, intermarriage and the changing face
  of Europe in the post war period.” The History of the Family, 14(1): 52–68.

Maddison, A. 2010. “World Population, GDP and Per Capita GDP, 1-2008 AD.” Groningen Growth
  and Development Centre mimeo.

Mavrogordatos, George Th. 1983. Stillbord Republic: Social Coalitions and Party Strategies in
  Greece, 1922-1936. Berkeley:University of California Press.



                                                24
 uller, Tobias, Pia Pannatier, and Martina Viarengo. 2021. “Labor Market Integration, Local
M¨
  Conditions and Inequalities: Refugees in Switzerland.” Unpublished Working paper. Commis-
  sioned as part of the “Preventing Social Conflict and Promoting Social Cohesion in Forced Dis-
  placement Contexts” Series. Washington, DC: World Bank Group.

Murard, Elie, and Seyhun Orcan Sakalli. 2018. “Mass Refugee Inflow and Long-Run Prosperity:
  Lessons from the Greek Population Resettlement.” IZA DP No. 11613.

                osef Sigurdsson, and J´
Nakamura, Emi, J´                     on Steinsson. 2016. “The gift of moving: Intergenera-
  tional consequences of a mobility shock.” National Bureau of Economic Research.

Pentzopoulos, Dimitri. 1962. The Balkan Exchange of Minorities and Its Impact Upon Greece.
  Paris:Mouton & Co.

Pettigrew, Thomas F. 1998. “Intergroup contact theory.” Annual review of psychology, 49(1): 65–
  85.

Pettigrew, Thomas F, and Linda R Tropp. 2006. “A meta-analytic test of intergroup contact the-
  ory.” Journal of personality and social psychology, 90(5): 751.

Pettigrew, Thomas F, Linda R Tropp, Ulrich Wagner, and Oliver Christ. 2011. “Recent advances
  in intergroup contact theory.” International journal of intercultural relations, 35(3): 271–280.

Powell, Steve, Rita Rosner, Willi Butollo, Richard G Tedeschi, and Lawrence G Calhoun. 2003.
  “Posttraumatic growth after war: A study with former refugees and displaced people in Sarajevo.”
  Journal of clinical psychology, 59(1): 71–83.

Pratsinakis, Emmanouil. 2005. “Aspirations and strategies of Albanian immigrants in Thessa-
  loniki.” Journal of Southern Europe and the Balkans Online, 7(2): 195–212.

Punamaki, Raija-Leena, Samir Qouta, and Eyad El Sarraj. 1997. “Relationships between trau-
  matic events, children’s gender, and political activity, and perceptions of parenting styles.” Inter-
  national Journal of Behavioral Development, 21(1): 91–110.

Putnam, Robert D. 2000. Bowling alone: The collapse and revival of American community. Touch-
  stone Books by Simon and Schuster.

Putnam, Robert D. 2002. Democracies in flux: The evolution of social capital in contemporary society.
  Oxford University Press, USA.

Putnam, Robert D. 2007. “E pluribus unum: Diversity and community in the twenty-first century
  the 2006 Johan Skytte Prize Lecture.” Scandinavian political studies, 30(2): 137–174.



                                                  25
Ruiz, Isabel, and Carlos Vargas-Silva. 2013. “The Economics of Forced Migration.” The Journal of
  Development Studies, 49(6): 772–784.

      aki, Matti, Roope Uusitalo, and Markus J¨
Sarvim¨                                       antti. 2019. “Habit formation and the misallo-
  cation of labor: evidence from forced migrations.” Available at SSRN 3361356.

Steinmayr, Andreas. 2020. “Contact versus exposure: Refugee presence and voting for the far-
  right.” Review of Economics and Statistics, 1–47.

Stolle, Dietlind. 2003. “The sources of social capital.” In Generating social capital. 19–42. Springer.

Tabellini, Marco. 2020. “Gifts of the immigrants, woes of the natives: Lessons from the age of
  mass migration.” The Review of Economic Studies, 87(1): 454–486.

Tedeschi, Richard G, and Lawrence G Calhoun. 1996. “The Posttraumatic Growth Inventory:
  Measuring the positive legacy of trauma.” Journal of traumatic stress, 9(3): 455–471.

Tellez, Juan F., and Laia Balcells. 2021. “The social, economic, and political effects of displace-
  ment: Evidence from rural Colombia.” Working paper. Commissioned as part of the “Preventing
  Social Conflict and Promoting Social Cohesion in Forced Displacement Contexts” Series. Wash-
  ington, DC: World Bank Group.

UNHCR. 2018. “UNHCR Global Trends Forced Displacement in 2018.” United Nations High Com-
  mission for Refugees, Geneva.

Vasilakis, Chrysovalantis. 2018. “Massive migration and elections: Evidence from the refugee
  crisis in Greece.” International Migration, 56(3): 28–43.

Verdier, Thierry, Alan Manning, Alberto Bisin, and Yann Algan. 2012. Cultural integration of
  immigrants in Europe. Oxford University Press.

Verme, Paolo, and Kirsten Schuettler. 2019. “The Impact of Forced Displacement on Host Commu-
  nities: A Review of the Empirical Literature in Economics.” World Bank Policy Research Working
  Paper 8727, The World Bank.

Voors, Maarten J, Eleonora EM Nillesen, Philip Verwimp, Erwin H Bulte, Robert Lensink,
  and Daan P Van Soest. 2012. “Violent conflict and behavior: a field experiment in Burundi.”
  American Economic Review, 102(2): 941–64.




                                                  26
  Figure 1: Distribution of refugees in 1928 at the community level

                                        Bulgaria
                        Macedonia
                        (FYROM)


         Albania




                                                                           Turkey




Share of refugees in 1928
     0
     (0, 2]
     (2, 5]
     (5, 10]
     (10, 20]
     (20, 30]
     (30, 40]
     (40, 60]
     (60, 80]




                                                                                Ü
     (80, 100]
     No data                                100    50   0      100 Kilometers



Notes. This figure maps the share of refugees in 1928 at the community level. The
map approximates the municipal boundaries as defined by the 1920 population
census of the Kingdom of Greece.
Source: Author’s elaboration on 1928 population census of the Kingdom of Greece.




                                      27
Figure 2: Literacy gap between refugee and native men in 1928, by level of school supply
increase in the province




          Notes. This figure presents the average literacy gap between refugee and native
          men in 1928, by birth cohors and by two groups of province: (i) 80 provinces in
          which the number of schools (per 1000 children) has increased by more than two
          between 1921 and 1927 (ii) the other 59 provinces, i.e., the ones in which the
          increase in school supply has been lower than two.
          Source: Author’s elaboration on 1928 population census of Greece


                         Figure 3: Intermarriage with Greek natives




          Notes. This figure shows the share of refugee women (born in 1908-1922, i.e.,
          0-15 years old in 1923) who are married with a husband born in Greece (Greek
          native). The sample consist of 293 refugee women who report to be married with
          a currently co-residing husband. The figure also shows the share of Albanian im-
          migrant women married to a Greek native. The sample consist of 452 Albanian
          immigrant women who arrived in Greece at an age younger than 15 and who re-
          port to be married with a currently co-residing husband.
          Source: Author’s elaboration on the 2001 Greek census.


                                                 28
       Table 1: Educational attainment of refugees and Greek natives

                                       Male                                          Female

                         Natives      Refugees        Diff         Natives         Refugees            Diff.
                          (1)           (2)         (2)-(1)         (1)              (2)              (2)-(1)

  Panel A: 1928 Census
 Literacy             0.751            0.666      -0.085***         0.336            0.355          0.019***
                     (0.000)          (0.000)      (0.000)         (0.000)          (0.000)          (0.000)

 N                      1,301,237     263,461     1,564,698      1,342,233          287,981         1,630,214

  Panel B: 2001 Census
 Literacy             0.919            0.931        0.013*          0.688            0.788          0.100***
                     (0.002)          (0.007)       (0.008)        (0.003)          (0.009)          (0.011)
 Primary school       0.693            0.706         0.014          0.431            0.497          0.066***
                     (0.004)          (0.013)       (0.013)        (0.004)          (0.012)          (0.012)
 Secondary school     0.137            0.189       0.052***         0.053            0.103          0.050***
                     (0.003)          (0.011)       (0.010)        (0.002)          (0.007)          (0.006)
 N                   14,371            1,270        15,641         18,325            1,890           20,215

  Panel C: European Social Survey
 Primary school       0.817      0.843                0.026         0.660            0.671            0.011
                     (0.015)    (0.033)             (0.038)        (0.015)          (0.039)          (0.043)
 Secondary school     0.293      0.256               -0.037         0.136            0.194           0.059*
                     (0.017)    (0.040)             (0.045)        (0.011)          (0.033)          (0.031)
 Tertiary education   0.076      0.050               -0.027         0.021            0.069          0.048***
                     (0.010)    (0.020)             (0.026)        (0.005)          (0.021)          (0.015)
 N                     693        121                  814           944              143             1,087
 Notes This table shows the average literacy and education of refugees and Greek natives in different dataset. In
Panel A, the sample consists of individuals born in 1858-1907, i.e., 16-65 adults in 1923. In Panel B, the sample
consists of individuals born in1908-1922, i.e., 0-15 children in 1923. In Panel C, the sample consists of second-
generation refugees and natives’ offspring born in 1923-1950. Standard errors in parentheses. *** p <0.01,
** p <0.05, * p <0.1;
Source: Author’s elaboration on the 1928 and 2001 Greek population censuses and European Social Survey
(ESS).




                                                      29
      Table 2: Political preferences, trust, and civic engagement among second-generation
      refugees and Greek natives

                                                                                  Natives    Refugees    Difference betw. refugees and natives
                                                                                   (1)         (2)        (2)-(1)    With age and gender FE
                                                                                                                                 standardized
 Trust and self-perceived discrimination
 Greek citizen                                                                     0.998      1.000         0.002       0.002          0.051
                                                                                  (0.001)    (0.000)      (0.003)      (0.003)        (0.062)
 Ever feel discriminated against                                                   0.033      0.030        -0.003       0.002          0.010
                                                                                  (0.004)    (0.010)      (0.012)      (0.012)        (0.066)
 Greek as language most often spoken at home                                       0.998      0.996        -0.001       -0.002         -0.045
                                                                                  (0.001)    (0.004)      (0.003)      (0.003)        (0.069)
 Most people can be trusted (11 pts.-scale)                                        3.499      3.353        -0.145       -0.117         -0.048
                                                                                  (0.060)    (0.144)      (0.161)      (0.163)        (0.067)
 Most people try to be fair and not take advantage of you (11 pts.-scale)          3.493      3.417        -0.077       -0.052         -0.022
                                                                                  (0.057)    (0.147)      (0.155)      (0.158)        (0.068)
 Most of the time people helpful (11 pts.-scale)                                   2.984      2.989         0.005       0.022          0.010
                                                                                  (0.056)    (0.146)      (0.152)      (0.154)        (0.068)
 Trust in the legal system (11 pts.-scale)                                         5.656      5.397        -0.259       -0.321         -0.107
                                                                                  (0.074)    (0.188)      (0.200)      (0.289)        (0.263)
 Trust in the police (11 pts.-scale)                                               6.674      6.864         0.190       0.106          0.038
                                                                                  (0.068)    (0.170)      (0.183)      (0.176)        (0.064)
 Trust in politicians (11 pts.-scale)                                              3.123      3.307         0.184       0.103          0.040
                                                                                  (0.064)    (0.166)      (0.174)      (0.163)        (0.063)
 Political preferences
 Voted for PASOK (Panhellenic Socialist Movement )                                 0.450      0.422        -0.028       -0.013         -0.026
                                                                                  (0.015)    (0.036)      (0.040)      (0.039)        (0.079)
 Voted for ND (New Democracy)                                                      0.476      0.508         0.032       0.010          0.019
                                                                                  (0.015)    (0.037)      (0.040)      (0.039)        (0.079)
 Voted for KKE (Communist Party)                                                   0.044      0.049         0.004       0.009          0.044
                                                                                  (0.006)    (0.016)      (0.016)      (0.017)        (0.081)
 Voted for SYN (Leftwing coalition)                                                0.012      0.005        -0.007       -0.005         -0.047
                                                                                  (0.003)    (0.005)      (0.008)      (0.008)        (0.078)
 Self-placement on a left right scale (11 pts.-scale)                              6.148      6.195         0.047       -0.035         -0.016
                                                                                  (0.060)    (0.157)      (0.160)      (0.161)        (0.073)
 Political values
 Government should reduce differences in income levels (5 pts.-scale)              4.345      4.415        0.070        0.066          0.082
                                                                                  (0.020)    (0.046)      (0.052)      (0.053)        (0.067)
 Important that people have equal opportunities (6 pts.-scale)                     5.191      5.489      0.298***     0.309***       0.345***
                                                                                  (0.022)    (0.040)      (0.057)      (0.058)        (0.065)
 Important to help people and care for others well-being (6 pts.-scale)            5.075      5.243      0.169***     0.184***       0.216***
                                                                                  (0.021)    (0.051)      (0.056)      (0.057)        (0.067)
 Important to understand different people (6 pts.-scale)                           4.764      5.008      0.243***     0.248***       0.239***
                                                                                  (0.025)    (0.055)      (0.068)      (0.068)        (0.065)
 Important that government is strong and ensures safety (6 pts.-scale)             5.305      5.459      0.153**      0.165**        0.168**
                                                                                  (0.024)    (0.052)      (0.064)      (0.064)        (0.066)
 Political participation and civic engagement
 Voted last national election                                                      0.935      0.940        0.005        -0.002         -0.008
                                                                                  (0.006)    (0.015)      (0.016)      (0.016)        (0.065)
 Interested in politics                                                            0.301      0.360       0.059*      0.060**        0.131**
                                                                                  (0.011)    (0.030)      (0.031)      (0.030)        (0.066)
 Political activism (petition, boycott, demonstration...)                          0.069      0.113       0.044**     0.050***       0.197***
                                                                                  (0.006)    (0.020)      (0.017)      (0.017)        (0.068)
 Participattion in civic organisation                                              0.107      0.154        0.047       0.060*         0.193*
                                                                                  (0.012)    (0.036)      (0.034)      (0.034)        (0.111)

 N                                                                                 1,652       264         1,916        1,916          1,916
 Notes This table shows the average attitudes and behaviors of second-generation refugees and Greek natives. In the fourth column, the table reports
regression estimates of the refugee-native gap after controlling for age and gender dummies. The fifth column reports the same coefficient using the
standardized outcome (Z-score) instead. Standard errors in parentheses. *** p <0.01, ** p <0.05, * p <0.1;
Source: Author’s elaboration on the European Social Survey (ESS).




                                                                        30
Table 3: Self-perceived discrimination and civic involvement of Albanian immigrants

                                                                     Natives        Albanians          Difference betw. refugees and Albanians
                                                                      (1)              (2)              (2)-(1)      With age and gender FE

  Greek citizen                                                       0.999              0.127         -0.872***               -0.869***
                                                                     (0.000)            (0.027)         (0.006)                  (0.006)
  Ever feel discriminated against                                     0.041              0.503         0.462***                0.454***
                                                                     (0.003)            (0.040)         (0.017)                  (0.018)
  Greek as language most often spoken at home                         0.990              0.541         -0.449***               -0.448***
                                                                     (0.001)            (0.040)         (0.011)                  (0.011)
  Political activism (petition, boycott, demonstration...)            0.144              0.052         -0.092***               -0.142***
                                                                     (0.005)            (0.018)         (0.028)                  (0.028)
  Participate in civic organisation                                   0.151              0.079           -0.072                 -0.148**
                                                                     (0.009)            (0.044)         (0.058)                  (0.060)
  N                                                                   4,596               157            4,753                    4,753
  This table shows average attitudes and behaviors of Albanian immigrants and Greek natives. In the fourth column, the table reports regression
 estimates of the refugee-native gap after controlling for age and gender dummies. Standard errors in parentheses. *** p <0.01, ** p <0.05,
 * p <0.1;
 Source: Author’s elaboration on the European Social Survey (ESS).




                               Table 4: Refugee resettlement and sport clubs

          Dep. var:                                                                 Presence of at least one sport club

                                                                    (1)            (2)           (3)           (4)      (5)         (6)

          Share of refugees in 1928                              0.031         0.035         0.049         -0.048      -0.053       -0.042
                                                                (0.033)       (0.036)       (0.036)       (0.034)     (0.039)      (0.040)
          Share of refugees in 1928 * small community                                                     0.151**    0.161***     0.157***
                                                                                                          (0.060)     (0.060)      (0.058)
          R2                                                       0.288          0.294         0.327       0.289       0.295       0.328
          Observations                                             4,393          4,393         4,393       4,393      4,393        4,393

          Average outcome                                          0.712          0.712         0.712      0.712      0.712        0.712

          Geography controls                                        Y               Y             Y             Y       Y            Y
          Population 1991                                           Y               Y             Y             Y       Y            Y
          Socio-Economic Development 1991                           Y               Y             Y             Y       Y            Y
          Region FE                                                                 Y             Y                     Y            Y
          Province FE                                                                             Y                                  Y

         Notes The table presents OLS regression estimates measuring the effect of the share of refugees in 1928 on the presence
        of sport club in the early 2000s, at the community level. Small communities are those in the bottom two terciles of the
        distribution of the 1991, which corresponds to less than 1100 inhabitants in 1991. Standard errors are clustered at the
        province level. *** p <0.01, ** p <0.05, * p <0.1;
        Source: Author’s elaboration on the 1928 Greek population census and General Secretariat of Sports’ inventory.




                      Table 5: Refugee resettlement and political fragmentation

          Dep. var:                                                            Political Fragmentation Index
                                                                   2012 election                        2015 election

                                                             (1)            (2)           (3)            (4)           (5)           (6)

          Share of refugees in 1928                        0.022           -0.005        -0.012        0.060**       0.043*         0.031
                                                          (0.018)         (0.017)       (0.016)        (0.027)       (0.025)       (0.021)
          R2                                               0.165            0.307         0.373         0.318         0.444         0.550
          Observations                                      275              275           275           275           275           275

          Average outcome                                  0.788          0.788          0.788          0.689        0.689          0.689
          Region FE                                         Yes            Yes            Yes            Yes          Yes            Yes
          Geography controls and Population 2011                           Yes            Yes                         Yes            Yes
          Socio-Economic Development                                                      Yes                                        Yes
          Notes The table presents OLS regression estimates measuring the effect of the share of refugees in 1928 on the political
        fragmentation of the electorate, at the municipality level. Standard errors are clustered at the department level. *** p <0.01,
        ** p <0.05, * p <0.1;
        Source: Author’s elaboration on the 1928 Greek population census and legislative election results (Ministry of Interior).




                                                                        31
                   Table 6: Refugee resettlement and crime


 Dep. var:                                Number of criminal offences in 2015 (log)

                                            (1)         (2)          (3)          (4)

 Share of 1928 refugees in pop1928         0.516     -2.468**    -3.871***     -2.497**
                                          (1.469)     (0.974)     (1.046)       (1.103)
 R2                                        0.002       0.618       0.675         0.725
 Observations                                55          55          55            55

 Average outcome                           5.908       5.908       5.908        5.908

 Population 2011                                         Y            Y            Y
 Geography controls                                                   Y            Y
 Socio-Economic Development                                                        Y
  Notes The table presents OLS regression estimates measuring the effect of the share of
refugees in 1928 on the number of crimes at the department level. *** p <0.01, ** p <0.05,
* p <0.1; Source: Author’s elaboration on the 1928 population census and crime data from
Vasilakis (2018).




                                           32
A      Main Appendix

                                 Table A.1: Timeline of events

            May 1919     Greek Army lands in Smyrna (Izmir)

      August 27, 1922    Turkish Offence

    September 9, 1922    Turkish Army enters Smyrna

 September 13, 1922      Persecutions trigger mass exodus of Greek Orthodoxes

           April 1923    Refugee census in Greece

            July 1923    Lausanne Peace Treaty and population exchange agreement

      November 1923      The Refugee Settlement Commission (RSC) starts its operations

            May 1928     Greek population census

        January 1930     Dissolution of the RSC



Figure A.1: Share of foreign-born population by birth cohort and origin country in the
2001 census of Greece




            Notes This figure shows on the left-axis the share of individuals born in
            Turkey, by birth cohorts in the 2001 census. It shows on the right-axis the
            share of foreign-born individuals born in another country than Turkey, by
            birth cohorts in the 2001 census.
            Source: Author’s elaboration on the 2001 and 1928 Greek censuses



                                                33
Table A.2: Share of second-generation refugees in Greece (European Social Survey: rounds
2002, 2004, and 2010)

                                                 Percentage of respondents with
                                               Parents born Foreign-born parents                         Number of
                    Decade of birth             in Turkey       (outside Turkey)                        respondents

                    1910–1920                       2.82                          2.82                          71
                    1920–1930                       10.91                         4.65                         559
                    1930–1940                       14.70                         3.09                        1102
                    1940–1950                       9.18                          3.13                         991
                    1950–1960                       3.59                          1.64                         976
                    1960–1970                       1.32                          1.23                        1137
                    1970–1980                       0.36                          1.19                        1097
                    1980–1990                       0.26                          2.08                         768
                    1990–2000                       0.00                          7.58                         211
                    Notes This table shows the share of respondents with at least one parent born
                   in Turkey and with at least parent born in another foreign country other than
                   Turkey across decades of birth.
                   Source: European Social Survey, Greece sample for rounds 2002, 2004, and
                   2010.


                 Table A.3: Descriptive statistics: numbers of schools in the 1920s

                                                                          Low-refugee Province      High-refugee Province
                                                                                  (1)                        (2)                 (2)-(1)

     Number of schools in 1921 per 1000 children                                  7.335                      3.184             -4.151***
                                                                                 (0.213)                    (0.382)              (0.456)
     Number of schools in 1927 per 1000 children                                  9.149                      6.807             -2.342***
                                                                                 (0.284)                    (0.533)              (0.602)
     Number of schools built betw. 1921 and 19727                                 6.111                     26.000             19.889***
                                                                                 (0.600)                    (3.732)              (2.277)
     Number of schools built betw. 1921-19727 per 1000 children                   1.814                      3.694              1.880***
                                                                                 (0.123)                    (0.517)              (0.357)

     N                                                                             108                         31                 139
     Notes The table presents the average number of schools by provinces. Standard errors in parentheses. *** p <0.01, ** p <0.05, * p <0.1;
    Source: Author’s elaboration on yearly statistical yearbook of Greece, 1926-1927.




                                                                     34
Table A.4: Educational outcomes of Albanian immigrants relative to Greek natives in 2001

     Dep. Var:                                      Primary Education            Secondary Education               Tertiary Education
                                                      (1)        (2)              (3)           (4)               (5)             (6)

     Albanian, adult when arrived                 -0.032***     -0.032***      -0.224***        -0.224***      -0.124***        -0.124***
                                                   (0.001)       (0.001)        (0.003)          (0.003)        (0.003)          (0.003)
     Albanian, below 12 when arrived              -0.020***                    -0.253***                       -0.181***
                                                   (0.002)                      (0.006)                         (0.008)
     Albanian, 6-12 when arrived                                -0.024***                       -0.259***                       -0.179***
                                                                 (0.002)                         (0.007)                         (0.010)
     Albanian, below 5 when arrived                             -0.014***                       -0.242***                       -0.189***
                                                                 (0.003)                         (0.010)                         (0.016)
     R2                                             0.00           0.00         0.10               0.10          0.01              0.01
     N                                             558,222       558,222       538,130           538,130        468,366          468,366

     Average outcome among Greek natives             0.984        0.984          0.718            0.718          0.192             0.192
     Age and gender FE                                 Y            Y              Y                Y              Y                 Y
      Notes The table presents regression estimates measuring the educational gap between Albanian immigrants and Greek natives. The
    sample consists of 16-60 individuals in the 2001 Greek census who are either native-born or born in Albania (26,681). Albanian immi-
    grants who arrived when adults are 19,879 individuals who report to have migrated when they were older than 20. There are 4,188 and
    2,614 Albanian immigrants who arrived at an age between 6 and 12 and below 5. The sample is further resstricted to migrants who have
    eat least spend 5 years in Greece. The outcome tertiatry educaion is only defined for indivdiuals older than 25. All regressions estimates
    include both gender and age fixed-effects Notes: *** p <0.01, ** p <0.05, * p <0.1; standard errors are robust to heteroskedasticity.
    Source: Author’s elaboration on the 2001 Greek census.




                                                                      35
Table A.5: Occupations and income of second-generation refugee, Albanian immigrants
and Greek natives


      Panel A: Second-generation refugees and natives (born in 1923-1950)

                                           Natives    Refugees     Difference         Difference
                                            (1)         (2)         (2)-(1)      With age & gender FE
     Occupation in last or current job:
     Farmer                                 0.420       0.327       -0.092***           -0.105***
                                           (0.014)     (0.031)       (0.035)             (0.035)
     Low-skill non-farmer occupation        0.127       0.159          0.032               0.025
                                           (0.009)     (0.024)       (0.024)             (0.024)
     Skilled occupation                     0.166       0.128         -0.037              -0.028
                                           (0.010)     (0.022)       (0.026)             (0.026)
     Deciles of household’s income          3.748       3.665         -0.083              -0.099
                                           (0.050)     (0.134)       (0.134)             (0.130)

     N                                      1,256         226         1,482               1,482


      Panel B: Albanian immigrants and natives (all ages)

                                           Natives    Albanians                 Differences
                                            (1)          (2)         (2)-(1)     With age & gender FE
     Occupation in last or current job:
     Farmer                                 0.230       0.091       -0.139***              0.008
                                           (0.007)     (0.025)        (0.037)             (0.035)
     Low-skill non-farmer occupation        0.103       0.318        0.215***            0.239***
                                           (0.005)     (0.041)        (0.028)             (0.028)
     Skilled occupation                     0.313       0.038       -0.275***           -0.383***
                                           (0.008)     (0.017)        (0.041)             (0.040)
     Deciles of household’s income          4.313       3.731       -0.582***           -1.111***
                                           (0.036)     (0.154)        (0.198)             (0.194)

     N                                      3,378         132         3,510               3,509
      Notes. The table presents the average occupation in the current or last job, and the self-declared
    household income in brackets corresponding to deciles. In the fourth column of Panel A, the table
    reports regression estimates of the gap in the outcome between natives and refugee (or between natives
    and Albanians in Panel B) after controlling for age and gender dummies. Standard errors in parenthesis;
    *** p <0.01, ** p <0.05, * p <0.1.
    Source: Author’s elaboration on the European Social Survey.




                                                     36
Table A.6: Social integration of Albanian immigrants arrived in Greece at different age


   Albania-born arrived in Greece when                                                  Young (<24)                  Older(≥24)           Difference
                                                                                            (1)                          (2)               (1)-(2)

   Greek citizen                                                                            0.149                       0.111               0.038
                                                                                           (0.044)                     (0.033)             (0.054)
   Ever feel discriminated against                                                          0.522                       0.489               0.033
                                                                                           (0.061)                     (0.053)             (0.081)
   Greek as language most often spoken at home                                              0.507                       0.567               -0.059
                                                                                           (0.062)                     (0.053)             (0.081)
   Political activism (petition, boycott, demonstration...)                                 0.045                       0.056               -0.011
                                                                                           (0.026)                     (0.025)             (0.036)
   Participate in civic organisation                                                        0.077                       0.080               -0.003
                                                                                           (0.077)                     (0.055)             (0.095)
   N                                                                                          67                          90                 157
   This table shows average attitudes and behaviors of Albanian immigrants arrived in Greece at different age. Standard
  errors in parentheses. *** p <0.01, ** p <0.05, * p <0.1;
  Source: Author’s elaboration on the European Social Survey (ESS).


        Table A.7: Robustness check: Sport clubs and share of refugees in 1923

         Dep. var:                                                          Presence of at least one sport club

                                                                    (1)           (2)           (3)            (4)             (5)         (6)

         Share of refugees in 1923                                0.066       0.120*        0.078           -0.015            0.039       0.023
                                                                 (0.051)      (0.062)      (0.070)         (0.049)           (0.068)     (0.075)
         Share of refugees in 1923 * small community                                                      0.266***          0.265***     0.187*
                                                                                                           (0.100)           (0.097)     (0.100)
         R2                                                        0.288       0.295          0.327          0.288            0.295       0.328
         Observations                                              4,393       4,393          4,393         4,393             4,393       4,393

         Average outcome                                           0.712       0.712          0.712           0.712           0.712      0.712

         Geography controls                                          Y              Y           Y               Y              Y           Y
         Population 1991                                             Y              Y           Y               Y              Y           Y
         Socio-Economic Development 1991                             Y              Y           Y               Y              Y           Y
         Region FE                                                                  Y           Y                              Y           Y
         Province FE                                                                            Y                                          Y
         Notes The table presents OLS regression estimates measuring the effect of the share of refugees in 1923 on the presence
        of sport club in the early 2000s, at the community level. Standard errors are clustered at the province level. *** p <0.01,
        ** p <0.05, * p <0.1;
        Source: Author’s elaboration on the 1928 population census, 1923 refugee census and General Secretariat of Sports’ inven-
        tory.



                             Table A.8: Sport clubs and Albanians immigrants

         Dep. var:                                                                       Presence of at least one sport club

                                                                              (1)         (2)           (3)           (4)        (5)       (6)

         Share of Albanian immigrants in 2001                               0.123        0.135         0.131          0.134     0.141     0.050
                                                                           (0.143)      (0.140)       (0.148)       (0.098)    (0.090)   (0.099)
         Share of Albanian immigrants in 2001 * small community                                                      -0.015     -0.008    0.112
                                                                                                                    (0.188)    (0.185)   (0.188)
         R2                                                                 0.288       0.295         0.328           0.288     0.295     0.328
         Observations                                                       4,393       4,393         4,393           4,393     4,393     4,393

         Geography controls                                                   Y           Y             Y             Y            Y        Y
         Population 1991                                                      Y           Y             Y             Y            Y        Y
         Socio-Economic Development 1991                                      Y           Y             Y             Y            Y        Y
         Region FE                                                                        Y             Y                          Y        Y
         Province FE                                                                                    Y                                   Y
         Notes The table presents OLS regression estimates measuring the joint effect of the share of refugees in 1928 and the share of
        Albanian immigrants in 2001 on the presence of sport club in the early 2000s, at the community level . Standard errors are clustered
        at the province level. *** p <0.01, ** p <0.05, * p <0.1;
        Source: Author’s elaboration on the 1928 Greek population census and General Secretariat of Sports’ inventory.




                                                                         37
                 Table A.9: Refugee resettlement and political fragmentation between the rightist and leftist
                 parties

                                     Dep. var:                                               Political Fragmentation Index between Rightist and Leftist Parties
                                                                                                    2012 election                     2015 election

                                                                                               (1)           (2)        (3)             (4)              (5)              (6)

                                     Share of refugees in 1928                                0.034       0.029        0.021         0.023           0.040*              0.031
                                                                                             (0.024)     (0.023)      (0.018)       (0.024)          (0.023)            (0.021)
                                     R2                                                       0.182       0.261        0.315         0.250            0.322              0.366
                                     Observations                                              275         275          275           275              275                275
                                     Average outcome                                          0.494       0.494        0.494         0.481            0.481              0.481

                                     Region FE                                                 Yes           Yes        Yes             Yes              Yes              Yes
                                     Geography controls and Population 2011                                  Yes        Yes                              Yes              Yes
                                     Socio-Economic Development                                                         Yes                                               Yes
                                     Notes The table presents OLS regression estimates measuring the effect of the share of refugees in 1928 on the political
                                   fragmentation of the electorate, at the municipality level. Standard errors are clustered at the department level. *** p <0.01,
                                   ** p <0.05, * p <0.1;
                                   Source: Author’s elaboration on the 1928 Greek population census and legislative election results (Ministry of Interior).




                                                     Table A.10: Robustness check: Political fragmentation

 Dep. var:                                                                                                         Political Fragmentation Index
                                                                             2012 election                                                                         2015 election
                                                    (1)        (2)        (3)       (4)                (5)           (6)          (7)              (8)            (9)        (10)             (11)           (12)

 Log Share of refugees in 1928                   0.003**      0.000      -0.001                                                0.008***         0.005**          0.002
                                                 (0.001)     (0.002)    (0.002)                                                 (0.002)         (0.002)         (0.002)
 Share of refugees in 1928                                                            0.083            0.004        -0.014                                                  0.178**          0.098           0.057
                                                                                    (0.052)          (0.058)       (0.054)                                                   (0.069)        (0.072)         (0.060)
 Share of refugees in 1928, square                                                   -0.099           -0.013         0.003                                                  -0.192**         -0.081          -0.038
                                                                                    (0.065)          (0.070)       (0.065)                                                   (0.078)        (0.078)         (0.066)
 R2                                               0.189       0.307      0.372        0.181            0.307         0.373       0.356           0.448           0.546        0.341           0.447           0.550
 Observations                                      275         275        275          275              275           275         275             275             275          275             275             275

 Region FE                                          Yes        Yes        Yes          Yes             Yes           Yes          Yes              Yes            Yes           Yes            Yes            Yes
 Geography controls and Population 2011                        Yes        Yes                          Yes           Yes                           Yes            Yes                          Yes            Yes
 Socio-Economic Development                                               Y es                                       Yes                                          Yes                                         Yes

 Dep. var:                                                                                    Political Fragmentation Index between Rightist and Leftist Parties
                                                                             2012 election                                                                  2015 election
                                                    (1)        (2)        (3)       (4)               (5)          (6)        (7)            (8)           (9)        (10)                    (11)           (12)

 Log Share of 1928 refugees in pop1928           0.002*       0.002      -0.000                                                  0.001           0.002           0.001
                                                 (0.001)     (0.001)    (0.001)                                                 (0.001)         (0.002)         (0.001)
 Share of 1928 refugees in pop1928                                                    0.098            0.089         0.066                                                    0.056          0.097           0.078
                                                                                    (0.065)          (0.066)       (0.056)                                                   (0.065)        (0.065)         (0.056)
 Share of 1928 refugees, Square                                                      -0.104           -0.090        -0.066                                                    -0.054         -0.085          -0.063
                                                                                    (0.071)          (0.070)       (0.061)                                                   (0.070)        (0.070)         (0.062)
 R2                                               0.175       0.254      0.310        0.194            0.268         0.319       0.245           0.313           0.356        0.253           0.327           0.369
 Observations                                      275         275        275          275              275           275         275             275             275          275             275             275

 Region FE                                          Yes        Yes        Yes          Yes             Yes           Yes          Yes              Yes            Yes           Yes            Yes            Yes
 Geography controls and Population 2011                        Yes        Yes                          Yes           Yes                           Yes            Yes                          Yes            Yes
 Socio-Economic Development                                               Y es                                       Yes                                          Yes                                         Yes
 Notes The table presents OLS regression estimates measuring the effect of the share of refugees in 1928 on the political fragmentation of the electorate, at the municipality level. Standard errors are clustered at
the department level. *** p <0.01, ** p <0.05, * p <0.1;
Source: Author’s elaboration on the 1928 Greek population census and legislative election results (Ministry of Interior).




                                                                                                        38
B    Research Design Appendix

                     Figure A.2: Description of the 1928 census: examples
                                           (a) Questionnaire




                                        (b) Municipal-level data




Notes. Data source: The 1928 population census of the Kingdom of Greece.



Identification of equation 1 We document that places of resettlement were neither more
prosperous nor growing faster before the population exchange. As presented in Table A.11,

                                                  39
we find no statically significant difference in the pre-settlement population density and ur-
banization between municipalities with varying shares of refugees in 1928, conditional on
geographic features. Moreover, Figure A.3 shows that, conditional on geographic charac-
teristics and province fixed effects, there is no statistically significant correlation between
pre-settlement population growth, i.e., between 1907/1913 and 1920, and the distribution
of refugees across communities.           31   While we cannot explore this question any further at
the community level due to lack of data, we can leverage richer data at the province-level.
Table A.12 shows that provinces that hosted more refugees in 1928 were less developed
in 1920, conditional on region fixed effects and on the covariates displayed in Table A.11.
The share of refugees in 1928 is correlated with a lower share of employment in the non-
agricultural sector and in manufacturing and with a lower literacy rate in 1920. We also
examine pre-settlement trends in the literacy of the 1920 population. Figure A.4 shows
that the increase in literacy rates across cohorts born between 1861 and 1905 follows a
very similar pattern between high- and low-refugee provinces. Reassuringly, a statistical
test of parallel pre-trends cannot be rejected, which suggests that places of settlement were
not growing faster prior to refugees’ arrival.




   31
      Greece conducted a population census in 1907 in all of its territories at that time, which comprised Central
Greece, Peloponnese, Thessaly, Ionian Islands, and Cyclades Islands. After the annexation of Macedonia,
Epirus, Crete, and Aegean Islands in 1913, an additional population census, similar to the one in 1907, was
conducted in these regions. Population data for 1907/1913 is not available for Thrace as it joined Greece only
after WWI.


                                                       40
Table A.11: Determinants of refugee settlement at the community level

       Outcome:                                        Share of refugees
                                                      in 1923      in 1928
                                                        (1)          (2)

       Log population density in 1920                  0.730          0.212
                                                      (0.717)       (0.370)
       Dummy: city in 1920                           10.476***      3.718*
                                                      (1.426)       (2.193)
       Dummy: island                                 24.736***     4.644***
                                                      (1.212)       (1.098)
       Distance to railway network in 1920 (km)        -0.012        -0.047
                                                      (0.022)       (0.033)
       Distance to shoreline (km)                    -0.082***       -0.005
                                                      (0.020)       (0.049)
       Distance to border (km)                         -0.009        -0.080
                                                      (0.017)       (0.062)
       Altitude (m)                                    -0.003        -0.003
                                                      (0.002)       (0.003)
       Mean annual precipitation                       0.003         -0.013
                                                      (0.007)       (0.012)
       Mean annual temperature                        -0.316*         0.211
                                                      (0.168)       (0.369)
       Crop suitability score                         1.251**      0.996**
                                                      (0.583)       (0.443)
       N                                               4,388          4,388
       R2                                               0.54           0.60

       Province FE                                        Y            Y

       Average of outcome                               4.19         7.05
       Notes. This table examines the determinants of the share of refugees
      in 1923 and 1928 and the change in the share of refugees between
      the two years across municipalities. The unit of analysis is the mu-
      nicipality as defined by the 1920 population census of the Kingdom
      of Greece. All specifications controls for province fixed effects and a
      set of municipal-level covariates, which comprises: log population den-
      sity in 1920; dummy variables indicating whether the municipality was
      a city in 1920 and whether it is an island; distances to railway net-
      work in 1920, shoreline, and national borders; altitude, mean annual
      precipitation and temperature, and crop suitability of the municipality.
      * p < 0.10, ** p < 0.05, *** p < 0.01.




                                        41
  Table A.12: Determinants of refugee settlement at the province level

 Outcome:                                                 Share of refugees in 1928
                                                        (1)          (2)          (3)
 Share of employment in non-agriculture in 1920      -0.198**
                                                      (0.091)
 Share of employment in manufacture in 1920                      -0.468***
                                                                   (0.163)
 Literacy rate in 1920 (15-65 year olds)                                       -0.368**
                                                                                (0.153)
 N                                                      176          176          176
 R2                                                     0.73         0.74        0.74

 Region FE                                               Y           Y            Y
 Controls                                                Y           Y            Y

 Average outcome                                       12.99       12.99        12.99
 Notes. This table examines the determinants of the distribution of refugees in 1928
across provinces. All specifications control for region fixed effects (for the following 10
regions of Greece: Crete, Peloponnese, Central Greece and Euboea, Cyclades Islands,
Aegean Islands, Ionian Islands, Epirus, Thessaly, Macedonia and Western Thrace) and
a set of province-level covariates, which comprises: log population density in 1920;
dummy variables indicating whether the observation unit was a city in 1920 and whether
it is an island; the distances to railway network in 1920, to shoreline, and to national
borders; altitude, mean annual precipitation and temperature, and crop suitability of
the province. Heteroskedasticity-robust standard errors are reported in parentheses.
* p < 0.10, ** p < 0.05, *** p < 0.01.




                                           42
Figure A.3: Correlation between the share of refugees and the population growth prior to
their arrival
                                                                    (a) Share of refugees in 1928




                           100         50
                Share of refugees in 1928
               -50          0
                           -100




                                            -100                 -50               0               50               100
                                                              Population growth between 1907/13 and 1920 (%)
                                             coef = .069, (robust) se = .047, t = 1.47, N = 1405, R-squared = .58



                                                                    (b) Share of refugees in 1923
                           100         50
                Share of refugees in 1923
               -50          0
                           -100




                                            -100                 -50               0               50               100
                                                              Population growth between 1907/13 and 1920 (%)
                                             coef = .002, (robust) se = .015, t = .14, N = 1405, R-squared = .51



        Notes. This figure presents the correlation between the share of refugees (in 1928 in
        panel (a) and in 1923 (in panel (b)) and the pre-treatment population growth, i.e., pop-
        ulation growth between 1907/1913 and 1920, before the arrival of refugees, controlling
        for province fixed effects and a set of municipal-level covariates, which comprises: log
        population density in 1907/1913; dummy variables indicating whether the municipality
        was a city in 1907/1913 and whether it is an island; distances to railway network, to
        shoreline, and to national borders; altitude, mean annual precipitation and temperature,
        and crop suitability of the municipality. The sample excludes outliers, defined as the top
        and bottom 2.5% of municipalities in terms of population growth between 1907/1913
        and 1920.. Data sources: The 1928 population census of the Kingdom of Greece.




                                                                                       43
                          Figure A.4: Pre-trends in literacy rates in 1920




Notes. This figure presents the average differences in trends in literacy rates across birth cohorts observed in
1920, i.e., before the arrival of refugees, and across provinces that received varying levels of refugees. The
figure plots the average literacy rate in 1920 by birth cohorts within high-refugee and low-refugee provinces
(defined as above and below of the median of the share of refugees in 1928, respectively).




                                                     44
Figure A.5: Distribution of refugees in 1923 and correlation with the share of refugees in
1928
                                                 (a) Share of refugees in 1923

                                                               Bulgaria
                                             Macedonia
                                             (FYROM)


                              Albania




                                                                                               Turkey




                     Share of refugees in 1923
                          0
                          (0, 2]
                          (2, 5]
                          (5, 10]
                          (10, 20]
                          (20, 30]
                          (30, 40]
                          (40, 60]
                          (60, 80]




                                                                                                    Ü
                          (80, 84.1]
                          No data                                  100    50   0   100 Kilometers




                                    (b) Correlation with the share of refugees in 1928




        Notes. Panel (a) of this figure visualizes the share of refugees in 1923 at the community
        level. Panel (b) presents the correlation between the share of refugees in 1928 and in
        1923 (in panel (b)) controlling for province fixed effects and a set of municipal-level
        covariates, which comprises: log population density in 1907/1913; dummy variables
        indicating whether the municipality was a city in 1907/1913 and whether it is an is-
        land; distances to railway network, to shoreline, and to national borders; altitude, mean
        annual precipitation and temperature, and crop suitability of the municipality. Data
        sources: The 1923 refugee census and the 1928 population census of the Kingdom of
        Greece.




                                                             45