You are on page 1of 26

Selectivity, Adaptation, or Disruption?

A Comparison of Alternative Hypotheses on the


Effects of Migration on Fertility: The Case of Brazil
Author(s): Hugo M. Hervitz
Source: International Migration Review, Vol. 19, No. 2 (Summer, 1985), pp. 293-317
Published by: The Center for Migration Studies of New York, Inc.
Stable URL: http://www.jstor.org/stable/2545774 .
Accessed: 17/06/2014 08:45

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Center for Migration Studies of New York, Inc. is collaborating with JSTOR to digitize, preserve and
extend access to International Migration Review.

http://www.jstor.org

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
or
Selectivity, Adaptation,

A of
Disruption? Comparison

Alternative on the
Hypotheses

Effects of on
Migration Fertility:

The Case of Brazil

Hugo M. Hervitz
Barry University,Miami

Several major hypotheses have been proposed to explain and predict


the effect of migration on fertility. This article conducts a highly
disaggregated analysis of migrant and stayer fertility, using Brazilian
data for 1976, to ascertain the extent to which the observed patterns of
fertility variations are compatible with one or another hypothesis. The
study controls for origin, destination, and recency of migration, as well
as for age, education, and marital status of migrants and stayers. The
regional breakdown consists of a cross-classification of rural/urban
status with modern/traditional categories based on the level of economic
development of the region.

The fertility impact of inter-regional migration in developing countries has


not received a great deal of attention in either the migration literature, or
the fertility-determinants literature. This is surprising given the potentially
large impact of internal migration on national fertility levels, as well as on
inter-regional fertility differentials.
At the theoretical level, a number of hypotheses have been proposed
suggesting various patterns of fertility behavior following migration.
However, at the empirical level the available evidence has been quite limited.
The four hypotheses most commonly examined in the literature focus on
processes of socialization, adaptation, selectivity, and disruption.
The socialization hypothesis premises the belief that the fertility behavior
of migrants reflects the fertility preferences prevalent in their childhood
environment. Rural-urban migrants are expected to exhibit levels of fertility
similar to those of rural stayers and convergence towards the lower fertility
levels of urban stayers is expected to occur only after at least a generation has
elapsed.
The adaptation hypothesis assumes that the fertilitypreferences of migrants
gradually adapt to the new economic, social and cultural environments at

IMR Volume xix, No. 2 293

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
294 International Migration Review

places of destination. Convergence to the fertility levels of stayers at


destination is expected to take place rather rapidly (usually meaning less
than ten years).
The selectivity hypothesis focuses on the well established fact that migrants
are not a random sample of the population at their place of origin. Since
migrants typically constitute a selected group in terms of age, education,
marital status, or occupation, among others, they should be expected to
possess fertility preferences different from those of the overall population at
origin. Only if relevant socioeconomic and demographic characteristics that
distinguish migrants from stayers are controlled would it be expected that
migrant-stayer comparisons would tend to be insignificant. Indeed, Ribe
and Schultz (1980) go further and have suggested that migrants have distinct
unobserved preferences for family size, quite in addition to other dif?
ferentiating characteristics, such that it is fertility propensities which
determine the choice of destination areas, rather than the choice of area
which determines fertility behavior. Thus, migrants with a preference for
smaller families are expected to migrate to low-fertility areas, where the
environment and relative costs or benefits are conducive to smaller families,
while high fertility areas should be expected to attract migrants who prefer
large families.
The disruption hypothesis suggests that in a period immediately following
a change of residence migrants would show a particularly low level of
fertility, due to disruptive factors associated with the migration process or
with the likelihood that women would not migrate while pregnant. Two
factors of disruption usually mentioned are the physiological consequences
of the stressful situation typically associated with moving and the fairly
common separation of spouses during early stages of the migration process.
The suggested drop in fertility attributable to disruption is expected to be
only temporary, and a more normal or even somewhat accelerated pace of
fertility is expected to be resumed gradually.
Whether fertility goals of migrants are similar to those of stayers at their
place of origin or to residents at the place of destination would presumably
depend on the extent to which migrants are subject to socialization, adaptation,
selectivity, or disruption effects. In principle, the four hypotheses could
lead to rather different conclusions concerning the long-run and short-run
fertility impacts of migration and the advisable policy interventions. Thus,
it is of extreme importance to examine the available empirical evidence in
search of support for one or another hypothesis. Unfortunately, existing
indications are quite limited in scope and with fairly mixed results.
Evidence supporting the socialization hypothesis was provided originally
by Goldberg (1959, 1960) and Duncan (1965), who analyzed the fertility
behavior of rural-urban migrants in the U.S. Edmonston (1976), in a study
of migrants to Bogota, also obtained evidence supporting this hypothesis.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 295

However, McGirr and Hirschman (1979), in a study of U.S. cohorts born


after 1910, did not find persuasive evidence to this effect.
Evidence supporting the adaptation hypothesis is provided in a recent
study by Lee and Farber (1982) on Korean migration, as well as in various
studies by Goldstein of Thailand (Goldstein, 1978; Goldstein and Goldstein,
1981). Other support is provided by Martine (1975), Park and Park (1976),
Macisco et al (1969), Hendershot (1976), and Hiday (1978), for Colombia,
Costa Rica, Korea, Puerto Rico and the Philippines, respectively.
Finally, Ribe and Schultz (1980) report what they regard as substantial
evidence supporting their particular version of the selectivity hypothesis,
applied to migration data for Colombia.
Evidence supporting the disruption hypothesis is presented in two studies
by Bach on Malaysia (1981), by Goldstein on Thailand (Goldstein, 1973 and
1978) and by Goldstein and Goldstein (1982).
A pervasive problem in the literature is the limited attention paid to
simultaneous disaggregations of migration flows by characteristics of migrants
and those relating to places of origin and destination. With respect to the
latter, most studies focus only on a broad urban-rural dichotomy, and even
then the main emphasis is usually placed on just one type of flow, from rural
to urban. Urban to rural, urban to urban and rural to rural flows are
disregarded as a rule, as are size-of-place disaggregations for any of these
alternative migration streams. Similarly with respect to migrant charac?
teristics, there is ample evidence to suggest the possibility of variable migrant
fertility behavior by education and marital status, as well as by length of stay
at place of destination. Presumably, studies that do not control for these
factors in combination with disaggregated origin-destination controls could
well be subject to significant biases.
To test this proposition is the purpose of this article, which presents
detailed comparisons of migrant and stayer fertility differentials in Brazil at
uncommonly high levels of disaggregation. Specifically the main objectives
of this study are:

1) to compare the fertility levels of migrants with those of stayers at both


origin and destination areas, after controlling for age, education, marital
status, and residence durations by migrants at areas of destination; and

2) to establish on this basis the extent to which observed patterns of fertility


differentials lend support to each of the four hypothesis discussed above.
Section III of this article presents a general overview of the major
demographic, regional, and economic characteristics of Brazil. Section III
describes the data set utilized and details the various measures to appear and
their definitions. Section IV reports the main empirical findings with respect
to the above objectives. Finally, Section V discusses some suggested fruitful
directions of future research.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
296 International Migration Review

0 VER VIEW OF THE BRAZILIAN DEMOGRAPHIC,


GEOGRAPHIC AND ECONOMIC DEVELOPMENT

Brazil is the most populous country in Latin America ? accounting for one
third of its population ? and one of the most populous countries in the
world. In the post-war period the rate of population growth has been high,
ranging from 2.5 to 3 percent, resulting in the more than doubling of the
? from 52 million in 1950 to 120 million in
population size in thirty years
1980.
In recent years, though, the country has been experiencing dramatic
declines in fertility rates. In 1980 the total fertility rate stood at 4.11, a full 30
percent frop from the 1970 rate of 5.83 (Merrick and Berquo, 1983). This
decline has been attributed to increased fertility control within marriage ?
? and
contraception, sterilization, and abortion especially to the spread of
these controls to lower-income regions and groups that had not participated
in previous fertility declines. Starting in the mid-1960s, these groups
experienced socioeconomic changes that were conducive to smaller family
norms, such as increased educational attainment, increased ownership of
consumer durables (especially TV sets), and increased female labor force
participation (Merrick and Berquo, 1983).
As is the case in many other developing countries, Brazil has a dual
economic structure with a significant degree of concentration of economic
activity in the Southeast, around the metropolis of Sao Paulo and Rio de
Janeiro, contrasting with poor and largely underdeveloped regions else?
where, especially in the Northeast. This region's per capita income stands at
less than half the national average of $1,600 (1980) and is vulnerable to
periodic droughts, which contributes to its persistent rural poverty. Gen?
eralized poverty and acute scarcity of land have resulted in substantial out?
migration from the Northeast for several decades. Traditionally, the des?
tination has been the metropolis of the Southeast, but in the last two decades
increasing numbers of migrants have turned to the newly-developed agri?
cultural frontier areas of the West and North.
The extreme degree of inter-regional dualism along economic development
lines, manifests itself as well in the demographic realm. Fertility rate
differentials between the Southeast and the Northeast have always been
large, and, moreover, have increased significantly in the last three decades.
Total fertility rates (TFR) in the Northeast have been 50 percent higher than
in the Southeast in 1950, 53 percent in 1960, 64 percent in 1970, and 73
percent in both 1976 and 1980.
The widening of the inter-regional fertility gap is due to the fact that the
Southeast has experienced more drastic overall fertility declines than the
Northeast: the TFR in the Southeast dropped by 37 percent between 1950
and 1980 ? from 4.94 to 3.10 ? while in the Northeast it dropped in the same

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 297

? from 7.39 to 5.36.


period by 27 percent
In addition to the significant inter-regional fertility gaps, Brazil has a
large rural/urban fertility gap. In 1950 the rural TFR was 65 percent higher
than its urban counterpart (7.70 and 4.68 respectively); in 1970 it was 67
percent higher (7.71 and 4.61): and in 1976 it was a whopping 83 percent
higher (6.36 and 3.48).l
One of the difficulties in assessing inter-regional differentials along
'modernY'traditional' lines versus urban/rural differentials is that it is usually
not clear from the data the extent of overlapping between the two classifi?
cations. This is important since the proper assessment of the importance of
regional levels of economic development on inter-regional fertility dif?
ferentials is essential to 'control' for urban and rural status. As discussed in
the next section, the regional breakdown of this study attempts to overcome
this problem by focusing on cross-classifications, i.e., 'modern-urban',
'modern-rural', 'traditional-urban', and 'traditional-rural' regions, as the
basic geographical units of analysis.

DA TA DESCRIPTION AND DEFINITIONS

The data used in this study come from a detailed and comprehensive
household survey carried out in 1976: the 'Pesquisa Nacional de Amostra
Domiciliar' or PNAD. Various versions of the PNAD surveys have been
carried out since 1967 by the Brazilian Bureau of the Census ('Fundacao
Instituto Brasileiro de Geografia e Estatistica'), but the 1976 survey has been
the most thorough one. The quality and reliability of the data are generally
considered to be good and above average for a developing country. The
PNAD surveys constitute the major source of demographic data in-between
decennial censuses and have become increasingly more relied upon given
the rapidly changing demographic environment in Brazil.
The 1976 PNAD covered 99,167 households with 393,871 individuals.
This represents an average sample fraction of .35 percent. Each household
entry included 27 data items, and each individual entry 94. The geographical
coverage was good, including all of Brazil's states and territories, and it was
based on a useful regional breakdown of the country into seven PNAD
regions:

(1) Rio de Janeiro (5) Northeastern States


(2) Sao Paulo (6) Brasilia
(3) Southern States (7) Frontier States
(4) Minas Gerais/Espirito Santo

1The figurespresentedin thelastthree


paragraphshave been derivedfromdata providedin
Merrickand Berquo (1983).Both'Southeast'and 'Northeast'have been definedin a broadsense,

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
298 International Migration Review

Rural areas in the first five regions were well covered. Region 6 is only
urban by definition, and in region 7 rural areas were not covered at all,
supposedly because of the sparse population and the difficulties of access to
rural jungle areas. Further technical details about the surveys are available
in Brazil (1978) and Brazil (1980).
The present study makes use of information for a subsample of 27,314
women in the childbearing age interval 15-44, all married at the time of the
survey. The PNAD regional breakdown was consolidated into three categories
reflecting three different levels and types of economic development and
demographic patterns:
1) the 'modern' states of the Center-South, incorporating PNAD regions 1,
2, and 3, which have historically constituted the pole of economic development
of the country, with relatively high per-capita income levels and relatively
low fertility rates;

2) the 'traditional' states of the Northeast, incorporating PNAD regions 4


and 5, which have historically been the most under-developed part of the
country, with low per-capita income levels and high fertility rates;2 and

3) the 'frontier' states of the West and North, incorporating PNAD regions
6 and 7, dominated by the jungle and only recently populated, with a
relatively low level of economic dvelopment and relatively high fertility
rates.
The above three-way classification was combined with a two-way urban/
rural breakdown giving rise to the following basic geographic units of
analysis:
1) urban areas of modern states (MU)
2) rural areas of modern states (MR)
3) urban areas of traditional states (TU)
4) rural areas of traditional states (TR)
5) urban areas of frontier states (FU)
Rural areas of frontier states could not be analyzed because they were not
covered by the survey.
Since each of the five basic units can be an origin or a destination for
migration, 25 different migration flows could be considered. Women were
classified as migrants if born in a municipio (county) different from their

withthe formerincludingthe Statesof Rio de Janeiroand Sao Paulo as well as the Southern
States,and thelatterincludingtheNortheasternStatesplus Minas Gerais and EspiritoSanto.
2
Region 4 is a 'borderline'case in-betweenmodernand traditional;it was included in the
traditionalcategorybecause of its historicaldemographicprofile,which,beforethe sharp
changesthattookplace in the 1970s,morecloselyresembledthetraditionalpattern.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 299

current municipio of residence. Otherwise they were classified as 'stayers'.


Migrants within a region (e.g. MU to MU) were those born in a municipio
different from their current one but within the same geographic unit among
the five considered.
The fertility measure utilized throughout is that of Children Ever Born
(CEB).
Migrants and stayers were disaggregated by characteristics as follows:

l)Age
?
young (Y), 15-24 years;
?
mid-age (M), 25-34 years;
? older (O), 35-44
years.
2) Education
? low (L), 0-4
years of schooling;
? some (S), 5-10
years of schooling;
? 11 4-
high (H), years of schooling.

Migrant women were further classified by duration of stay at their place of


residence.

3) Duration
? moved to current municipio less than 2
very recent migrant (VRM),
years ago;
?new migrants (NEW), moved to current municipio 2 or more years
ago but less than 6 years ago;
?veteran migrants (VET), moved to current municipio 6 or more
years ago.
In the next section, the main findings of the study are presented and the
evidence in favor or against the four migration-fertility hypotheses discussed
earlier is analyzed.

EMPIRICAL FINDINGS

Analysis of Migration Patterns

Before we turn to the analysis of fertility rates, it is illustrative to examine the


overall pattern of migration by region of origin and destination, for the
subsample utilized in this study, i.e. married women aged 15 to 44. Table 1
presents the corresponding place-to-place migration matrix.
From Table 1 we can assess the relative importance of the rural/urban
dimension versus the modern/traditional breakdown. With respect to the
former it is important to note that only 31 percent of all migrants moved
from rural to urban areas while 45 percent of the moves were betwen urban
areas. Also, a not so insignificant 20 percent of moves took place between
rural areas. This finding highlights the inappropriateness of the almost

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
300 International Migration Review

TABLE 1
Number of Migrants from Origin X to Destination Y
as a Percentage of Total Migration

Region ofDestination
Modern Modern Traditional Traditional Frontier
Origin Urban Rural Urban Rural Urban

MU 14.27 1.24 .45 .06 .74


MR 10.56 6.66 .12 .10 .36
TU 13.89 2.71 7.99 1.22 1.62
TR 9.52 3.54 8.81 8.23 1.58
FU 2.29 .54 .16 .02 3.31

exclusive concern in the literature with rural-urban flows.


A further revealing finding is that among the rural-to-urban migrants
only a third migrated from traditional to modern regions. The remaining
two-thirds moved within modern and traditional regions. This suggests that
a significant fraction of migrants do not cross either rural/urban or
modern/traditional lines. Actually, that fraction can be computed by adding
up the values of the diagonal in Table 1, which shows that two fifths of all
migrants are in this category.
Among urban-to-urban migrants, 60 percent moved within their own
region, 30 percent moved from traditional to modern areas, and 10 percent
moved to frontier areas.
Among rural-to-rural migrants, 80 percent moved within their own region,
while 20 percent moved from traditional to modern areas.
To obtain a different perspective on Table 1 consider the modern/tradi?
tional breakdown first. Thirty percent of total migration took place from
traditional to modern areas while less than one percent occurred in the
opposite direction. Thirty percent of migrants moved between modern
areas, and the same percentage between traditional areas.
Among the traditional-to-modern movers half were urban-to-urban movers
and only a third moved from rural to urban areas.
It is clear from the table that the modern-urban region is a large net-
recipient of migration serving as a destination for half the migrant population
while generating only a sixth of total migration. On the other hand the
traditional-rural region is a large net-generator of migration since one third
of all migrants originate in that region while only one tenth choose it as their
destination.
It is interesting to note that the migration deficit of the traditional-rural
region amounts to 65 percent of the migration surplus of the modern-urban

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 301

? as has been
region. However, it is erroneous to conclude from this figure
done in the past ? that the migration flow from TR to MU must account for
two-thirds of total migration. In fact, Table 1 shows that migration from TR
to MU accounted for just 10 percent of total migration. The simple expla?
nation of this apparent paradox lies in the fact that even though the
traditional-rural region experienced substantial out-migration, only 40
percent of it was directed towards the modern-urban region; similarly, the
modern-urban region experienced substantial in-migration but only 26
percent of it came from TR.

Partially-Disaggregated Analysis of
Migrant and Stayer Fertility

Before we engage in the fully-disaggregated comparison of the fertility of


migrants and stayers, controlling for age, education, recency of migration,
origin, and destination, it might be helpful to obtain a general overview of
the fertility differentials by controlling just for region of origin and destina?
tion. Table 2 presents the fertility levels of migrants in each of the 25
place-to-place cells with significant migration levels (greater than 1% of total
migration), as well as the fertility levels of stayers in each of the five regions.
To facilitate comparisons between migrant and stayer fertility levels, a
statistic "H" is introduced to measure the relative magnitude of the fertility
change experienced by migrants, with respect to the fertility of non-migrants
at origin and destination. Specifically, "H" compares the deviation of migrant
fertility from the "norm" of comparable individuals back in their region of
origin, with the overall "fertility-gap" between the regions of origin and
destination.
As an illustration, consider two hypothetical regions "X" and "Y". Suppose
that the average fertility rate (CEB) at X is 5.0 and at Y is 3.5 Thus, the
fertility-gap between the two regions is 1.5. Now suppose that the fertility
rate of migrants from X living in Y is 4.0. We see that these migrants have a
fertilityrate which is 1.0 lower than the "norm" at their region of origin, which
suggests a fertility reduction of 1.0. The H-statistic assesses the magnitude of
that suggested fertility reduction in relation to the overall fertility gap
between the two regions (in this example 1.5). The statistic is computed as
1.0/1.5 = .66, and is interpreted as meaning that these migrants have "bridged"
two-thirds of the fertility-gap between the two regions.

Formally, H is defined as follows:


=
H(i,j,k) (FS(i,k) - FM (i,j,k))/(FS(i,k) - FS
(j,k))
where,
i: region of origin
j: region of destination

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
302 International Migration Review

k: type of population analyzed (by socioeconomic or other charac?


teristics)
FS(i,k): fertility rate of stayers of type k in region i
FM (i,j,k): fertility rate of individuals of type k who migrated from
region i to region j
Each of the hypotheses discussed earlier (selectivity, adaptation, social?
ization, and disruption) can be usefully assessed in terms of the H-statistic.
The "strong" selectivity hypothesis proposed by Ribe and Schultz (1980)
would imply that H should be close to, or even larger than 1.00, since
migrants from high to low fertility areas are hypothesized to have a prefer?
ence for low fertility even before migrating, and this should be true for both
recent and veteran migrant
The adaptation hypothesis suggests that migrants gradually adjust to the
fertility levels prevalent at destination, thus H should be expected to be low
(close to zero) for recent migrants and move closer to unity for veteran
migrants.
The disruption hypothesis would imply that for recent migrants from
high to low fertility areas the H-statistic should be high (close to unity), but
for more veteran migrants - less affected by temporary disruptive effects - H
should be quite low (close to zero).
Finally, the socialization hypothesis predicts no significant change in the
fertility of migrants compared to that of stayers at origin, thus H should be
close to zero for both recent and veteran migrants.

TABLE 2
Children Ever Born to Migrants By Region of Origin
and Destination, and to Stayers by Region of Residence

REGION OF DESTINATION
Modern Modern Tradit. Tradit. Frontier
Urban Rural Urban Rural Urban

MU 2.598 2.966 a a a
MR 3.240 3.916 a a a
TU 2.945 3.567 3.548 4.144 4.120
TR 3.566 4.703 4.450 5.042 4.598
FU 2.796 a a a 3.381
STAYERS
AT: MU MR TU TR FU

2.276 3.658 3.629 4.664 3.346


Note: a migrationflow(less than1% oftotalmigration).
Insignificant

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 303

Table 3 presents the values of the H-statistic for the fertility rates in Table
2. Notice that the statistic is useful only to analyze situations in which there is
a fertility gap between origin and destination; otherwise, the statistic is
undefined.
The interpretation of the information in Table 3 can only be partial and
tentative since it includes no controls for age, education, and recency of
migration. In particular, the latter omission prevents a full assessment of the
four hypotheses at this point. Yet, the main patterns that emerge from the
table are the following:

a) There appears to be no support for the strong selectivity hypothesis,


since in not one case is the H-statistic anywhere close to unity.

b) Migrants to the modern-urban region appear to experience partial


adaptation to the fertility levels prevalent at destination. The degree of
adaptation, however, appears to vary according to the region of origin,
with migrants from urban areas bridging 50 percent of the fertility gap
and those from rural areas bridging, on average, only 38 percent of the
gap.

c) Migrants from traditional areas moving to regions other than modern-


urban appear to experience hardly any significant fertility reduction.
This finding is compatible with the socialization hypothesis. However,
other migrant groups do not show a similar pattern.

d) Urban-to-rural migrants constitute a special case since they are moving

TABLE 3
Values of the H-Statistic By Region of Origin and Destination

region of destination
Modern Modern Tradit. Tradit. Frontier
Urban Rural Urban Rural Urban

ORIGIN

MU b .50 a a a
MR .30 b a a a
TU .51 b b .50 -1.73
TR .46 -.04 .21 b .05
FU .51 a a a b

Note: a Insignificant
migrationflows(less than1% oftotalmigration)
D Statisticundefined.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
304 International Migration Review

to destinations with higher fertility levels than their origin. The fact
that they show a pattern of half-way adaptation to the higher fertility
levels (comparable to the magnitude discussed in [c] above), suggests
that adaptation appears to be at work quite symmetrically for moves to?
ward lower as well as higher fertility regions.

e) Migrants from traditional-rural areas to frontier-urban areas show a


surprisingly high fertility rate, even higher than that prevalent at their
region of origin.

f) Nothing can be said at this point about the disruption hypothesis since
there is no breakdown of migrants by recency of migration.

In the following section, migrant and stayer fertility will be analyzed in a


fully-disaggregated way. The analysis will focus only on those categories
which involved a significant number of migrants, moving between regions
with different fertility levels.

Fully-Disaggregated Analysis of
Migrant and Stayer Fertility

To be able to uncover the main patterns of migrant and stayer fertility


differentials, the major flows of migrants by origin and destination were
disaggregated by recency of migration, education, and age. Likewise, for
comparison purposes, stayers at origin and destination were disaggregated
by education and age. The definitions of the various categories was presented
in section III above.
Tables 4 through 11 present the fertility levels of migrants and stayers for
the following origin/destination flows:

Table # Origin Destination

4 Modern-Rural Modern-Urban
5 Traditional-Urban Modern-Urban
6 Traditional-Rural Modern-Urban
7 Frontier-Urban Modern-Urban
8 Modern-Urban Modern-Rural
9 Traditional-Urban Traditional-Rural
10 Traditional-Rural Modern-Rural
11 Traditional-Rural Traditional-Urban
12 Traditional-Rural Frontier-Urban
The information in these tables is conveniently summarized by means of
the H-statistic. Table 13 presents the values of the H-statistic by region of
origin and destination, recency of migration, age, and education. The main

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 305

patterns observed in Table 13 are these:

1) There appears to be little support for the strong selectivity hypothesis


since for none but two of the 17 categories analyzed is the H-statistic
consistently close to or above unity.

2) The first exception to the above is the pattern observed for young and
older migrants with some education, who moved from the traditional-
rural to the traditional-urban sector; for this group H is consistently
larger than unity.

3) The second exception is the case of urban-to-rural migrants in the


modern sector. This is a special case since it involves migration towards a
higher fertility region. Young and mid-age migrants in this category
show H-values close to or above unity.

4) The one additional important rural-to-urban flow?in the traditional


sector?supports to some extent the pattern orbserved in 3) since the
H-value for mid-age veteran migrants is significantly above unity. It is
important to note that older urban-rural migrants in both traditional
and modern sectors do not share the same paterrn. Rather, their H-values
are in the .30 to .60 range.

5) In many categories the H-value is high ? close to or above unity?for very


recent and new migrants but it declines substantially for veteran migrants
(in several cases even to negative values). This pattern is compatible with
the presence of significant disruption effects which cause a temporary
decline in migrant fertility.

6) Most migrants to the modern-urban region with low or some education


show a pattern of partial adaptation to the lower fertility levels at destina?
tion. The values of the H-statistic for veterans are in the .50 to .70 range,
implying that migrants bridge half to two-thirds of the fertility gap
between origin and destination. However the path of this adaptation is
not typically smooth but rather it is subject to fluctuations for very recent
and new migrants that appear to be attributable to disruption effects.

7) Migrants from the frontier-urban sector, as well as rural-to-rural mi?


grants moving from traditional to modern areas, appear to maintain their
high fertility levels without any lasting reduction. This is suggested by
H-values which are close to zero or negative, and is compatible with the
socialization hypothesis.

8) A somewhat similar pattern of no lasting fertility reduction is apparent


also for two additional categories. Better educated migrants moving from
traditional to modern urban areas, appear to be subject to short-term
disruption effects but hardly any permanent reduction in fertility is

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
306 International Migration Review

evident, since the H-values become quite low or even negative for veteran
migrants. The same pattern appears for young migrants with low education
moving from modern or traditional rural areas to modern urban areas.
9) It is evident from a comparison of the main findings above with those
derived from the non-disaggregated analysis in the previous section, that
some of the patterns observed in the latter case do not really hold for all
categories once disaggregation is carried-out.

OVERVIEW OF EMPIRICAL FINDINGS

The above analysis of migrant and stayer fertility differentials, carried-out


at an uncommonly detailed degree of disaggregation, permits the uncov?
ering of interesting patterns related to fertility differentials in general, and
evidence supporting the main migration-fertility hypothesis in particular.
The major overall conclusions can be summarized as follows:

1) National dichotomous urban-rural classsifications are likely to be insuffi?


cient and be misleading for judging probable migration impacts on
fertility, in that fertility differentials between migrants and stayers may
well vary significantly according to regional levels or patterns of economic
development.
2) It appears that rural-to-urban migration flows need to be disaggregated
into various modern/traditional cross-classifications, and that more em?
phasis should be placed on rural-to-urban, urban-to-urban, and rural-
to-rural flows.

3) Duration-of-stay information appears to be essential for assessing


existing main hypotheses concerning migration effects on fertility.
4) No robust quantitative measures of migrant-stayer fertility differentials
were observed to hold across migrant groups. It follows that migrants
differing in terms of age, education, origin, and destination, are likely to
behave in significantly variable fashion with respect to stayer standards of
fertility behavior.
5) Migrant groups with overall lower fertility levels such as the young and
the better educated, are less likely to experience significant fertility
reduction to bridge the origin/destination fertility gap.

6) Rural-to-rural migrants do not appear to experience any lasting fertility


reduction even when they move to areas with lower overall fertility rates.

7) Urban-to-rural migrants tend to bridge a larger fraction of the "uphill"


fertility gap than rural-to-urban migrants do with the "downhill" gap.

8) The evidence quite strongly suggests the presence of significant

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 307

disruption effects affecting many?but not all ?migrant categories.

9) There is little support for the strong selectivity hypothesis proposed by


Ribe and Schultz.

10) There is substantial evidence of partial, but not complete, adaptation


for most migrant categories, once disruption effects disappear.

11) Evidence consistent with the socialization hypothesis (no fertility


reduction for at least a generation) was apparent for migrants origi?
nating in the least developed parts of the country, the frontier region
and the traditional-rural region. In the latter case, however, the
rigidity had a clear destination pattern since it appeared only for
migrants to other rural areas.

SUGGESTIONS FOR FUTURE RESEARCH

The essential purpose of this paper was to see whether a significant increase
in the degree of disaggregation made possible by an unusually rich data set
could help identify new factual regularities ? or assess expected regulari-

TABLE 4
CEB of Migrants From the Modern-Rural to Modern-Urban Sector
by Recency of Migration, Education, and Age, and CEB of Stayers
at Origin and Destination by Education and Age

MID-
YOUNG AGE OLDER

LOW EDUCATION

Stayersat Origin 1.551 3.605 5.631


VeryrecentMigrants 1.267 3.357 4.763
New Migrants 1.241 2.801 5.041
VeteranMigrants 1.550 3.151 4.441
Stayersat Destination 1.343 2.756 3.849
SOME EDUCATION

Stayersat Origin 1.039 2.513 4.731


VeryRecentMigrants .786 1.421 a
New Migrants 1.226 1.872 3.300
VeteranMigrants 1.263 2.247 3.132
Stayersat Destination 1.036 2.043 2.840
HIGH EDUCATION a

Note: a Insignificant
category.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
308 International Migration Review

ties?in LDC migration patterns. Despite the depth of the available data, the
analysis was beset by a number of limitations, most especially its inability to
distinguish between a lifetime-oriented definition of migration and more
time-specific orientation. Lack of information on pre-migration and post-
migration economic characteristics and fertility histories was another cause
of major analytic limitations.
Future work should try to incorporate one or more of the following leads
to needed further disaggregation.

1) Urban-rural classifications should be refined to reflect modern-tradi?


tional distinctions and information on size of place.

2) Within urban sectors, formal and informal sub-sectors should be


identified, to examine migration-work career-fertility interrelations.

3) Migration-fertility interrelations should be examined in terms of current


or recent fertility measures, in addition to cumulative fertility measures.

4) Longitudinal data sources are needed to permit adequately precise


analysis of the determination of migration decisions and their relation
to other life cycle events, whether migration is one-time and permanent,
temporary, return, repeat or sequential.

TABLE 5
CEB of Migrants from the Traditional-Urban to the Modern-Urban Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at
Origin and Destination by Education and Age

MID-
YOUNG AGE OLDER

LOW EDUCATION

Stayersat Origin 1.810 4.067 6.522


VeryRecentMigrants 1.349 2.661 5.286
New Migrants 1.333 2.910 5.949
VeteranMigrants 1.597 3.370 4.735
Stayersat Destination 1.343 2.756 3.849
SOME EDUCATION

Stayersat Origin 1.155 2.776 4.577


VeryRecentMigrants .678 2.111 3.609
New Migrants 1.139 2.315 4.289
VeteranMigrants 1.385 2.616 4.004
Stayersat Destination 1.036 2.043 2.840

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 309

TABLE 5 (Continued)
CEB of Migrants from the Traditional-Urban to the Modern-Urban Sector
by Regency of Migration, Education, and Age, and CEB of Stayers at
Origin and Destination by Education and Age

MID-
YOUNG AGE older

HIGH EDUCATION
Stayersat Origin .851 1.751 2.899
VeryRecentMigrants .737 1.389 2.333
New Migrants .814 1.441 2.970
VeteranMigrants .918 1.887 2.910
Stayersat Destination .545 1.439 2.239

TABLE 6
CEB of Migrants from the Traditional-Rural to the Modern-Urban Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin and
Destination by Education and Age

MID-
YOUNG AGE OLDER

LOW EDUCATION
Stayersat Origin 1.742 4.510 7.234

VeryRecentMigrants 1.105 4.083 6.800


New Migrants 1.455 3.221 6.427
VeteranMigrants 1.821 3.450 4.942

Stayersat Destination 1.343 2.756 3.849


SOME EDUCATION
Stayersat Origin 1.436 3.254 5.091

VeryRecentMigrants .600 1.789 3.609


New Migrants 1.280 2.000 4.111
VeteransMigrants 1.194 2.650 3.747

Stayersat Destination 1.036 2.043 2.840


HIGH EDUCATION3

Note: a Insignificant
category.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
310 International Migration Review

TABLE 7
CEB of Migrants from the Frontier-Urban to the Modern-Urban Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin
and Destination by Education and Age

MID-
YOUNG AGE OLDER

LOW EDUCATION
Stayersat Origin 4.241 6.024
VeryRecentMigrants a a
New Migrants 6.077 a
VeteranMigrants 4.074 5.540
Stayersat Destination 2.756 3.849
SOME EDUCATION
Stayersat Origin 3.067 4.532
VeryRecentMigrants a a
New Migrants 3.688 a
VeteranMigrants a 3.013 4.196
Stayersat Destination a 2.043 2.840
HIGH EDUCATION
Stayersat Origin .918 1.828 2.974
VeryRecentMigrants a 1.800 a
New Migrants .684 1.857 3.000
VeteranMigrants .875 1.783 2.852
Stayersat Destination .545 1.439 2.239

Note: a Insignificant
category.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 311

TABLE 8
CEB of Migrants from the Modern-Urban to the Modern-Rural Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin
and Destination by Education and Age

MID-
YOUNG AGE older

LOW EDUCATION

Stayersat Origin 1.343 2.756 3.849


VeryRecentMigrants 1.650 3.167 4.300
New Migrants 1.947 3.542 3.800
VeteranMigrants 1.536 3.830 4.489
Stayersat Destination 1.551 3.605 5.631

HIGH EDUCATION a

SOME EDUCATION a

Note: a Insignificant
category.

TABLE 9
CEB of Migrants fromthe Traditional-Urban to the Traditional-Rural Sectors
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin
and Destination by Education and Age

MID-
YOUNG AGE OLDER

LOW EDUCATION

Stayersat Origin 4.067 6.522


VeryRecent Migrants 2.174 a
New Migrants a a
VeteranMigrants 4.825 6.917
Stayersat Destination 4.510 7.234
SOME EDUCATION a

HIGH EDUCATION Q

Note: a Insignificant
category.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
312 International Migration Review

TABLE 10
CEB of Migrants from the Traditional-Rural to the Modern-Rural Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin and
Destination by Education and Age

MID-
YOUNG AGE OLDER
LOW EDUCATION
Stayersat Origin 1.742 4.510 7.234

VeryRecentMigrants 1.564 4.783 6.865


New Migrants 2.000 4.449 7.313
VeteranMigrants 1.910 4.448 6.933

Stayersat Destination 1.551 3.605 5.631


SOME EDUCATION a
HIGH EDUCATION a

Note: a Insignificant
category

TABLE 11
CEB of Migrants from the Traditional-Rural to the Modern-Rural Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin and
Destination by Education and Age
MID-
YOUNG AGE OLDER

LOW EDUCATION

Stayersat Origin 1.742 4.510 7.234


VeryRecentMigrants 1.667 4.134 7.635
New Migrants 1.582 4.307 7.329
VeteranMigrants 2.020 4.214 6.429
Stayersat Destination 1.810 4.067 6.522
SOME EDUCATION
Stayersat Origin 1.436 3.254 5.091
VeryRecentMigrants .625 a 4.571
New Migrants .826 2.933 a
VeteranMigrants 1.121 2.980 4.527
Stayersat Destination 1.155 2.776 4.577
HIGH EDUCATION a

Note:a Insignificant
category.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 313

TABLE 12
CEB of Migrants from the Traditional-Rural to the Modern-Rural Sector
by Recency of Migration, Education, and Age, and CEB of Stayers at Origin and
Destination by Education and Age

MID-
YOUNG AGE OLDER

LOW EDUCATION
Stayersat Origin 4.510 7.234

VeryRecentMigrants 4.478 6.667


New Migrants 4.175 7.031
VeteranMigrants 4.025 5.538

Stayersat Destination 4.241 6.024


SOME EDUCATION a
HIGH EDUCATION a

Note: a Insignificant
category.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
TABLE 13
Values of the H-Statistic by Region of Origin and Des
Recency of Migration, Age, and Education

AGE AND DESTINATION


YNG MID OLD YNG MID OLD YNG MID OLD
ORIG. EDUC. RECENCY MU MR TU

MU Low VERY REC. 1.48 .48 1.25


NEW 2.90 .93 .03
VET .93 1.27 .36

MR Low VERY REC. 1.37 .29 .49


NEW 1.49 .95 .33
VET 0.00 .53 .67
Some VERY REC. b 2.32 a
NEW b 1.36 .76
VET b .57 .85

TU LOW VERY REC. .99 1.07 .46


NEW 1.02 .88 .21
VET .46 .53 .67
Some VERY REC. 4.01 .91 .56
NEW .13 .63 .17
VET -1.93 .22 .33
High VERY REC. .37 1.16 .86
NEW .12 .99 -.11
VET .22 -.44 -.02

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
TABLE 13 (continued)
Values of the H-Statistic by Region of Origin and D
Recency of Migration, Age, and Education

AGE AND DESTINATIO


YNG MID OLD YNG MID OLD YNG MID OLD
PRIG. EDUC. RECENCY MU M* TU

TR Low VERY REC. 1.60 .24 .13 .93 .30 .23 b - .56
NEW .72 .73 .24 -1.35 .07 - .05 b .46 - .13
VET - .20 .60 .68 .07 .19 b .67 1.13
Some VERY REC. 2.09 1.21 .66 2.89 a 1.01
NEW .39 1.04 .44 2.17 .67 a
VET .61 .50 .60 1.12 .57 1.10

FU Low VERY REC. a a a


NEW a -1.24 a
VET a .11 .22
Some VERY REC. a a
NEW a - .61
VET a .05 .20
a .07 a
High VERY REC.
NEW .63 - .07 .04
VET .12 .12 .17

Note: a Insignificant
category
b StatisticUndefined

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
The Effect of Migration on Fertility 316

REFERENCES

Bach, R.L.
1981 "Migrationand Fertilityin Malaysia: A Tale ofTwo Hypotheses".International
Migration
Review,15(3):502-521.
Brazil
1980 "Sistemade InformacaoparaAnalisede Dados: Pesquisa Nacionalde AmostraDomiciliar
1976", InstitutoBrasileiro de Geografiae Estatistica,Superintendenciade Estudos
Geograficose Socio-Economicos,Rio de Janeiro.

1978 "PesquisaNacionalporAmostrade Domicilios? 1976",InstitutoBrasileirode Geografia


Primarias,Rio de Janeiro.
Superintendenciade Estatisticas
e Estatistica,
Duncan, O.D.
1965 "FarmBackgroundand Differential
Fertility", 2:240-249.
Demography,
Edmondston,B.
1976 "Migrant-Nonmigrant Fertility in MetropolitanAreasofLatinAmerica".In
Differentials
TheDynamicsofMigration: InternalMigration Occasional MonographSeries,
and Fertility.
Communications
Vol. 1, No. 5, Interdisciplinary Program,SmithsonianInstitution.
Goldberg,D.
1960 "AnotherLook at the Indiana Policy FertilityData", MilbankMemorialFund Quarterly,
38:23-36.

Studies,12:214-222.
1959 "The FertilityofTwo GenerationUrbanites",Population
Goldstein,S.
in Thailand, 1960-1970",
1978 "Migrationand Fertility 5:167-180.
CanadianStudiesinPopulation,

10:225-241.
BetweenMigrationand Fertilityin Thailand. Demography,
1973 "Interrelations
Goldstein,S. and A. Goldstein
1982 "The ImpactofMigrationon LengthofBirthIntervals:An AnalysisUsing LifeHistory
Data forMalaysia". Presentedat theAnnual MeetingofThe PopulationAssociationof
America,San Diego, California.

1981 "The Impact of Migrationon Fertility:An 'Own Children' Analysis forThailand",


35:265-284.
Studies,
Population
Goldstein,S. and P. Tirasawat
1977 "The Fertility ofMigrantsto UrbanPlaces in Thailand", East-WestPopulationInstitute,
Paper No. 77.

Hendershot,G.E.
1976 "Social Class, Migration,and Fertilityin thePhilippines".In TheDynamicsofMigration:
InternalMigration OccasionalMonographSeries,Vol. 1, No. 5, Interdiscipli?
andFertility,
naryCommunications Program,SmithsonianInstitution.

Hiday, V.
in thePhilippines",International
1978 "Migration,Urbanization,and Fertility Review,
Migration
12(3):370-383.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions
317 International Migration Review

Lee, B.S. and S. Farber


1982 "FertilityAdaptationby Rural-UrbanMigrantsin Developing Countries:The Case of
Korea". Presentedat theAnnualMeetingofThe PopulationAssociationofAmerica,San
Diego, California.
Macisco,J.,L. Bouvierand MJ. Renzi
1969 "MigrationStatus,Education,and Fertility
in PuertoRico, 1960",MilbankMemorial
Fund
Quarterly,47:167-187.
Martine,G.
1975 "MigrantFertility Adjustmentand UrbanGrowthin LatinAmerica",International
Migra?
tionReview,9(2):179-191.
McGirr,N.J.and C. Hirschman
1979 "The Two GenerationUrbaniteHypothesisRevisited",Demography,
16:27-35.
Merrick,T.W. and E. Berquo
1983 "The Determinantsof Brazil's Recent Rapid Decline in Fertility".Washington,D.C:
NationalAcademyPress.
Park,J.Y.and I.H. Park
1976 "Migrationand Female Labor ForceImpacton Korean Fertility".In theDynamics ofMi?
Internal
gration: andFertility.
Migration OccasionalMonographSeries,Vol. 1, No. 5, Inter?
disciplinaryCommunicationsProgram,SmithsonianInstitution.
Ribe, H. and T.P. Schultz
1980 "Migrantand Native Fertilityin Colombia: Are MigrantsSelectedAccordingto their
ReproductivePreferences?".
Yale UniversityEconomicGrowthCenterDiscussionPaper
No. 355.

This content downloaded from 185.44.77.34 on Tue, 17 Jun 2014 08:45:43 AM


All use subject to JSTOR Terms and Conditions

You might also like