Migration, Cumulative Causation and Gender:

Evidence from Thailand

Sara R. Curran
Princeton University
United States of America

Filiz Garip
Princeton University
United States of America

Chang Chung
Princeton University
United States of America

Kanchana Tangchonlatip
Mahidol University
Thialand

Abstract:

In our study we examined longitudinal data from Thailand to replicate studies of cumulative causation and
to extend our theoretical understanding of the concept by examining the gendered content of migration
networks at multiple levels of aggregation. Our research and previous research shows that migration
processes are gendered, but there has been little empirical evaluation of how cumulative causation may
differentially influence men and women or how the gendered content of different kinds of migrant
accumulated experiences might be more or less important for predicting migration behaviors. In this study
we find that cumulative causation is also important for explaining rural-urban migration in the Thai
context, net of other factors, and that there are important gender dimensions related to those effects.

Acknowledgements:

The authors are grateful for the support of the Program in Urbanization and Migration at Princeton
University for a research fellowship to Tangchonlatip and to the Center for Migration and Development for
summer research support to Curran, Garip, and Chung. The research could not have been completed
without the data collection efforts of the research team from the Carolina Population Center at the
University of North Carolina and the Institute for Population and Social Research at Mahidol University
and the cooperation of the villagers of Nang Rong district, Buriram province, Thailand. We are extremely
grateful to all of them for their work and contribution.

Paper prepared for Conference on African Migration in Comparative Perspective, Johannesburg, South
Africa, 4-7 June, 2003.

0
Introduction
The idea that migrant networks can evolve, accumulate, and generate higher than
expected levels of migration out of communities of origin has yielded a considerable
number of empirical studies and some policy attention (Massey 1990a; Massey 1990b;
Massey and Garca-Espaa 1987; Massey, Goldring, and Durand 1994; Massey and
Zenteno 1999). This idea, cumulative causation, describes a process by which migration
propensities, among those who are in origin communities (whether they are return
migrants or otherwise), grow with each additional migrant in a migrant stream. Further,
the theory holds that the importance of other factors predicting migration propensities
becomes less important in a context of high levels of migration (Massey 1990a; Massey
1990b; Massey and Garca-Espaa 1987; Massey et al. 1994; Massey and Zenteno 1999).
Much of this research focuses upon Mexico-U.S. migration patterns and Mexican
migration behaviors (Davis, Stecklov, and Winters 2002; Espinoza and Massey 1999;
Massey et al. 1994; Massey and Zenteno 1999; Massey and Espinoza 1997; Winters, de
Janvry, and Sadoulet 2001). Because of the data requirements necessary to test the
theory of cumulative causation extensions to other settings outside of the Mexico-U.S.
case have not been undertaken.
In our study we examined longitudinal data from Thailand to replicate studies of
cumulative causation and to extend our theoretical understanding of the concept by
examining the gendered content of migration networks at multiple levels of aggregation.
Our research and previous research shows that migration processes are gendered, but
there has been little empirical evaluation of how cumulative causation may differentially
influence men and women or how the gendered content of different kinds of migrant
accumulated experiences might be more or less important for predicting migration
behaviors. Our study is also different from previous studies evaluating the effect of
cumulative causation, not only because of the setting, but also because our analysis
concerns internal migration, mostly rural-urban migration. Previous work has suggested
that cumulative causation will be less meaningful for internal migration processes(Taylor
1986), although recent analyses suggests that gender may explain differences in the
effects of migrant networks for internal and international migration(Curran and Rivero
2003). In this study we find that cumulative causation is also important for explaining

1
rural-urban migration in the Thai context, net of other factors, and that there are
important gender dimensions related to those effects.
In what follows, we provide a background review of the literature describing our
theoretical framework and related empirical evidence. We also outline the qualitative
evidence for why it is important to account for the content of migrant accumulated
experiences, in our case the gendered content of those experiences. We conclude our
review of the literature with a discussion of the Thai case and why it provides an
interesting opportunity to consider the role of cumulative causation and the importance of
knowing the gender content of migrant accumulated experiences for explaining migration
patterns.
Focusing upon gender as a critical aspect describing the content of migrant
networks is important because of what is assumed about the way cumulative causation is
understood to affect individual propensities to migrate and the overall rate of migration.
Migrant networks and accumulated migrant experiences decrease the costs of migration,
or increase the demonstrative effects (diminishing psychic costs or familial resistance), or
increase information about and access to labor market opportunities. If men and women
live the migration experience differently, then the magnitude of the decrease in costs, the
demonstrative effects, or the amount of information and access to labor markets will vary
depending on the gender content of the accumulated experiences. In turn, this variation
may yield gendered migration outcomes which, given the momentum implied by
cumulative causation, may yield profound implications for the social organization of life
in places of origin and destination.
In origin communities, this could influence the organization of care work
provided to the elderly or children, or the redistribution of wealth (shifting inheritance
preferences away from daughters in the case of Thailand, for example) or investments
(abandonment of upland agricultural plots that require significant male labor). If male
and female migrants behave significantly differently with regards to maintaining contact
with places of origin (either through remittances or visits), then a gender difference in
migration rates and accumulated migration experience may also affect the amount and
level of resource flows directed by migrants back to their villages of origin.

2
 Alternatively, destination communities, if they are defined significantly by
gendered migration patterns, may also be transformed. If, for example, males
predominate then the organization of domestic work may be affected. In some cases men
may take on household work (Hondagneu-Sotelo 1994) or may outsource the work to
entrepreneurs, if they can afford to do so. Or, if females predominate and establish a
sector and occupational niche for themselves, they may affect a redefinition of womens
and mens work. For example, in the case of Hong Kong, Filipina migrant predominance
in domestic work has meant that Filipino male migrants following their female relatives
are more likely channeled into domestic service, an occupation they would otherwise not
have joined, either in other migrant destinations or in the Philippines (Constable 1997;
Tyner 1996). Although we do not quantitatively evaluate these particular possibilities in
this paper, we offer them as justification for why it is important to examine the influence
of gender composition of migrant networks upon migrant behavior.
Our study, instead, examined 10-year retrospective longitudinal data from twenty-
two villages in one district in Northeastern Thailand to evaluate the influence of
cumulative trips and experience upon the probability of being a migrant at any point in
time over the ten years, net of alternative explanations. We evaluate gender differences
in three ways. First, we compare the differential effect of number of migrant trips and
migrant experience upon mens and womens migration propensities. Second, we
evaluate the effect of the number of female trips and experience and the number male
trips and experience, net of overall migration prevalence upon anyones migration
propensities. And, third, we evaluate whether these gendered networks matter more or
less for womens and mens migration propensities. In our discussion of results, we
provide some insights from our qualitative fieldwork about the possible effects of
gendered migration patterns upon the reorganization of social life in origin villages.

Background
To our knowledge there are very few studies that have quantitatively evaluated
the role of gender content of accumulated migrant experiences, but numerous review
articles and ethnographic research have pointed to the importance of considering gender
relations when studying migration processes (Chant 1992; Grieco and Boyd 1990;

3
Hondagneu-Sotelo 1994; Pessar 1999a; Pessar 1999b; Tienda and Booth 1991). Before
discussing studies of gender, migration and cumulative causation, we briefly describe the
concept of cumulative causation and its effect upon migration behavior.
Social networks, or social capital, in relation to migration are commonly
understood as the links between residents in a community of origin and individuals living
in another place, or with individuals who previously migrated regardless of their current
residence (Hugo 1991; Massey and Garca-Espaa 1987) (Massey 1990). These ties to
migrants are incorporated in most theories used to explain international migration
including world systems theory (Portes 1978), dual labor market theory(Piore 1979),
social capital(Massey 1990) and new economics models (Stark 1991; Stark and Levhari
1982) (Taylor 1986). The concept of networks for explaining migration has strong
theoretical and intuitive purchase, as well as empirical support. Typically, most studies
measure migrant networks as simply counts of other people who have already migrated
from a common social unit, either a village or family. The network tie is presumed based
on the common social unit. Whether or not prior migrants actually participate in a
network of relationships with the members of the social unit of origin is usually not
quantified, although there is plenty of ethnographic evidence to suggest that on the whole
a vast majority of migrants do participate in these networks. Therefore, we use the terms
migrant network and migrant social capital interchangeably, reflecting the current
literature and the presumption of ours and others measures of accumulated migrant trips
and experiences.
Social networks facilitate migration in several ways. Contact with migrants
shows individuals they may be better off in a place other than their current residence
(Hugo 1991). Migrant networks reduce travel costs through information on safe and
cheap routes or smugglers and reduce emotional costs. Migrant networks lessen
assimilation shock if immigrants arrive in an environment where others speak their
language (Choldin 1973) and where living among other foreigners can easily prevent
deportation (Massey 1990). Migrant networks increase the expected benefits of
migration when contact with previous migrants helps in the job search process. Finally,
migrant networks also help to save and reduce living expenses and provide financial
assistance upon arrival.

4
 Empirical evidence supporting the importance of migrant networks for predicting
migration and for influencing a dynamic rate of migration are numerous and mostly focus
upon Mexican-U.S. migration (Davis et al. 2002; Espinoza and Massey 1999; Massey et
al. 1994; Massey and Zenteno 1999; Massey and Espinoza 1997; Winters et al. 2001).
Qualitative evidence from numerous settings throughout the world further support these
findings (see for example: (Arizpe 1975; Arizpe 1980; Arizpe 1981; Arizpe 1985;
Menjivar 1995; Menjvar ; Menjvar 2000; Tyner 1996)). None of the quantitative
studies, however, systematically distinguish between individual level accumulated
experiences, family migrant accumulated experiences, and community migrant
accumulated experiences. There is some suggestion in the qualitative studies that there
are differences in the kinds of information available from a personal experience versus
those from family members versus those from a wider array of community members.
Not only might information vary, but also the quality of the social capital will be
different. Trust may distinguish social capital quality between family and community
networks, but community networks may yield greater amounts and a wider array of
information. In our study we have distinguish migrant social capital along these three
levels of aggregation.
Related to the quality of information or the resources available through migrant
networks to those that remain behind in origin villages, is the frequency of trips and the
accumulated length of time in destinations. The number of preceding trips may indicate
greater ties to a destination, but not necessarily high quality information. The length of
time migrants spend in a destination may increase the quality of information and the array
of resources they may be able to offer potential migrants, net of the number of trips made
by migrants. The return flow of both amount and quality of information will depend on
the nature of the ties between the migrant in a place of destination and the migrants
home village or community. In many different contexts these ties have been shown to be
differentiated by gender (Curran and Saguy 2001). In some cases, female migrants may
be more likely to maintain ties to places of origin (Curran and Saguy 2001) and in other
cases men may be more likely to do so (Grasmuck and Pessar 1991). To date, we know
of no studies that have systematically disaggregated by sex the effects of trips and
experience upon migration propensities. We suggest that in particular contexts gender

5
may serve as a proxy for measuring the quality of migrant social capital or migrant
networks. In Thailand, as we describe in the next section, men and women have very
different social ties to their villages and natal households. These relationships may
systematically affect the content of migrant networks or the quality of migrant social
capital and thereby differently influence migrant behavior.
Similarly, systematic analyses of the differences between how migrant social
capital might differentially affect mens and womens migration behavior have been
limited. Numerous studies have suggested that the factors affecting mens and womens
migration differ and that men and women live the migration process differently. If
womens migration is perceived as fraught with more risks, then several effects are
anticipated. First, family migrant networks may be more important for ensuring high
quality information and diminishing the risks associated with migration. Second, women
need a greater accumulated social capital at the village or community to overcome
barriers to migration. So, womens migration increases after the migration process has
developed significantly (Boyd 1989). Or, womens migrant networks may be very
different and offer significantly different information thereby differentially affecting both
mens and womens propensities to migrate (Curran and Rivero 2003; Hondagneu-Sotelo
1994; Tyner 1996). Despite fairly strong expectations about differences, few quantitative
studies have examined whether migrant social capital matters more or less to men or
women and all of these studies have focused upon the Mexican-U.S. migration case.
Kanaiaupuni (2000) analyzes womens and mens first migration and finds that in
a constrained model, family networks matter more for womens first move than for
mens. Further, village networks appear to have a stronger effect upon mens migration
than upon womens. And, finally, that the sex composition (a greater proportion of
women in a network) positively affects womens first move, but discourages mens.
However, with time the positive effect of sex composition disappears for women,
although the significant discouraging effect remains for men(Kanaiaupuni 2000).
Cerutti and Massey (2001) find that migration prevalence in the community has
little effect on husbands or wives migration, but does significantly influence an eldest
sons migration (and not the eldest daughters). Family moves are more important for
predicting migration than are community moves, in general. Sons and daughters prior

6
moves raise the probability of both migrating and husbands are significantly more
influenced by their childrens moves than are wives. Cerruti and Massey (2001) also
find that a mothers prior migration increases the probability of both a daughter and son
migrating, but has a greater effect upon a daughters migration. A fathers prior
migration significantly affects a sons migration, but not a daughters.
Curran and Rivero (2003) find that the prior internal migration of women from a
household facilitates the migration of both men and women, but international migration
of female household members only facilitates the migration of other women, not men.
Mens internal migration has no influence upon either mens or womens migration to
internal destinations, but has a significantly greater influence upon mens international
migration probabilities than upon womens(Curran and Rivero 2003).
In all of the above cases, the empirical evidence shows a strongly gendered story
about the migration process, which is supported by numerous ethnographic accounts.
Kanaiaupunis (2000) analysis comes closest to our analysis by taking into both family
and community networks. We extend these analyses of the migration process, by
analyzing the extent to which cumulative migration at the individual, household, and
village level is differentially important for men and women. Further, we examine the
extent to which the gendered content of family and village trips and accumulated
experience, net of overall migration prevalence, might affect any migration, as well as
mens and womens migration. Despite the fact that our analysis is conducted using Thai
data, we find surprisingly similar results. Our finding also offer further insights on how
distinguishing between different types of migrant social capital illuminates the dynamics
of the migration process. The Thai case offers an interesting opportunity to test a theory
in a different setting where there are very different gender dynamics. The data and
setting provide an opportunity for generalizations and further refinement of how migrant
networks influence migration outcomes.

Thailand
Although migration from rural Thailand to urban Bangkok may not seem as
dramatic as a move from rural Mexico to the United States, for many rural Thais during
the mid-1980s it was. Although the dangers of the trip are not as dramatic or deadly as

7
those for Mexicans crossing the border illegally, the experience in the place of destination
in unsafe work environments and poor living quarters was similar. During our field
work, rural Thai villagers told tales of failed migrants who had returned home and told of
not being paid for weeks of work, of living in six foot by six foot cubicles, of being
physically mistreated, and of losing all their money to drinking and gambling. Even
more devastating for parents are the children who migrate and are never heard from
again. During our fieldwork we learned of one such case and heard stories of others.
Being able to trust someone and distinguishing between good information about jobs
from acquaintances versus close friends and family were critical for determining whether
a migrant trip would be worthwhile.
Seasonal migration is not unheard of in Thailand, especially in the Northeastern
region (the site for these data) as the monsoon rains are often preceded by lengthy
droughts that have often required the movement of people to find alternative livelihoods,
at least for a few months (Nartsupha 1999; Phongphit 1990). However, migration took
on added significance in Thai livelihoods from the mid 1980s onward. It was then that
Thailands shift from an agriculture-based export economy to a manufacture-based export
economy took place (Bello, Cunningham, and Poh 1998; Phongpaichit 1980;
Phongpaichit and Baker 1996; Phongpaichit and Baker 1998; Warr 1993; Warr and
Nidhiprabha 1996). From the mid-1980s to the mid-1990s some experts estimate that
Thailands economy grew on average 10 percent per year (Bello et al. 1998; Warr and
Nidhiprabha 1996). Much of this growth fueled by production in export manufacturing,
which was a result of the rising value of the Yen, rising wages in nearby newly
industrialized countries (NICs), changes in textile import quotas to the United States, and
dramatic increases in foreign direct investment, primarily from Japan (Nidhiprabha 1994;
Phongphaichit and Baker 1998). By 1985 Thai manufacturing exports had outpaced rice
and other agricultural exports in value (Nidhiprabha 1994; Warr and Nidhiprabha 1996).
With the growth in manufacturing export came an increased demand for labor. Much of
this labor was met by rural migrants, many from the Northeastern part of the country,
many of them young, and many of them women (Chamratrithirong et al. 1995; Mills
1997; Phongpaichit and Baker 1996). According to the 1992 National Migration Survey
most migrants to the Bangkok metropolitan area were in their teens or early twenties and

8
at least half of these migrants were women (Chamratrithirong et al. 1995). These
women were not moving to urban areas for marriage, theirs were economically motivated
moves, primarily to help support their family of origins household economy (De Jong,
Richter, and Isarabhakdi 1996).
Thai rural women have long played an important role in household
economies(Singhanetra-Renard and Prabhudhanitisarn 1992). They work next to their
husbands and brothers in the rice fields and are often described as holding the purse
strings with regards to financial planning (Yoddumnern-Attig 1992). Also many
historic and ethnographic studies describe womens relations with their husbands as
egalitarian (Knodel, Chamratrithirong, and Debavalya 1987). Womens participation in
rural-urban migrant streams is considerable, reaching as high as 60 % of all migrants
(Tantiwiramanond 1995; Chamratrithirong et al.1995). These rates are only surpassed in
Asia by the migration rates of women from the Philippines and Japan (Tantiwiramanond
1995). It is important to note that these moves are rarely associational (family moves)
but primarily for jobs for the women themselves (Chamratrithirong et al. 1995).
These positive characterizations of womens status are increasingly questioned by
recent scholarship. These research note that women predominate in the low wage, low
skill sectors of the economy, including low wage service jobs, prostitution, agricultural
wage labor, and low skill manufacturing (like textiles, parts assembly for electronics, and
food processing plants) (Sussangkarn 1993; Tantiwiramanond 1995). Importantly women
consistently earn one-third to one-half as much as men in similar occupations
(Phananiramai 1993; Richter and Havanon 1994; Tantiwiramanond 1995). Outside of
seasonal construction labor, much of the destination labor market is sex segregated. Men
tend to work in heavier industries, taxi driving and motorcycle services, automobile
servicing, and construction (Sussangkarn 1993). Hence, the kinds of information migrant
men can provide for women may not be as helpful to women as the information from
migrant women and vice versa.
Further, patterns of marriage and settlement adjudicate against strong household ties
with sons and tend towards strong ties with daughters, especially in Northeastern Thailand
(Blanc-Szanton 1990; De Jong et al. 1996; Yoddumnern-Attig 1992). Thai households are
matrilocal. That is, a husband moves to the wife's family's house for 2 days to 5 years (usually

9
until the first child is one year old). In the ideal situation the husband provides an important
source of farm labor. The inheritance norm is bilateral, but sons usually sell their rights to
their sisters or brothers-in-law since they will be moving to their wife's household. Youngest
daughters are particularly advantaged with regards to household resources because they often
inherit the homestead and a larger portion of the land. Access to these resources comes at a
price. These daughters and their husbands are expected to care for the daughter's parents in
their old age (Keyes 1984; Yoddumnern-Attig 1992). On the other hand, sons are encouraged
to develop outside, non-kin based networks and daughters are socialized to cultivate their
kinship ties. For men, this cultural emphasis has lead to an interesting phenomenon of patron-
client relations that crosscut kinship ties (Hanks and Hanks 1963; Limpinuntana et al. 1982
). The non-kin based network for men provides access to political patronage, jobs, and
resources.
These gendered patterns of household and family relations also suggest different
ways in which the social capital of migrant networks might influence migration
probabilities. Different access to different types of labor markets may be more or less
beneficial to members of the opposite sex. For example, migrant trips by women for
clearly advertised female export manufacturing jobs may be less influential upon mens
migration probabilities, especially if recruitment networks are highly gendered. During
previous fieldwork we found that factory managers often used their factory workers as
recruiters back in their home villages. This strategy is not unusual and can be productive,
providing a labor force that is controlled by social relations of obligation and minimizing
training expenses since recruiters are responsible for training their kin and friends. Wolf
(1994) also describes this type of management strategy in her study of women laborers in
Indonesia factories. Besides the quite sex segregated export manufacturing work, women
are also employed in domestic and other types of services, which are even more sex
segregated.
Working against this sex-segregated effect of migrant social capital, might be the
strength of ties between destination and village of origin and the length of time a migrant
spends in a destination. Time in a place of destination serves to broaden the local ties a
migrant has to information and contacts regarding other jobs. With time a migrant may
be just as able to help the next migrant from their village regardless of their sex. This

10
information may only be useful to potential migrants if current migrants maintain ties to
places of origin. If women are more likely to maintain ties to their natal villages because
of cultural expectations or material reasons (maintaining access to their future
inheritance), then the longer their experience in places of destination the greater the
amount of information they may have and the more helpful they may be to potential men
or women interested in migration. On the other hand, mens time in places of destination
may be less useful to households and villages of origin, because they maintain less
contact and are seen to be less vested in the household and village economies.
The data for this study comes from one district in Northeastern Thailand, Nang
Rong, in Buriram province. The Northeastern region of Thailand is known for supplying
seasonal and permanent migrants to both rural and urban destinations (Chamratrithirong
et al. 1995), mostly because of the relative poverty of the region. Rice paddies dominate
the landscape, which are irrigated with rain. Frequent droughts and poor soil quality are
severe limitations to the agriculturally based economy. The region, a frontier until the
1970's (Phongphit 1990), no longer has unclaimed arable land (Siamwalla, Setboonsarng,
and Patamasiriwat 1993). The northeast region is also known for its relative poverty
compared with other parts of the country(Phongphit 1990). Because of poverty, past high
fertility and limited arable land for future development, the region has become an
important source of migrants to urban centers in Thailand, primarily Bangkok. Nang
Rong is located in the southern portion of the Northeast region along a major highway
built during the 1970s by the United States, linking Bangkok with the Laotian border.
This highway serves as an important conduit of people and resources to and from central
Thailand. Figure 1 displays a map of the study site and its relative location within
Thailand and the province of Buriram. In 1990, Nang Rong was one of the largest
districts in the country with a population of over 200,000 within an area of about 240,000
hectares.

- Figure 1 About Here

11
The Nang Rong Surveys and Analysis File
The Nang Rong Surveys are a longitudinal data collection effort conducted by the
Carolina Population Center at the University of North Carolina and the Institute for
Population and Social Research at Mahidol University in Thailand.1 We employ the first
two waves of data for this analysis, the data for the 1984 and 1994 survey rounds. The
1984 data collection was a census of 50 villages and included information individual
demographic data, household assets and village characteristics. Figure 1 shows the
location of the study site in relation to the country as a whole and within Buriram
Province. The 1994 data collection not only replicated the 1984 survey, including a
census of all households and information about former 1984 village members.

- Figure 2 About Here

Two other dimensions of the 1994 data collection are of interest for this study.
First, the 1994 data collection included 10-year retrospective life histories of some
members (those 18-35 years old in 1994), including information about their work,
migration and education experiences. Second, the 1994 data collection included a
migrant follow-up component where members of a 1984 household who had since left a
village for more than two months were interviewed if they came from one of twenty-two
original villages and went to one of four destinations: the provincial capital, Buriram; the
regional capital, Korat or Nakhon Ratchasima; Bangkok and the Bangkok Metropolitan
Area; and the Eastern Seaboard. Figure 2 shows the location of the four destinations of
interest. In related project manuscripts it has been documented elsewhere how successful
the surveys were at following households and individuals (Rindfuss, Kaneda,
Chattopahyay, and Sethaput 2002). On average, for the twenty-two villages, about 43%
of the migrants were successfully interviewed at some point in the six months following
the 1994 village surveys (see Table 1).

1
The data and information about the surveys are available at http://www.cpc.unc.edu/projects/nangrong/.

12
 In our analysis we build a data file that starts with 1984 household members that
are 8-25 years old from the twenty-two migrant follow-up villages2 and are matched with
information from the 1994 surveys. We use the life history information to construct a
person-age file that begins with those individuals that are 13-25 years old in 1984 and
adds persons to the record as they become 13 years old. We chose 13 years old as the
lower bound because it marks the end of primary schooling and the beginning moment of
being at risk of moving as an independent adult. Our analyses will be conducted on two
sets of data, one for the entire set of 13-25 year olds and subsequent individuals that age
into the file over the 10-year period and one that is only a subset of those that are 13-15
years old (to limit the historical bias of prior migration experience among those older
individuals). As you will see, the results are similar for both sets of analyses. Figure 3
shows the age structure and sample sizes within each cohort for each of the data files.
- Figure 3 -

These data are similar to the Mexican Migration Project data in the following
ways.3 First, the data come from a variety of communities and the data include measures
that can account for levels of development or other explanations for patterns of migration.
Second, the data include information about migration of similarly aged co-residents
enabling us to aggregate counts of migrant trips and experiences at the household and
village level. The data are different from the Mexican Migration Project in the following
ways. The villages were all surveyed at the same time, whereas the MMP surveys were
conducted over a period of ten years. The surveys among households and individuals
were complete censuses, rather than randomized samples of households as in the MMP
data. The 1984 survey serves as a baseline allowing us to view migration prospectively
and we can take into account some of our sample bias as a result of attrition, a factor that
cannot be adjusted for in the retrospective MMP data. The Nang Rong life history data
selects only those household members that were 18-35 years old in 1994 (8-25 years old

2
The sample selection bias resulting from migration significantly affects the composition of our life history
data file if we were to examine data from all fifty of the original villages. Out migrants from the twenty-
two villages in the survey have significantly different patterns of migration, education, and work
experiences than those surveyed in the villages in 1994.
3
Data and information from the MMP are available at: http://lexis.pop.upenn.edu/mexmig/welcome.html.

13
in 1984), whereas the primary focus of the life history data for the MMP was male heads
of households (understandable, given the Mexican-U.S. migration context). The MMP
life history accounts include lifetime events, not just a ten-year retrospective as with the
Nang Rong Surveys. Finally, the Nang Rong surveys do not record age at first migration,
whereas the MMP data allow estimates of first moves. Importantly for this study, the
Nang Rong data allow for an extended analysis of the sex disaggregated effects of
accumulated trips and experience at the individual, household and community level,
otherwise not available with the MMP.
Migration prevalence grows dramatically over the 10-year time period. The top
image in Figure 4 displays the overall migration prevalence rate for all individuals for
which life history information was collected. A migration prevalence rate measures the
proportion of people that have ever migrated up to a point in time (Massey et al. 1994).
In the Thai case, there is considerable variation across the twenty-two villages and that
variation is maintained over time. In 1985, at the low end, in one village only 5% had
ever migrated in the past year and at the higher end 50% had ever migrated in the past
year. By 1994 all villages had increased their prevalence rates, but the wide range
between high and low villages was still apparent. The middle and bottom images display
villages female and male migration prevalence rates, respectively. There is similarly
wide variation across villages in both images, although female prevalence rates are
generally lower than male migration prevalence rates and male migration rates appear to
rise more steeply over the period. The patterns for male and female migration
prevalence in a particular village do not always mirror each other either. Figure 5 shows
two villages disaggregating the pattern of male and female migration prevalence. One
village is a high female migration village in 1985 and the other is a high male migration
village. In the high female migrant prevalence village, the proportion of those ever
migrating is 50% in 1985 and increases steadily to 75%. For men in that village the rates
are lower and do not mirror the pattern for female migration prevalence. On the other
hand, in the high male migration village, the proportion of males who have ever migrated
rises from 29% to 85% and the female migration prevalence rate, although lower, follows
a similar upward trend. It is this significant variability in migration prevalence over

14
time and across the sexes that is of interest and provides an opportunity to evaluate their
impact on individual migration propensities.

Measures and Analytic Approach

Our analytic approach builds on a model developed in Massey and Zenteno
(1999) to measure the dynamics of mass migration. We employ their approach for a
number of reasons. Our data are limited in that we do not know the date of first
migration and without making some very large assumptions cannot presume to estimate
it. We do have a population of villagers 8-25 years old in 1984 and observe them
forward in time, much like Massey and Zenteno (1999) do with their communities, taking
into account sample attrition. We build on their model by adding a set of baseline
attributes of individuals that might account for possible unobserved heterogeneity related
to our explanatory factors and the dependent variable. We are interested in predicting
whether a person is living outside of Nang Rong district or not in time t. Our model takes
the following form and takes into account the correlated error structure of multiple
observations from individuals (we estimate a random effects logistic equation):
(1) Prob(Migit) = f(Itripsit-1, Iexpit-1, Htrips it-1, Hexp it-1, Ctripsit-1, Cexpit-1, Variantit ,
Invarianti )

where Prob(Migit) is a person i's probability of migrating in year t, Migit is 1 if i migrated

in year t and 0 otherwise, Itripsit-1 is the number of trips made by person i up through year
t-1, Iexpit-1 is the number of months experienced as a migrant by person i up through year
t-14, Htripsit-1 is the number of trips made by other people in is household up through
year t-1, Hexpit-1 is the number of months experienced as a migrant by other people in is
household up through year t-1, Ctripst-1 is the number of trips made by other community
members up through year t-1, Cexpt-1 = the months of experience accumulated by other
community members through year t-1.

4
Exact time in a place of destination was not measured by the survey. The survey did allow for multiple
trips within one year (up to six trips including return trips) and the survey links trips with work,
education, and destination information. To measure months of migrant experience we took the number of
trips taken within one year and divided it into 12 months. For one trip in one year we calculated the
amount of experience as six months, for two trips we counted it as four months for each trip. Only 10
percent of the sample ever made more than round one trip in one year.

15
 We include as controls a vector of time varying factors, including age, educational
attainment, marital status, village electrification, and the migration prevalence rate.
Except for age, these are all included in the models as measured in time t-1. We also
include as controls a vector of time invariant measures5, including: sex, household land
ownership in 19846, whether the person was a temporary migrant in 19847, whether there
were any temporary migrants in their household in 1984, the persons villages
proportion of 13-30 year olds in 1984 that were temporary migrants, the female
temporary migration rate, and whether the person lived in a very remote village in 1984.
A village is considered very remotely located the more obstacles there are to travel. The
obstacles we include in our measure are the presence of a portion of the route to the
district town that is a cart path (unpaved, rutted, and narrow), the lack of public
transportation to the district town, travel to the district town takes an hour or more (as
reported by a village headman or key informant), that during the year there are four or
months of difficult travel to leave the village (this is also a measure of road conditions
and susceptibility to flooding), and it is 20 or more kilometers to the district town. If all
four of these characteristics are present we code the village as very remote.
In this specification, the rate of out-movement depends not only on the age and
sex of the individual, but also on a persons prior migratory experience (i.e., on his or her
accumulated human capital) and on the degree to which he is surrounded by other
villagers with migratory experience (the quantity of social capital). Our model is
different from Massey and Zentenos model in the following ways: first, we add a
measure of household migrant trips and experience (because we suspect that the quality
of information available to potential migrants is different at the household level than at
the village level); second, we add a larger array of time varying and invariant factors that
might be related to the migration decision; third, we control for sample attrition due to
our the data collection design and success in following migrants over time; and fourth,
we include a measure of village migration prevalence as suggested by Massey, Goldring

5
All of these, except for the measure of sex, were data collected during the 1984 survey.
6
This is measured with three categories of land ownership: landless, near landless 1-10 rai, somewhat
landed (11-25 rai), and landed (more than 25 rai). A rai is 0.46 acres.
7
A temporary migrant is a person identified in the household roster as a permanent household member,
who migrated out of the house for two or more months, but plans to return.

16
and Durand (1994). We do so because we are particularly interested in the effect of the
quantity and quality of social capital, net of the exposure to migration experiences.
Finally, we estimate our equation on two separate samples of the data. We do so
because the larger sample includes many older young adults, some of whom may have
had significant amounts of unobserved prior migration experience, biasing our results.
The direction of the bias is unclear. If rural out migration takes a life course path, this
may diminish the possible effects of cumulative migration. Significant unobserved prior
migration may also be related to some of the right hand side variables increasing the
possibility of spurious results. We do have some reason to suspect that these effects are
relatively minimal. The baseline data captures the very start of the period of rapid
economic growth and the beginning of large movements of young people out of rural
villages in northeastern Thailand. We also include measures of migration experience in
1983 that should help to diminish the possible biases. The younger sample upon which
we replicate our models, reflects the youngest cohorts in the data. Given the
overwhelming rise in migrant prevalence in the villages, these migrants may be much less
influenced by the increase in quantity and quality of social capital, given high levels
migrant experience saturation. We expect that our results from the analysis of the overall
sample will be significantly dampened when evaluating the effects on the younger
cohorts.

- Table 1 -

Table 1 displays the means for all the variables in the model. The time invariant
factors are shown in the 1984 panel and the time variant factors displayed in the 1994
panel. In 1984 men are slightly younger in both samples. In the overall sample, this
probably reflects life course related outmigration for marriage. The overall sample is
more likely to have married individuals than the sample of the younger cohort and
women are more likely to be married than men in the sample. Again, this could be
because of sex differences in age at marriage, but it could also be that men are more
likely to leave the village to marry following a tradition of matrilineal postnuptial
residence. A majority of young people come from households with some land. A vast

17
majority were not temporary migrants (gone for 2 or more months in the last year in
1984). On the other hand almost a third of the individuals in the sample live in
households that did have a temporary migrant in 1983-1984. On average the individuals
in the sample came from villages where approximately 46% of the 13-30 year olds were
temporary migrants in 1984 and about 35% of women were migrants. So, there were
fairly high levels of pre-existing levels of migration at the start of the ten-year period. In
1984 most lived in very remote villages and only a third live in a village with electricity.
In 1984, most respondents had only a primary education, but some secondary education
was not unusual.
Although very few had been temporary migrants between 1983 and 1984, by
1994 more than 60% had ever migrated out of the Nang Rong district. This rate was
higher among men in both groups, but highest among the younger group of men. Forty
percent of women had gotten married by 1994 in the younger group and over half of
women in the older group had done so. For men the proportion of married individuals is
much lower for the younger cohort (just over 10% had gotten married) and in the overall
sample (just over a third). By 1994 all the villages had electricity. With regards to
schooling, ten percent of women had completed secondary schooling in both samples and
15% of men in the overall sample had done so. The highest completion rate was among
the younger cohort of men.
On average women had made a little more than one trip out of the district,
whereas men had made closer to 1.5 trips. Most had spent more than two years out of the
district over the last ten years, and the men in the overall sample had spent almost three
years outside of the district. Most came from household where other household members
had every migrated, although other members were more likely to have spent fewer
months outside of Nang Rong (this measure is standardized by the number of people in
the household contributing to the measure). The measure of village trips and months is
standardized by the number of people in the village. The average number of trips by
individuals in each individuals village is about 1 and a third and the average number of
months outside of the village are more than two years. The gendered version of each of
these variables shows that the contribution of men and women to each of these patterns is
roughly equal.

18
 Our modeling approach introduces a set of baseline controls, the vectors of time
varying and invariant factors, we then introduce the measures of accumulated migration
experience, and we include a measure of year effects to control for changes in economic
and structural opportunities (results are found in Table 2 for the overall sample and Table
6 for the younger cohort). Our next step evaluates these models separately for men and
women (Tables 3 and 4 for the overall sample and Tables 7 and 8 for the younger cohort).
Finally, we disaggregate our measures of migration experience and evaluate models by
sex to and evaluate models that includes the variables on the importance of the gendered
content of social capital for the a pooled sample of men and women and for men and
women separately (Table 5 shows results for the overall sample and Table 9 shows
results for the younger cohorts). First, we briefly discuss our baseline model and then
move on to discuss the results evaluating the effect of migrant social capital on the
probability of migrating in the Thai context. In general, we evaluate all of our results for
the overall sample first and then
discuss differences with regards to the younger cohort.

Results From A Baseline Model

In Table 2 the results for Model 1 show that as a person gets a year older their
odds of being a migrant (in time t) increase by 3.4 times, a dramatic and significant
increase. This age effect, however, attenuates as shown by the lowered odds ratio for the
squared term. Men are 2.3 times more likely to be migrants than are women. Some
secondary schooling also increases the odds of being a migrant relative to primary
schooling, but completing secondary school does so dramatically (by almost 3 times).
Not surprisingly, marriage decreases the odds of being a migrant quite dramatically, by
almost 65%.
Land ownership is not significantly associated with being a migrant although the
pattern of the effect is curvilinear. Individuals from households that are near landless or
have middle-sized landholdings (11-25 rai) have higher odds of being migrants, relative
to landless households. However, individuals from larger landholding households are
less likely to be migrants than are individuals from landless households.

19
 If individuals are temporary migrants in 1984 then the odds that they are also a
migrant in any subsequent year are dramatically and significantly higher. Similarly, if
the household they live in has a temporary migrant the odds are 50% greater that they
will be migrants in any subsequent year. Village level rates of temporary migration have
no effect on the odds of subsequent migration, but the female migration rate has a
marginally significant effect, reducing the odds of being a migrant.
Living in a remote village significantly increases the odds of being a migrant by
2.3 times. We suspect that this is an effect of not being near local labor market
opportunities available in the district town. Electrification has no significant effect on the
odds of being a migrant. The success of the 1994 field work in following migrants from
each village also has an effect on the odds of being a migrant at a point in time8.

- Tables 2 and 3 About Here

Table 3 displays the results of the same model, but evaluated on separate samples
of men and women. We find that although age has a greater effect on the odds of
migration for men than for women, the effect is positive for both. Schooling, however, is
different for men and women. Having some secondary schooling increases the odds of
mens migration, but only completing secondary schooling increases the odds of
womens migration. Marriage is a greater deterrent of migration for women than for
men. Land ownership, especially near landlessness approaches significance for men,
increasing the odds of out migration. Prior migration experience in 1984 dramatically
increases the womens odds of moving and also increases the odds of mens migration,
but much less dramatically. The presence of temporary migrants in a household only
increases the odds of womens migration. On the other hand, village temporary
migration rates increase the odds of mens migration, but not womens. Further, high
rates of womens temporary migration are a significant deterrent for mens out migration.
All of these effects are similar for the equivalent models tested on the sample of younger

8
Our prior expectation for this would not have predicted this. Finding migrants is probably
related to the frequency by which they return to the village for visits or their stability in places of
destination. I also expect that the migrant follow-up rate is related to some of these right hand
side variables (see Rindfuss et al. 2002), and the patterns of cumulative migration.

20
cohorts except that land ownership becomes a significant factor. Individuals from larger
landholding households are less likely to be migrants. In addition, the effect of village
migration is attenuated for men (see Tables 6 and 7) in the models evaluated on the
sample of younger cohorts.

Migrant Social Capital and Its Effects on Men and Women

Table 2 also shows the results for the overall sample for the effects of migrant
social capital upon the odds of being a migrant. We discuss the results for Model 3,
which controls for the year effects in the data. We find that each additional trip (net of
village migration prevalence rates) increases the odds of being a migrant by 47%. And
an additional month of experience significantly increases the odds by about 1.5%. Trips
and months of experience by other household members also have similar effects on the
odds of migration. Trips by other villagers have no significant effect upon the odds of
being a migrant and months of migrant experience has a negative effect. The odds of
being a migrant are reduced by about 4% if all village members increased their months of
migrant experience by one month. It would appear that household level migrant social
capital is more important than village level social capital. And time away by other
villagers may actually diminish the influence of migrant networks upon the odds of
migrating. Before speculating further, we describe the results of the models evaluated
upon separate samples of men and women (see Table 4)

- Table 4 About Here -

Individual migration experience is important for both men and women, but more
so women. Women are 80% more likely to be a migrant with each additional trip they
have made previously. Men are only 27% more likely with each additional prior trip.
Trips by other household members are also important for both men and women, but more
so for men. Each additional trip taken by other household members increases the odds of
man being a migrant by 56% and only increases the odds for a woman by 27%.
Interestingly, trips by other villagers has a marginally significant and positive effect upon
womens migration, but a strong negative effect upon the odds of being a male migrant.

21
For both men and women the number of migrant months experienced by other villagers
significantly decreases the odds of migration.
These last puzzling results, regarding the contradictory effects of village trips
upon mens and womens behavior, suggested to us that there may be further information
not yet distilled about the influence of migrant social capital, particularly about the
gendered content of migrant networks. So, before discussing these results further we will
turn to the effects of the gendered content of migrant social capital. Briefly, though, we
find similar results for the analysis conducted on the younger cohorts (results found in
Tables 6 and 8).

Engendering Migrant Social Capital

When we decomposed our migrant social capital measures into those capturing
mens and womens experiences, several interesting findings stand out. These results are
found in Table 5.

- Table 5 About Here -

Table 5 presents the results for a full model, including baseline controls and year
effects. As with the results in Table 4, the effects of an individuals prior migrant
experiences have a strong influence upon the probability of being a migrant. What is
more revealing is that household and village migrant social capital has significantly
different effects depending on gender composition. Because the results for the pooled
sample mask important differences in effects for men and women we describe the results
for the models evaluating results separately for men and women.
Migrant trips by other female household members have no significant effects on
the odds of being a migrant for men or women. However, months of experience do
significantly increase the odds of being a migrant for both men and women. As
suggested earlier, we suspect that this effect is a result of strong ties between female
migrants and their households of origin and the quality of information they are able to
share (through frequent interactions) about opportunities and provision of care for other
migrants upon those migrants arrival in a destination. Migrant trips by women villagers

22
has little effect on a womans odds of migration, suggesting to us that the quality of
social capital is lower than that provided by other women household members who are
migrants. On the other hand, the effect upon a mans odds of migration is strongly
negative. Again, we suspected this possibility given the sex segregated labor market in
places of destination. On the other hand, as women villagers gain migrant experience and
because they maintain ties to their villages of origin they are able to provide additional
information to their fellow male villagers that facilitate these mens migration. Contrary
to this effect, however, as other women villagers migration experiences lengthen this
lowers the odds that women will be migrants. Although somewhat unexpected, we
suspect that the loss of women from a village may increase the hold of a village on
current women villagers limiting their opportunities for migration (or to put it more
positively, increasing their opportunities in the village of origin).
Male migrant social capital has somewhat different effects for men and women.
In general, male migrant social capital, measured as trips or experience or measured at
the household or village level, has no significant effect upon the odds of a woman being a
migrant. In some cases, male migrant social capital does influence the odds of a man
being a migrant. Specifically, the number of trips by male household members does not
significantly influence the odds of a woman migrating, although the effect is positive and
approaches significance. Trips by other male household members however do increase
the odds of a man being a migrant. An additional trip by all other male household
members would increase the odds of a man being a migrant by 2 times. Months of
experience by other male household members has no significant effect upon either men
or women and neither does the number of migrant trips by male villagers. Trips by male
villagers have a positive effect upon the odds of being a migrant, but are not significant.
However, contrary to the effect of months of migrant experience among female villagers,
which positively influences the odds of migration, the effect of male villagers migrant
months of experience has a negative effect upon the odds of a man being a migrant. If all
male villagers increased their months of migrant experience by one month then the odds
that a man will be a migrant fall by 14%. The results for similar models for the younger
cohorts show that most effects hold, although slightly attenuated, except for the effect of
other male household members migrant trips. In this case, an increase in the number of

23
trips by other male household members lowers the odds of a woman migrating, but not
the odds of a man migrating. The year effects for the younger cohort are strong and
reflect the aging of the cohort. Interestingly, these aging effects are only strongly
significant for men. The odds of being a migrant decline dramatically with each year for
men and seem to stay relatively flat for women.

- Tables 6-9 About Here

In the next section we briefly summarize our results and discuss them in light of
the literature and the Thai context. We suggest that these very different patterns of male
and female migrant social capital and related migration behavior may indicate important
nuances about the migration process, drawing attention to the possibly varying impact
upon the organization of social life and policy. We also discuss the limitations of this
study and suggest next steps for future research.

Discussion of Results: Gendered Nuances of Migrant Social Capital

In this study we sought to extend a model of migration previously tested in only
one other location (Mexico-U.S. migration) to the case of Thailand. Our hypothesis was
that although the theory suggested that migrant social capital was particularly important
for cross-national migration, the process of out migration was similarly profound and
fraught with risks in the Thai case and that migrant social capital would also be a
meaningful influence upon migration behavior. We also expected that the very different
context of gender relations in Thailand, a relatively high status of women with some
freedom to migrate, dramatic economic growth creating jobs outside of the local
economy for both men and women, and gendered social ties with natal households would
create very different types of migrant social capital with different effects for mens and
womens probabilities of being migrants.
Our results show that migrant social capital measured at a household or village
level has an important influence upon migrant behavior. Our results also show that an
exponential growth in migration, growing momentum, or cumulative causation may not
always happen and may depend on the sex composition of prior migrants, whether the

24
migrant is a household member or a village member, and whether migrant social capital
is measured as number of trips or months of experience. Importantly, we also show that
migrant social capital is still important even after controlling for migrant prevalence
rates. That is, even as villages approach saturation with regard to migration, additional
trips or months of experience are important determinants of migrant behavior.
For all of the analyses, an individuals prior experience is an important
determinant of migration, whether it is months of experience or number of trips.
Household-based migrant social capital appears to have stronger positive effects than
village-based migrant social capital for all individuals. This suggests that the quality of
information from household-based migrant networks may be significantly better or more
helpful than that from village-based migrant networks. This is not surprising and has
also been found in studies of Mexican-U.S. studies of migration (Cerrutti and Massey
2001; Espinoza and Massey 1999; Kanaiaupuni 2000; Massey and Espinoza 1997).
Stronger ties and greater degrees of trust may characterize household- or family-based
migrant social capital significantly decreasing the costs of migration. Beyond these
conclusions, we cannot discuss the results for generalize migrant social capital without
discussing the gendered nuances of migrant social capital.
We find several interesting and different patterns depending on whether migrant
social capital is male or female-based and whether men or women are considered at risk
of migrating. Female-based migrant social capital finds its greatest impact when it comes
to months of experience, especially at the household level. The longer women spend as
migrants the more helpful they can be to future migrants. This is not the case with regard
to male-based migrant social capital. We suspect that since women tend to maintain ties
to natal households, at least in the Thai context, and men do not, that female migrants
enhance the quality of migrant social capital as women develop contacts and resources in
places of destination. The longer men are away from natal households the weaker their
ties and the lower the quality of migrant social capital transmitted through them. If,
however, men make frequent trips and thereby demonstrate a strong tie to their natal
household the quality of their migrant social capital improves the likelihood that other
household members will migrate, particularly other men.

25
 The gender nuances of village migrant social capital are particularly striking. The
greater the frequency of village trips is important for predicting womens migrant
behavior, but not so for men. When we explored this finding further and decomposed
village effects by sex, we find that one of the reasons for the negative effect of village
trips upon men is because of the strong negative influence of female-based village
migrant social capital. The number of trips by village women significantly dissuades
men from migrating. We suspect that this is a result of sex segregated labor markets in
urban destinations. Export manufacturing or service jobs overwhelmingly hire women
and offer few opportunities for men, especially if they do not have a secondary education.
On the other hand, as with months of experience at the household level, female-based
village migrant social capital as measured by months of experience reverses this effect
and significantly improves the odds that men will migrate. Again, because women
maintain ties to villages of origin, the longer they stay in a place of destination the more
they know about migrant opportunities for both men and women and the greater the
information they transmit back to other villagers about migrant opportunities. There are
no such effects for male-based village migrant networks.
In conclusion our study has demonstrated that migrant social capital is also
important in the Thai context, even when considering rural-urban migration. However,
because of the very different patterns of migration in the Thai case and the data available
to us, specifically that there is significant variation in the gendered patterns of migration
and that data were collected from both men and women, we can also offer some
refinements on the theories about migrant social capital. We confirm what ethnographic
researchers have previously found in their qualitative studies and expand upon the few
prior studies of gender and migration. Gender relations imbue the quality of migrant
social capital with very different results for mens and womens migration outcomes.
The patterning of these effects can be quite profound and have implications for
understanding the rate and pace migration out of rural areas. Although we do not explore
these effects, we suspect that this patterning will influence gendered social organization
in villages and households of origin, as well as differentially influence the impact of
migration upon village and household economies. One example might be that high rates
of migration by women may yield very different flows of remittances back to villages of

26
origin than if men predominate in a migrant stream. In the Thai context women are
considered more reliable remitters than are men(Curran and Saguy 2001). On the other
hand, high rates of migration by women, may also weaken their ties to natal villages and
shift patterns of care provision from daughters to sons, especially if high rates of female
migration are a significant deterrent of male out migration (Curran and Saguy 2001).
These are offered as tentative speculations that require more exploration.
Our study adds to the small but growing collection of quantitative research which
builds on what ethnographers have been suggesting for a long time: ignoring the
gendered content of migrant social capital may mask important insights about the
dynamics of the migration process. Our analysis presumes that because gender is such a
profound social category, organizing many aspects of social life, that it inevitably
influences the quality of migrant social capital, as well. We speculate that this is because
men and women live the migrant experience very differently. In part because men and
women face different barriers to moving, maintain different relationships with
households and villages of origin, and experience completely different opportunities in
places of destination. Future research should explore these particular dimensions more
explicitly, taking into account variability across destinations, work experiences, patterns
of remittances and visits home, and varying barriers to migration.

27
 Reference List

1. Arizpe, Lourdes. 1975. Indgenous En La Ciudad De Mxico: El Caso De Las

Maras. Mxico D.F., Mxico: Secretara de Educacin Pblica.

2. . 1980. La Migracion Por Relevos y La Reproduccion Social Del

Campesinado. Mexico D.F., Mexico: Centro de Estudios Sociologicos, El
Colegio de Mexico.

3. . 1981. "Relay Migration and the Survival of the Peasant Household." Why
People Move, editor Jorge Balan. Paris, France: UNESCO.

4. . 1985. Campesinado y Migracin. Mexico D.F., Mexico: Secretara de

Educacin Pblica.

5. Bello, Walden, Shea Cunningham, and Li K. Poh. 1998. A Siamese Tragedy:

Development and Disintegration in Modern Thailand. London: Zed
Books, Ltd.

6. Blanc-Szanton, M. C. 1990. "Gender and Inter-Generational Resource Allocation

Among Thai and Sino-Thai Households." Pp. 79-102 in Structures and
Strategies: Women, Work and Family, vol. 3, editors Leela Dube and
Rajni Palriwala. New Delhi and London: Sage Publications.

7. Boyd, Monica. 1989. "Family and Personal Networks in International Migration:

Recent Developments and New Agendas." International Migration
Review 23(3):638-70.

8. Cerrutti, Marcela and Douglas Massey. 2001. "On the Auspices of Female
Migration From Mexico to the United States." Demography 38(2):187-
200.

28
 9. Chamratrithirong, Aphichat, Kritaya Archavanitkul, Kerry Richter, Philip Guest,
Thongthai Varachai, Wathinee Boonchalaksi, Nittaya Piriyathamwong,
and Panee Vong-ek. 1995. National Migration Survey of Thailand.
Bangkok, Thailand: Institute for Population and Social Research, Mahidol
University.

10. Chant, Sylvia. 1992. Gender and Migration in Developing Countries. London,
UK: Belhaven Press.

11. Choldin, Harvey M. 1973. "Kinship Networks in the Migration Process."

International Migration Review 10:163-75.

12. Constable, Nicole. 1997. Maid to Order in Hong Kong: Stories of Filipina
Workers. Ithaca: Cornell University.

13. Curran, Sara and Estela Rivero. 2003. "Engendering Migrant Networks: The
Case of Mexican Migration." Demography 40(2).

14. Curran, Sara and Abigail C. Saguy. 2001. "Migration and Cultural Change: A
Role for Gender and Social Networks." Journal of International Women's
Studies 2(3):54-77.

15. Davis, Benjamin, Guy Stecklov, and Paul Winters. 2002. "Domestic and
International Migration From Rural Mexico: Disaggregating the Effects of
Network Structure and Composition." Population Studies 56:291-309.

16. De Jong, Gordon, Kerry Richter, and Pimonpan Isarabhakdi. 1996. "Gender,
Values, and Intentions to Move in Rural Thailand." International
Migration Review 30(3):748-70.

17. Espinoza, Kristin and Douglas Massey. 1999. "Undocumented Migration and the
Quantity and Quality of Social Capital." Migration and Transnational
Social Spaces, Ludger ed. Pries. Aldershot: Ashgate.

18. Grasmuck, Sherri and Patricia R. Pessar. 1991. "Between Two Islands:

29
 Dominican International Migration." Pp. 1-161 in Berkeley//Oxford:
University of California Press.

19. Grieco, Elizabeth M. and Monica Boyd. 1990. "Women and Migration:
Incorporating Gender Into International Migration Theory." Working
Paper Series. College of Social Science, Florida State University.

20. Hanks, Lucien M., Jr. and Jane R. Hanks. 1963. "Thailand: Equality Between the
Sexes." Pp. 424-51 in Women in the New Asia: The Changing Social Roles
of Men and Women in South and South-East Asia, editor Barbara E. Ward.
Paris and Amsterdam: UNESCO.

21. Hondagneu-Sotelo, Pierrette. 1994. Gendered Transitions: Mexican Experiences

of Immigration. Berkeley, Calif.: University of California.

22. Hugo, Graeme. 1991. "Village-Community Ties, Village Norms, and Ethnic and
Social Networks: A Review of the Evidence From the Third World." Pp.
186-225 in Migration Decision Making, editors Gordon DeJong and
Robert W. Gardner. New York: Pergamon Press.

23. Kanaiaupuni, Shawn M. 2000. "Reframing the Migration Question: An Analysis

of Men, Women, and Gender in Mexico." Social Forces 78(4):1311-48.

24. Keyes, Charles. 1984. "Mother or Mistress but Never a Monk: Buddhist Notions
of Female Gender in Rural Thailand." American Ethnologist 11(2):223-41.

25. Limpinuntana, Viriya, Aran Patanothai, Terd Charaoenwatana, Gordon Conway,

Manu Seetisarn, and Kanok Rerkasem. 1982
. An Agroecosystem Analysis of Northeastern Thailand. Khon Kaen,
Thailand: Khon Kaen University.

26. Massey, Douglas. 1990a. "The Social and Economic Origins of Immigration."
Annals of the American Academy of Political and Social Science
510(July):60-72.

30
27. . 1990b. "Social Structure, Household Strategies, and the Cumulative
Causation of Migration." Population Index 56(1):3-26.

28. Massey, Douglas and Felipe Garca-Espaa. 1987. "The Social Process of
International Migration." Science 237:733-38.

29. Massey, Douglas, L. Goldring, and Jorge Durand. 1994. "Continuities in

Transnational Migration: An Analysis of Nineteen Mexican
Communities." American Journal of Sociology 99:1492-533.

30. Massey, Douglas and Rene Zenteno. 1999. "The Dynamics of Mass Migration."
Proceedings of the National Academy of Sciences :5328-35.

31. Massey, Douglas S. 1990. "The Social and Economic Origins of Immigration."
ANNALS, AAPSS 510:60-72.

32. Massey, Douglas S. and Kristin E. Espinoza. 1997. "What's Driving Mexico-U.S.
Migration? A Theoretical, Empirical, and Policy Analysis." American
Journal of Sociology 102(4):939-99.

33. Menjivar, Cecilia. 1995. "Kinship Networks Among Immigrants: Lessons From a
Qualitative Comparative Approach." International Journal of
Comparative Sociology 36(3-4):219-32.

34. Menjvar, Cecilia. "Immigrant Kinship Networks: Vietnamese, Salvadoreans and

Mexicans in Comparative Perspective." Journal of Comparative Family
Studies .

35. . 2000. Fragmented Ties: Salvadoran Immigrant Networks in America.

Berkeley, CA: University of California.

36. Mills, Mary B. 1997. Thai Women in the Global Labor Force: Consuming
Desires, Contested Selves. New Brunswick, NJ: Rutgers University .

37. Nartsupha, Chatthip. 1999. The Thai Village Economy in the Past. Chiang Mai,

31
 Thailand: Silkworm Press.

38. Nidhiprabha, Bhanupong. 1994. Determinants of Private Investment Expenditures

and Driect Foreign Investment in Thailand. Bangkok: Thailand
Development Research Institute.

39. Pessar, Patricia. 1999a. "Engendering Migration Studies: The Case of New
Immigrants in the United States ." American Behavioral Scientist
42(4):577-600.

40. . 1999b. "The Role of Gender, Households, and Social Networks in the
Migration Process: A Review and Appraisal." Pp. 53-70 in Handbook of
International Migration: The American Experience, editors Charles
Hirschman, Philip Kasinitz, and Josh DeWind. New York, New York:
Russell Sage Foundation.

41. Phongpaichit, Pasuk. 1980. "The Open Economy and Its Friends: The
"Development" of Thailand." Pacific Affairs 53(3):440-460.

42. Phongpaichit, Pasuk and Chris Baker. 1996. Thailand's Boom! New South Wales,
Australia: Allen and Unwin.

43. Phongpaichit, Pasuk and Chris Baker. 1998. Thailand's Boom and Bust. Chiang
Mai, Thailand: Silkworm Press.

44. Phongphit, Seri. 1990. Thai Village Life: Culture and Transition in the Northeast.
Bangkok, Thailand: Mooban Press.

45. Piore, Michael. 1979. Birds of Passage: Migrant Labor in Industrial Societies.
Cambridge: Cambridge University Press.

46. Portes, Alejandro. 1978. "Migration and Underdevelopment." Politics and Society
8(1):1-48.

47. Rindfuss, Ronald, Toshiko Kaneda, Chattopahyay, and Chanya Sethaput. 2002.

32
 "Panel Studies and Migration." Carolina Population Center, University of
North Carolina.

48. Siamwalla, Ammar, Suthad Setboonsarng, and Direk Patamasiriwat. 1993.

"Agriculture." Pp. 81-117 in The Thai Economy in Transition, editor Peter
Warr. Hong Kong: Cambridge University Press.

49. Singhanetra-Renard, Anchalee and Nitaya Prabhudhanitisarn. 1992. "Changing

Socio-Economic Roles of Thai Women and Their Migration." Pp. 154-73
in Gender and Migration in Developing Countries, editor Sylvia Chant.
New York: Belhaven Press.

50. Stark, Oded. 1991. The Migration of Labor. Cambridge, MA: Basil Blackwell.

51. Stark, Oded and Mario Levhari. 1982. "On Migration and Risk in LDCs."
Economic Development and Cultural Change 31:191-96.

52. Sussangkarn, Chalongphob. 1993. "Labour Markets." Pp. 355-400 in The Thai
Economy in Transition, editor Peter G. Warr. Cambridge, UK: Cambridge
University Press.

53. Taylor, Edward. 1986. "Differential Migration, Networks, Information and Risk."
Research in Human Capital and Developmnt: Migraiton, Human Capital
and Development 4:147-71.

54. Tienda, Marta and Karen Booth. 1991. "Gender, Migration and Social Change."
International Sociology 6(1):51-72.

55. Tyner, James A. 1996. "The Gendering of Philippine International Labor

Migration*." Professional Geographer 48(4):405-16.

56. Warr, Peter, editor. 1993. The Thai Economy in Transition. New York:
Cambridge University Press.

57. Warr, Peter and Bhanupong Nidhiprabha. 1996. Thailand's Macroeconomic

33
 Miracle: Stable Adjustment and Sustained Growth. Washington, D.C.: The
World Bank.

58. Winters, Paul, Alain de Janvry, and Elisabeth Sadoulet. 2001. "Family and
Community Networks in Mexico-US Migration." Journal of Human
Resources 36(1):159-84.

59. Yoddumnern-Attig, Bencha. 1992. "Thai Family Structure and Organization:

Changing Roles and Duties in Historical Perspective." Pp. 8-24 in
Changing Roles and Statuses of Women in Thailand, editors Bencha
Yodumnern-Attig, Kerry Richter, Amara Soonthorndhada, Chanya
Sethaput, and Anthony Pramualrathana. Bangkok, Thailand: Institute for
Population and Social Research, Mahidol University.

34
Figure 1: Map of Setting and Study Site

Provincial Map of Thailand

0 150 300
Myanmar
Kilometers
Population size
Chang Rai ! 5,000,000 and greater ! 50,000 to 100,000
!
Laos ! 100,000 to 250,000 ! Less than 50,000
Mae Hong Son
! Phayao

Chiang Mai
!
Provincial Boundary
Nan
! !
Lamphun!
Lampang
! Phrae
! South
Uttaradit
! Loei
!
Nong Khai
Udon Thani
China
Sea
!
! Nakhom Phanom !
Sukhothai !
Tak ! Phitsanulok Sakon Nakhon
!
!
Kamphaeng Phet Phetchabun Khon Kaen Kalasin
! !
Phichit
! ! !
Maha Sarakham !
!Roi Et
Nakhon Sawan Chaiyaphum
! !Yasothon
! Thailand
Uthai Thani
! Ubon Ratchathani
Chainat! Nakhon Ratchasima Buriram Sisaket
!
!
Sing Buri
! Thahanbok Lop Buri !!
!
! Surin
Ang Thong
! Saraburi )
"
Supham Buri ! ! Nang Rong
Andaman Kanchanaburi
! Phra Nakhon Si Ayutthaya
Nakhon Nayok
Pathum Thani ! Prachin Buri
Sea !
Nakhon Pathom!
! !
! Bangkok
! ! Nonthaburi
Area of
Ratchaburi
! ! !Samut Prakan
! Samut detail
! Chon Buri
Samut
Songkhram
!
Sakhon Vietnam
Phet Buri
Rayong Chanthaburi
! !
Trat
! Cambodia
! Prachuap Khiri Khan

Gulf
of
Thailand
! Chumphon

!
Ranong
Map of Study Site
Road
!
Surat Thani 0 30 60

Phangnga Nakhon Si Thammarat Kilometers

! !
Krabi
! Buriram
Phuket !
Phatthalung !
!
South
!
Trang
Songkhla
!
Pattani
! China
Satun Nang Rong
! Yala
!
! Narathiwat
! Sea

Malaysia
Created by Tsering Wangyal Shawa

35
Figure 2: Map of Migrant Destinations Where Migrants Were Followed for the Nang
Rong, 1994 Survey

Map of Migrant Destinations

0 250 500
Myanmar
Kilometers
! Provincial Capital Bangkok Metropolitan Area
Laos ! Regional Capital Eastern Seaboard
U.S. Friendship Highway

Nakhon Ratchasima Buri Ram

! !
Nang Rong
"

[ Bangkok
Andaman
Sea
Cambodia
Area of
detail
Vietnam
Gulf
of
Thailand

Provinces in the
Bangkok Metropolitan Area
Pathum Thani and Eastern Seaboard
Nakhon
Pathom Nonthaburi
Krung
Mahanakhon

Samut Chachoengsao
Samut Prakan
Sakhon

Gulf Chon Buri

of
Thailand

Rayong
0 30 60
Malaysia Kilometers

Created by Tsering Wangyal Shawa

36
Figure 3: Sample of Life History Information

Year
84 85 86 87 88 89 90 91 92 93 94
8 9 10 11 12 13 14 15 16 17 18
9 10 11 12 13 14 15 16 17 18 19
10 11 12 13 14 15 16 17 18 19 20
11 12 13 14 15 16 17 18 19 20 21
12 13 14 15 16 17 18 19 20 21 22
13 14 15 16 17 18 19 20 21 22 23
14 15 16 17 18 19 20 21 22 23 24
15 16 17 18 19 20 21 22 23 24 25
Age

16 17 18 19 20 21 22 23 24 25 26
17 18 19 20 21 22 23 24 25 26 27
18 19 20 21 22 23 24 25 26 27 28
19 20 21 22 23 24 25 26 27 28 29
20 21 22 23 24 25 26 27 28 29 30
21 22 23 24 25 26 27 28 29 30 31
22 23 24 25 26 27 28 29 30 31 32
23 24 25 26 27 28 29 30 31 32 33
24 25 26 27 28 29 30 31 32 33 34
25 26 27 28 29 30 31 32 33 34 35

Overall Sample
13-15 Year Old Sample

Year
84 85 86 87 88 89 90 91 92 93 94
140 140 140 140 140 140
267 267 267 267 267 267 267
251 251 251 251 251 251 251 251
Sample Size for Each Age- Year

260 260 260 260 260 260 260 260 260

223 223 223 223 223 223 223 223 223 223
183 183 183 183 183 183 183 183 183 183 183
198 198 198 198 198 198 198 198 198 198 198
212 212 212 212 212 212 212 212 212 212 212
156 156 156 156 156 156 156 156 156 156 156
158 158 158 158 158 158 158 158 158 158 158
130 130 130 130 130 130 130 130 130 130 130
108 108 108 108 108 108 108 108 108 108 108
120 120 120 120 120 120 120 120 120 120 120
105 105 105 105 105 105 105 105 105 105 105
112 112 112 112 112 112 112 112 112 112 112
105 105 105 105 105 105 105 105 105 105 105
89 89 89 89 89 89 89 89 89 89 89
80 80 80 80 80 80 80 80 80 80 80

37
Figure 4: Variation in Migration Prevalence Rates Across Villages and Time
Female Migrant Prevalence Rates by Village and Year

100%

90%
Migrant Prevalence Rates (% Ever Migrated

80%

70%

60%

50%

40%

30%

20%

10%

0%
1982 1984 1986 1988 1990 1992 1994 1996

Male Migration Prevalence Rate by Village and Year

100%

90%
Migration Prevalence Rate (% Ever Migrated

80%

70%

60%

50%

40%

30%

20%

10%

0%
1982 1984 1986 1988 1990 1992 1994 1996

Migration Prevalence Rates by Village and Year

90%
Migration Prevalence Rates (% Ever Migrated

80%

70%

60%

50%

40%

30%

20%

10%

0%
1982 1984 1986 1988 1990 1992 1994 1996

Each colored dot represents a village coded to reflect a villages ranking in 1984 (blue
lowest and brown highest). Lines are drawn from highest, mid-point, and lowest values
in 1994.

38
Figure 5: Migration Prevalence Rates for Two Sample Villages High Female Migrant
Village and High Male Migration Village.

100%

90%
Migration Prevalence Rates (Ever Migrated

80%

70%

60%

50%

40%

30%

20%

10%

0%
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995

High Female Prevalence Rate Village (Female Rate)

High Female Prevalence Rate Village (Male Rate)
High Male Prevalence Rate Village (Female Rate)
High Male Prevalence Rate Village (Male Rate)

39
 Table 1: Descriptive Statistics for All Variables in 1984 and 1994
Individual Level Means in 1984
13-15 Year Olds Overall Sample
Women Men Women Men
Age 14.074 14.021 18.055 17.827
Married in 1984 0.004 0.009 0.228 0.131
Land Owned <10 rai 0.119 0.134 0.133 0.116
Land 11-25 rai 0.205 0.216 0.192 0.187
Land >25 rai 0.290 0.312 0.279 0.287
Temporary Migrant in 1984 (0/1) 0.051 0.053 0.081 0.122
HH Any Temporary Migrant (0/1) 0.321 0.317 0.317 0.327
% 13-30 yr old Temporary Migrants in Village 47.182 45.651 46.358 46.479
% Female in Temporary Migrants in Village 35.091 35.282 35.449 35.400
Remote (0/1) 0.881 0.819 0.844 0.817
Electricity (0/1) 0.327 0.299 0.321 0.307
No Education (0/1) 0.006 0.007 0.005 0.010
Primary Education (0/1) 0.830 0.776 0.685 0.600
Some Secondary Schooling (0/1) 0.151 0.210 0.258 0.336
Completed Secondary School (0/1) 0.006 0.007 0.038 0.057
Sample Size (person-age) 312 281 968 788
Individual Means in 1994
Ever Migrated Out of Nang Rong by 1994 0.662 0.736 0.592 0.724
# Migration Trips Among Indiv. 1.275 1.501 1.143 1.557
# Migrant Months Among Indiv. 27.302 30.366 27.592 35.030
# Migrant Trips for HH Members Per HH 0.686 0.703 0.659 0.661
# Migrant Months for HH Members 16.671 15.796 16.145 15.921
# Migrant Trips for Vill. Members Per Person 1.340 1.343 1.342 1.345
# Migrant Months for Vill. Members 31.264 31.053 31.338 31.136
# Migrant Trips by Women, Indiv. 1.275 0.000 1.143 0.000
# Migrant Months by Women, Indiv. 27.302 0.000 27.592 0.000
# Migrant Trips for Women HH Members 0.333 0.309 0.299 0.307
# Migrant Months for Women HH Members 8.385 7.318 7.671 7.766
# Migrant Trips for Women Vill. Members 0.607 0.615 0.613 0.619
# Migrant Months for Women Vill. Members 14.728 14.828 14.823 14.887
# Migrant Trips by Men, Indiv. 0.000 1.501 0.000 1.557
# Migrant Months by Men, Indiv. 0.000 30.366 0.000 35.030
# Migrant Trips for Men HH Members 0.354 0.394 0.360 0.353
# Migrant Months for Men HH Members 8.286 8.478 8.474 8.155
# Migrant Trips for Men Vill. Members 0.733 0.729 0.729 0.726
# Migrant Months for Men Vill. Members 16.536 16.225 16.514 16.250
Age 21.533 21.414 25.067 24.651
Married by 1994 (0/1) 0.411 0.169 0.551 0.361
Village Migrant Follow-up Rate 43.733 43.577 43.976 43.684
Village has Electricity (0/1) 1.000 1.000 1.000 1.000
No Education (0/1) 0.031 0.029 0.021 0.025
Primary Education (0/1) 0.818 0.748 0.711 0.634
Some Secondary Schooling (0/1) 0.072 0.093 0.182 0.216
Completed Secondary School (0/1) 0.106 0.153 0.097 0.145
Migration Prevalence Rate (% Ever Migrated) 65.352 65.421 65.203 65.357
Sample Size (person-age) 906 828 1562 1335

40
 Table 2: Random Effects Logistic Estimation of the Odds of Being a Migrant
(Living Outside of Nang Rong), Overall Sample
Model 1 Model 2 Model 3
OR z OR z OR z
Controls
Age 3.429 27.43 *** 2.834 23.24 *** 2.719 22.05 ***
Age*Age 0.975 -27.01 *** 0.978 -23.69 *** 0.978 -22.94 ***
Men 2.346 8.45 *** 1.606 4.97 *** 1.606 5.00 ***
Some 2ndary School 1.408 3.02 ** 1.361 2.91 ** 1.251 2.11
Completed 2ndary School 2.921 12.22 *** 2.731 11.40 *** 2.282 8.61 ***
Married 0.351 -13.06 *** 0.387 -12.29 *** 0.392 -12.14 ***
Own 0-10 Rai of Land 1.260 1.34 1.556 3.09 ** 1.551 3.15 **
Own 10-25 Rai of Land 1.029 0.21 1.216 1.62 1.204 1.53
Own More than 25 Rai 0.887 -0.97 0.906 -0.83 0.897 -0.92
Temporary Migrant 8.543 9.17 *** 4.305 7.09 *** 4.381 7.25 ***
Temporary Migrants in Household 1.495 3.46 ** 1.398 2.90 ** 1.424 3.07 **
% 13-30 Year Olds That are Migrants 1.006 1.64 1.014 4.75 *** 1.011 3.14 **
% 13-30 Year Old Women Migrants 0.991 -2.17 0.998 -0.51 0.999 -0.28
Remote Village 2.247 6.18 *** 1.479 2.80 * 1.472 2.78 *
Migrant Follow Up Rate 1.011 1.22 1.014 1.73 1.019 2.23
Village has Electricity 1.134 1.61 1.079 1.03 1.094 1.11
Migration Prevalence Rate 1.042 19.53 *** 1.031 6.11 *** 1.052 7.35 ***
Cumulative Migration Experience
# Migration Trips Among Indiv. 1.476 11.03 *** 1.477 11.08 ***
# Migrant Months Among Indiv. 1.015 6.91 *** 1.015 7.25 ***
# Migrant Trips for HH Members Per HH 1.405 4.49 *** 1.394 4.45 ***
# Migrant Months for HH Members 1.010 2.80 * 1.010 2.99 **
# Migrant Trips for Vill. Members Per Person 0.987 -0.04 0.510 -1.93
# Migrant Months for Vill. Members 0.977 -2.69 * 0.960 -4.53 ***
Year Effects
1984 1.389 0.90
1985 0.656 -1.37
1986 0.787 -0.85
1987 0.711 -1.33
1988 0.795 -0.98
1999 0.865 -0.69
1990 0.802 -1.21
1991 0.905 -0.65
1992 1.047 0.36
1993 1.424 3.40 **
s.e. (u) 2.703 2.043 2.026
Rho 0.879 0.807 0.804
Wald Chi-square 2623.6 *** 3050.97 *** 3113.11 ***
p<.05, *p<.01, **p<.005, ***p<.001

41
Table 3: Random Effects Logistic Estimation of the Odds of Being a Migrant For Men
and Women, Baseline Model (Overall Sample)
Women Men
OR z OR z
Controls
Age 2.966 17.25 *** 3.999 22.21 ***
Age*Age 0.977 -17.21 *** 0.972 -21.20 ***
Some 2ndary School 1.032 0.20 1.823 4.34 ***
Completed 2ndary School 2.440 7.55 *** 3.348 9.41 ***
Married 0.198 -15.12 *** 0.717 -2.74 *
Own 1-10 Rai of Land 1.448 1.87 1.691 2.33
Own 11-25 Rai of Land 0.819 -1.13 1.411 1.81
Own More than 25 Rai 1.085 0.45 0.940 -0.34
Temporary Migrant 10.909 7.03 *** 2.569 3.21 **
Temporary Migrants in Household 2.052 4.85 *** 1.297 1.29
% 13-30 Year Olds That are Migrants 1.006 1.28 1.017 3.62 ***
% 13-30 Year Old Women Migrants 1.002 0.39 0.975 -3.21 **
Remote Village 2.356 4.89 *** 1.840 3.31 **
Migrant Follow Up Rate 1.005 0.44 1.047 4.10 ***
Village has Electricity 1.127 1.08 1.153 1.28
Migration Prevalence Rate 1.057 18.93 *** 1.035 11.04 ***
s.e. (u) 2.789 2.579
Rho 0.886 0.869
Wald Chi-square 1431.48 *** 1289.73 ***
p<.05, *p<.01, **p<.005, ***p<.001

42
 Table 4: Random Effects Logistic Estimation of the Odds of Being a Migrant For Men and Women, Evaluating
Cumulative Experience and Year Effects, Net of Baseline Controls (Overall Sample)
Women Men
Model 2 Model 3 Model 2 Model 3
OR Z OR Z OR Z OR Z
Cumulated Migration Experience
# Migration Trips Among Indiv. 1.802 9.64*** 1.795 9.54*** 1.266 5.26*** 1.271 5.31***
# Migrant Months Among Indiv. 1.017 5.43*** 1.017 5.40*** 1.008 2.57 1.009 2.92**
# Migrant Trips for HH Members Per 1.272 2.33 1.265 2.27 1.579 3.90*** 1.563 3.91***
HH
# Migrant Months for HH Members 1.006 1.37 1.007 1.46 1.015 2.89** 1.016 2.94**
# Migrant Trips for Vill. Members Per 2.529 2.16 1.432 0.71 0.324 -2.71* 0.160 -3.70***
Person
# Migrant Months for Vill. Members 0.970 -2.54 0.952 -3.78*** 0.988 -1.01 0.972 -2.25
Year Effects
1984 1.801 1.14 1.142 0.25
1985 0.667 -0.92 0.654 -0.96
1986 0.824 -0.48 0.779 -0.62
1987 0.806 -0.59 0.662 -1.12
1988 0.895 -0.33 0.761 -0.81
1989 1.106 0.34 0.727 -1.06
1990 0.896 -0.43 0.754 -1.09
1991 1.145 0.62 0.755 -1.28
1992 1.089 0.47 1.041 0.22
1993 1.530 2.86** 1.374 2.17
Baseline Controls Also Included
s.e. (u) 1.964 1.948 2.19 2.157
Rho 0.79 0.792 0.83 0.823
Wald Chi-square 1658.99 *** 1689.69 *** 1411.6 *** 1435.39 ***
p<.05, *p<.01, **p<.005, ***p<.001

43
Table 5: Evaluating the Gendered Content of Migrant Networks (Random Effects Logistic Estimation of the Odds of
Being a Migrant (Living Outside of Nang Rong)), Overall Sample
All Women Men
OR Z OR z OR z
Cumulative Migration Experience
# Individual Trips 1.476 10.90 *** 1.799 9.47*** 1.279 5.32***
# Individual Months Experience 1.015 7.07 *** 1.017 5.24*** 1.009 2.84*
Female Cumulative Migration Experience
# Migrant Trips for Women HH Members 1.274 2.24 1.202 1.18 1.276 1.52
# Migrant Months for Women HH Members 1.023 4.49 *** 1.016 2.23 1.034 4.21***
# Migrant Trips for Women Vill. Members 0.360 -2.48 1.529 0.72 0.071 -4.54***
# Migrant Months for Women Vill. Members 1.001 0.03 0.944 -2.69* 1.071 3.06**
Male Cumulative Migration Experience
# Migrant Trips for Men HH Members 1.508 4.01 *** 1.303 1.95 2.001 4.25***
# Migrant Months for Men HH Members 1.000 0.09 1.000 -0.01 0.998 -0.27
# Migrant Trips for Men Vill. Members 1.838 1.04 1.146 0.16 2.970 1.31
# Migrant Months for Men Vill. Members 0.910 -5.29 *** 0.960 -1.61 0.858 -5.89***
Year Effects
1984 1.475 1.03 1.699 1.00 1.381 0.61
1985 0.719 -1.03 0.630 -1.01 0.841 -0.39
1986 0.866 -0.49 0.782 -0.59 1.000 0.00
1987 0.788 -0.88 0.767 -0.70 0.862 -0.39
1988 0.886 -0.49 0.855 -0.45 0.994 -0.02
1989 0.961 -0.18 1.062 0.19 0.936 -0.22
1990 0.895 -0.58 0.865 -0.54 0.975 -0.10
1991 1.000 0.00 1.113 0.47 0.954 -0.21
1992 1.137 0.96 1.065 0.33 1.268 1.26
1993 1.497 3.78 *** 1.509 2.70* 1.553 2.93**
Baseline Controls Also Included
s.e. (u) 2.044 1.95 2.173
Rho 0.807 0.792 0.825
Wald Chi-square 3123.08 *** 1691.3 *** 1441.31 ***
p<.05, *p<.01, **p<.005, ***p<.001

44
 Table 6: Random Effects Logistic Estimation of the Odds of Being a Migrant
(Living Outside of Nang Rong), 13-15 Year Olds
Model 1 Model 2 Model 3
OR z OR z OR z
Controls
Age 9.775 19.20 *** 7.261 16.58 *** 7.404 15.28 ***
Age*Age 0.950 -16.58 *** 0.955 -14.91 *** 0.956 -14.00 ***
Men 1.604 3.59 *** 1.232 1.88 1.230 1.87
Some 2ndary School 1.407 2.01 1.369 1.98 1.321 1.65
Completed 2ndary School 2.647 9.70 *** 2.375 8.72 *** 2.226 6.58 ***
Married 0.381 -8.41 *** 0.436 -7.51 *** 0.425 -7.70 ***
Own 0-10 Rai of Land 2.279 4.10 *** 2.026 3.92 *** 2.088 4.07 ***
Own 10-25 Rai of Land 1.234 1.21 1.305 1.76 1.350 1.94
Own More than 25 Rai 1.027 0.14 1.019 0.13 1.056 0.36
Temporary Migrant 1.382 0.93 1.323 0.71 1.211 0.48
Temporary Migrants in Household 1.413 2.03 1.255 1.56 1.207 1.32
% 13-30 Year Olds That are Migrants 1.011 2.08 1.016 3.78 *** 1.006 1.28
% 13-30 Year Old Women Migrants 0.983 -3.30 ** 0.994 -1.39 0.994 -1.37
Remote Village 1.948 3.09 ** 1.442 2.10 1.287 1.43
Migrant Follow Up Rate 1.030 2.84 ** 1.021 2.03 1.011 1.04
Village has Electricity 0.929 -0.65 0.881 16.58 0.907 -0.85
Migration Prevalence Rate 1.029 7.34 *** 1.042 5.36 *** 1.059 0.01
Cumulative Migration Experience
# Migration Trips Among Indiv. 1.614 9.01 *** 1.624 9.27 ***
# Migrant Months Among Indiv. 1.011 3.11 ** 1.013 3.53 ***
# Migrant Trips for HH Members Per HH 1.345 3.18 ** 1.342 3.14 **
# Migrant Months for HH Members 1.013 2.87 ** 1.013 2.80 *
# Migrant Trips for Vill. Members Per Person 0.551 -1.51 0.608 -1.12
# Migrant Months for Vill. Members 0.963 -3.37 ** 0.963 -3.16 **
Year Effects
1984 11.431 4.34 ***
1985 4.081 2.97 **
1986 4.293 3.45 **
1987 3.417 3.26 **
1988 3.119 3.38 **
1989 2.831 3.52 ***
1990 2.028 2.82 *
1991 1.866 3.00 **
1992 1.641 2.96 **
1993 1.661 3.77 ***
s.e. (u) 2.607 1.928 1.879
Rho 0.872 0.788 0.779
Wald Chi-square 1898.39 *** 2145.04 *** 2187.43 ***
p<.05, *p<.01, **p<.005, ***p<.001

45
 Table 7: Random Effects Logistic Estimation of the Odds of Being a Migrant For Men
and Women, Baseline Model (13-15 Year Olds)
Women Men
OR z OR z
Controls
Age 7.753 12.23 *** 12.971 14.38 ***
Age*Age 0.955 -10.76 *** 0.945 -12.57 ***
Some 2ndary School 1.247 0.79 1.492 1.70
Completed 2ndary School 2.291 6.03 *** 2.770 6.46 ***
Married 0.261 -9.13 *** 0.802 -1.08
Own 1-10 Rai of Land 2.657 3.26 ** 2.089 2.44
Own 11-25 Rai of Land 0.901 -0.43 1.741 2.29
Own More than 25 Rai 0.943 -0.22 1.136 0.48
Temporary Migrant 1.856 . . 1.395 0.71
Temporary Migrants in Household 1.819 2.15 1.296 1.07
% 13-30 Year Olds That are Migrants 1.007 0.75 1.019 2.62 *
% 13-30 Year Old Women Migrants 0.994 -0.79 0.980 -2.74 *
Remote Village 2.242 2.99 ** 0.795 -0.58
Migrant Follow Up Rate 1.013 0.85 1.048 3.27 **
Village has Electricity 0.791 -1.48 1.073 0.42
Migration Prevalence Rate 1.044 7.53 *** 1.016 3.02 **
s.e. (u) 2.619 2.569
Rho 0.873 0.868
Wald Chi-square 955.00 *** 940.54 ***

p<.05, *p<.01, **p<.005, ***p<.001

46
 Table 8: Random Effects Logistic Estimation of the Odds of Being a Migrant For Men and Women, Evaluating
Cumulative Experience and Year Effects, Net of Baseline Controls (13-15 Year Olds)
Women Men
Model 2 Model 3 Model 2 Model 3
OR Z OR Z OR Z OR Z
Cumulated Migration Experience
# Migration Trips Among Indiv. 1.726 7.08*** 1.729 7.15*** 1.481 5.45*** 1.477 5.45***
# Migrant Months Among Indiv. 1.023 4.72*** 1.024 4.85*** 0.996 -0.77 0.998 -0.34
# Migrant Trips for HH Members Per 1.378 2.54 1.368 2.49 1.289 1.91 1.314 2.00
HH
# Migrant Months for HH Members 1.008 1.39 1.008 1.35 1.019 2.49 1.018 2.50
# Migrant Trips for Vill. Members Per 1.008 0.01 0.984 -0.03 0.308 -2.01 0.404 -1.39
Person
# Migrant Months for Vill. Members 0.966 -2.18 0.963 -2.24 0.957 -2.73* 0.960 -2.38
Year Effects
1984 6.898 2.63* 22.190 3.59***
1985 2.326 1.35 7.671 2.82*
1986 2.713 1.79 7.119 3.05**
1987 2.303 1.68 5.311 2.91**
1988 2.057 1.63 5.043 3.16**
1989 2.218 2.05 3.779 2.96**
1990 1.430 1.08 3.027 2.91**
1991 1.581 1.65 2.309 2.66*
1992 1.172 0.70 2.400 3.51***
1993 1.323 1.48 2.189 4.04***
Baseline Controls Also Included
s.e. (u) 1.693 1.672 2.238 2.218
Rho 0.741 0.736 0.834 0.831
Wald Chi-square 1141.03 *** 1154.83 1004.21 *** 1034.88 ***
p<.05, *p<.01, **p<.005, ***p<.001

47
 Table 9: Evaluating the Gendered Content of Migrant Networks (Random Effects Logistic Estimation of the Odds of
Being a Migrant (Living Outside of Nang Rong)), 13-15 Year Olds
All Women Men
OR Z OR z OR z
Cumulative Migration Experience
# Individual Trips 1.627 9.25 *** 1.734 7.11*** 1.464 5.33***
# Individual Months Experience 1.013 3.49 *** 1.024 4.91*** 0.998 -0.29
Female Cumulative Migration Experience
# Migrant Trips for Women HH Members 1.449 2.82 * 1.603 2.54 1.287 1.28
# Migrant Months for Women HH Members 1.015 2.33 1.004 0.49 1.030 2.09
# Migrant Trips for Women Vill. Members 0.258 -2.57 0.648 -0.59 0.119 -2.80*
# Migrant Months for Women Vill. Members 1.019 0.94 1.017 0.60 1.015 0.48
Male Cumulative Migration Experience
# Migrant Trips for Men HH Members 1.012 1.92 0.901 -3.28** 1.343 1.63
# Migrant Months for Men HH Members 1.233 1.65 1.014 1.69 1.007 0.82
# Migrant Trips for Men Vill. Members 5.324 2.22 3.882 1.31 7.561 1.77
# Migrant Months for Men Vill. Members 0.899 -4.58 *** 1.193 1.07 0.897 -2.92**
Year Effects
19849 14.548 4.68 *** 7.649 2.73* 32.763 3.84***
1985 5.315 3.44 ** 2.622 1.52 11.721 3.22**
1986 5.555 3.95 *** 3.060 1.97 10.719 3.49***
1987 4.429 3.84 *** 2.617 1.90 7.925 3.41**
1988 4.009 4.00 *** 2.336 1.87 7.386 3.70***
1989 3.575 4.18 *** 2.504 2.31 5.361 3.54***
1990 2.548 3.59 *** 1.613 1.39 4.247 3.57***
1991 2.274 3.79 *** 1.759 1.97 3.090 3.36**
1992 1.927 3.76 *** 1.278 1.04 3.046 4.17***
1993 1.820 4.36 *** 1.395 1.73 2.498 4.53***
Baseline Controls Also Included
s.e. (u) 1.892 1.67 2.242
Rho 0.781 0.736 0.834
Wald Chi-square 2185.57 *** 1151.92 *** 1027.97 ***
p<.05, *p<.01, **p<.005, ***p<.001

9
Omitted year category for women is 1984, not 1994.

48