Implications of economic reform and spatial mobility for fertility in vietnam

Population Research and Policy Review 20: 207–228, 2001 © 2001 Kluwer Academic Publishers Printed in the Netherlands 207 Implications of economic reform and spatial mobility for fertility in Vietnam MICHAEL J WHITE1 , YANYI K DJAMBA2 & DANG NGUYEN ANH3 Brown University, USA; Southeastern Louisiana University, USA; Institute of Sociology, Vietnam Abstract Vietnam has registered a dramatic decline in fertility during the last decades While the causes of such a sustained decline are still not well documented, many observers believe that government policies adopted in the 1980s have contributed to lower fertility This article focuses on the implications of the Doi Moi program of market reforms on fertility, taking into account the influences of migration and population policy The analysis is based on a sequential logit model of birth histories of ever married women interviewed in Vietnam in 1997 The results show a substantial decline in fertility since the Doi Moi program was introduced The disruptive effects of migration are less pronounced, although migrants generally exhibit lower childbearing rates, and a somewhat different pattern of parity progression We argue that the economic reforms of 1986, and the two-child policy initiated two years later, have reinforced Vietnamese women’s desire for smaller families Keywords: Economic reform, Fertility, Migration, Vietnam Introduction The population of Vietnam, which was about 13 million at the beginning of the century, attained 27 million in 1955 (Barbieri et al 1995) and 76.7 million in 1997 (NCPFP 1998) This rapid population growth and the war conditions of the past generated substantial regional differences in population distribution After the Vietnam War, the government adopted a series of measures both to control fertility and to redistribute the actual population These measures overlapped with the Doi Moi program, which transformed the country from a centrally planned system to a market-driven economy Although these programs were not associated with substantial economic growth, fertility has dramatically declined from 5.6 children per woman in 1979 (Haughton 1997) to about 2.3 in 1997 (GSO 1999) To understand how this spectacular fertility decline occurred in such a poor country, it is necessary to examine the goals of the following three programs: (1) the two-child policy, (2) the population redistribution program, and (3) the Doi Moi economic reform plan After several years of encouraging 208 M J WHITE, Y K DJAMBA & D N ANH family planning, the Vietnamese government officially adopted a comprehensive population policy in 1988 (Goodkind 1995) This policy stipulated that couples should have a maximum of two children, with three to five years spacing between births (Hoa et al 1996) To reinforce this policy, penalties would be imposed upon violators, including modest fines and possible loss of jobs for government employees (Goodkind 1995) In practice, however, the two-child policy was not implemented in the same way across the country In fact, many researchers have noted a discrepancy between proximate determinants of reproductive behavior and actual fertility (Allman et al 1991: 312; GSO 1995; Goodkind 1994) Haughton (1997), for example, argues that the proximate determinants of low fertility, including contraception and abortion, would imply a value close to two children per woman in early 1990s The gap, he notes, is probably due to the fact that people overstate the degree to which they use contraception (Haughton 1997: 208) In addition to possible measurement errors, this demographic puzzle suggests that the causes of sustained decline in fertility observed in Vietnam might also be associated with other factors, such as economic reform and migration, which can affect the proximate determinants of fertility Besides the two-child policy, the government also initiated a population redistribution program in 1980s This was the first major population policy initiated since the North–South unification in mid-1970s This program was designed to counter the great disparities between manpower and natural resources, focusing on rural-rural and urban-rural migration rather than encouraging rural-urban migration (Dang et al 1997: 319) Such a relocation program has changed the composition of both urban and rural populations, but its effect on fertility remains unknown The third important program that may have influenced the reproductive behavior of the Vietnamese population is the Doi Moi policy Officially adopted in December 1986, this program opened the doors to the free market economy and the non-socialist business world; new forms of ownership were introduced with the creation of private sectors and markets Yet, like the population distribution program, the effects of economic change on fertility in Vietnam remain unclear Moreover, the fact that all these three programs overlap in time and space makes the analysis of their individual effects on fertility difficult The purpose of this research is to examine the interrelations between migration, economic transition, and reproductive behavior in Vietnam The major question is whether the economic reform and the population redistribution programs initiated in the 1980s have contributed to further decline in fertility Previous studies on the implications of the economic reform on fertility in Vietnam have produced contradictory results Some scholars argued ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 209 Figure Total fertility and gross domestic product that children would be more wanted under the decollectivization period, as parents needed more labor for their newly acquired family farms (Allman et al 1991: 308) Yet, the analysis of fertility trends in Vietnam reveals a continuous decline, making this latter mechanism less plausible Conversely, the introduction of private entrepreneurship is said to lead to lower fertility, as parents struggle to devote their resources to business for upward mobility (Bryant 1998) This is because, under the new economic regime, the economic returns from children may be negative or, at least, less substantial than in the planned economy Parents must now cope with rising costs of education and other expenses, if they want to guarantee the social mobility of their children This strategy of grooming children for good jobs started among the rich and urban residents (Bryant 1998: 262), but it expanded to other groups across the society This assumption is in line with data in Figure (Data are in Table A1.) The examination of the relationship between fertility trends and economic data, as measured by average annual growth rates of real Gross Domestic Product (GDP), suggests that the economic reform of 1986 may have contributed to lower fertility Although these data are based on gross estimations of fertility from various censuses and different surveys (Haughton 1997), they reveal a substantial decline since the introduction of Doi Moi Also according to these data, Vietnam has registered a significant improvement in the growth of the real GDP from 1989, suggesting that the Doi Moi program has produced positive results (Harvie & Hoa 1997: 175) The fluctuations observed in 1991-1992 were mostly due to temporary decline in industrial 210 M J WHITE, Y K DJAMBA & D N ANH production The relationship between economic reform and fertility will be examined in more detail in this study by comparing childbearing behavior before and after the initiation of the Doi Moi program The influence of migration on fertility has been documented in several studies under the disruption hypothesis model (Ram & George 1990; Beaujot 1991; Ford 1990; Goldstein & Goldstein 1981) The conclusion emerging from most of these studies is that the fertility of migrant women will be lower at the time of arrival, higher shortly after arrival and will then converge to that of the resident women (Ng & Nault 1997) We examine the way in which migration alters the childbearing pattern of women The analysis is based on data from the Vietnam Migration and Health Survey conducted in 1997 Recent forms of migration and their contexts Over the past three decades the government has organized the relocation of population to new areas, called New Economic Zones, to achieve a more rational distribution of labor and population among various regions During the period between 1975 and 1995, a total of 4.7 million people were officially relocated (Do 1998) Most of these migrants moved from Red River Delta and Northern Central regions, and went to the Central Highlands and Southeast areas where more arable land and development programs were available The goal of this resettlement program was to improve the standard of living of the population, but it was also used to move political opponents to remote areas In some New Economic Zones, as many as half of the migrants have been reported to have moved again or returned to their place of origin soon after they arrived (Desbarats 1987; Dang 1997) Parallel to this planned program of resettlement, a spontaneous migration movement also took place in the 1990s These migrants were not under government control; they relocated on their own Spontaneous migration is often selective of more economically motivated individuals The Central Highlands and Southeast regions, known for their rich natural environments and large amounts of arable land, attracted many of these spontaneous migrants The Ministry of Agriculture and Rural Development claimed that spontaneous or unregistered migration has risen in recent years to include 212,000 households comprising more than one million people Of these figures, 97,000 households migrated to Southeastern provinces, more than 70,000 households moved to Central Highland provinces, and 37,000 households migrated to southern provinces (Do 1998) Another major redistribution in Vietnam is from rural to urban areas This movement is usually explained by economic factors Urban bias and general perceptions may also suggest that life would be better in cities than in rural ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 211 areas Sayings such as “It is better to be poor in the city than rich in the countryside” (Giau nha que khong bang ngoi le pho), illustrate the perceived advantage of urban residents over their rural counterparts in Vietnam today (Dang 1998) Moreover, income inequality between rural and urban areas and the freedom of movement since the initiation of the Doi Moi policy increased rural-urban migration Such movements were mostly temporary, with migrants working in the tertiary sector and in low-paid jobs often rejected by native workers (e.g., bricklayer, porters, weavers, vendors, cyclo drivers, etc.) Out-migration from rural areas seems beneficial to the migrants, their families, and their rural communities in several ways It relieves population pressure on limited land areas Many migrants send remittances to their families, augmenting rural well-being In addition, temporary migrant labor from rural provinces has become an important resource required for meeting the increasing demand for labor in both manufacturing and non-manufacturing industries We now investigate how these different forms of spatial mobility – urban and rural, temporary and permanent - affect fertility Data and methods The Vietnam Migration and Health Survey (VNMHS97) was conducted in 1997 by the Institute of Sociology of the National Center for Social Sciences and Humanities of Vietnam, under an international cooperation project on migration research and training funded by the United Nations Population Fund (UNFPA) The Vietnam survey is part of a four-country study designed and directed by the Population Studies and Training Center at Brown University (USA) The main purpose of the survey was to examine the interrelations between the Doi Moi program, the migration process initiated through the population redistribution policy of the 1980s, women’s status, and reproductive behavior Six provinces were covered by the survey, taking into account the differentials in geographical and rural/urban sectors in the country A purposive sampling method was applied to randomly selected areas in order to obtain an adequate number of respondents in each of the following three migrant categories: non-migrants, permanent migrants, and temporary migrants As a result, our findings cannot provide estimates of the relative numbers of each kind of migrant, nor can we aggregate the three categories Our focus, therefore, is on the comparative differentials in characteristics and reproductive behavior among the groups A total of 2502 individuals were interviewed in 1847 households, 1265 women and 1237 men Of the 1265 women, 1015 were ever married and 250 212 M J WHITE, Y K DJAMBA & D N ANH never married Our analysis is limited to ever married women of different migration statuses Migration is difficult to define Its definition depends on the research questions and the analytical strategies involved For example, some scholars may emphasize the time dimension (duration) of the move, while others consider the crossing of geographic borders as a major element in defining migration The definitions used here derive from intensive discussions between Brown University researchers and their colleagues in Vietnam These definitions have also been used in other work published on the same Vietnamese data (Djamba et al 1999) The three migrant categories are defined as follows: (1) Non-migrants were persons who were born and registered in the place of interview and who were never away for more than months since age 13; (2) Permanent migrants had been (a) born elsewhere, but registered at place of interview, or (b) born in place of current residence and registered there, but away from that place for at least six months since age 13; (3) Temporary migrants were born other than in place of interview and held permanent registration elsewhere In Vietnam, a large percentage of permanent migrants in rural areas were in fact return migrants – persons who had been either mobilized into the army or displaced during the many years of war Three different instruments were used to collect information on individual respondents, household members, and community data Although most variables used here are available in the individual respondents’ file, a few others are found only in the household file To get a working file for our fertility analysis, we combined the two data sets by matching the respondent’s information using two key identification variables (province and code) available in both files Among the 1015 ever married women interviewed, some have missing data on age at first marriage (16 cases) and migration status (4 cases) So in terms of migration status, there are 1011 valid cases distributed as follow: 488 non-migrants, 361 permanent migrants, and 162 temporary migrants (see Table 1) But our working file includes all the 1015 ever married women Our analysis focuses on the impact of economic reform and spatial mobility on the birth history of women with different migration statuses, which we analyze using an event history approach The individual questionnaire contains information on each of the first five moves (and last move, for those with more than five moves), type of place of residence or destination of each move (rural, town, city), and the waiting time, or each year of exposure to childbearing From the migration history, we constructed a dummy variable – called move – which takes the value of for each year for which a person moved, and otherwise Note that the move variable differs in many ways from the migration status variable, which divides respondents into the three categories defined ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 213 earlier The move variable records every trip of one month or more made after age 13, regardless of registration status In contrast, migration status is cross-sectional information, reflecting the respondent’s current residence and registration status According to these definitions, even non-migrants can have valid data on the move variable, if they made some short duration moves So, move, place of residence in the interval, and birth are time-varying variables, because their values may vary over the event history We also included attributes that are constant over time, such as education, place of birth, and age at marriage Another major variable used in this study is policy period This variable serves as a proxy for the economic reform Dang (1999) has identified the following important features of the Doi Moi periods The first period (1987– 1991) was characterized by the diversification of market opportunities and the emergence of the private sector and new forms of ownership During the second period (1992–1997), policy measures were taken to strengthen the achievements of the economic reform and explicitly promote foreign investments The second period also observed a substantially increased movement of rural labor to the major cities and urban centers (Dang, 1999) However, because the two-child policy was introduced only two years after the Doi Moi program, the policy period variable also measures the effect of the population control program of 1988 Following the approach used in previous work in China (Goldstein et al 1997), and the usual practice in the analysis of life history data (Allison, 1984; Blossfeld et al 1989: 22–25), we constructed a person year data file described below Each woman contributed one record for each year from age 15 onward until age at survey, or age 44, whichever is smaller We analyze these data in the form of sequential logit (discrete time) models predicting the likelihood of having a first birth since marriage, having a second birth since first birth, and having a higher-order birth since second birth Covariates available in each record include age of the woman in each person-year, a censoring indicator (1 if yes, otherwise), age of the woman at marriage, parity of the woman, place of residence, mobility status during the year (the move variable), duration of exposure to childbearing (waiting time), and policy period The resulting event history data file constructed from the 1015 respondents has a total of 23,211 records Results Before discussing the results of multivariate analysis, it is important to note some differentials observed in the original sample of ever married women The descriptive statistics in Table show that although most respondents were 214 M J WHITE, Y K DJAMBA & D N ANH Table Characteristics of sample: ever married women, Vietnam 1997 Characteristic All women Nonmigrants Permanent migrants Temporary migrants Average Age at Survey 40.5 (16.6) 22.0 (3.9) 2.8 (1.8) 24.0 (4.3) 27.1 (4.6) 31.9 (5.7) 0.9 (1.1) 1.8 17.9 40.4 28.7 11.2 55.6 15.3 29.1 26.0 13.3 60.7 1011 39.8 (13.0) 21.2 (3.7) 2.9 (1.8) 23.5 (4.0) 26.6 (4.7) 31.5 (6.2) 0.2 (0.6) 2.0 20.5 48.8 21.6 7.1 38.9 14.1 46.9 30.3 15.0 54.7 488 43.0 (11.7) 23.0 (4.2) 2.8 (1.8) 24.8 (4.6) 28.0 (4.7) 32.2 (5.0) 1.8 (1.0) 2.0 13.7 30.7 35.7 17.8 64.5 18.0 17.5 21.9 11.7 66.4 361 37.0 (12.3) 22.1 (3.7) 2.3 (1.6) 23.9 (4.5) 26.7 (3.7) 32.2 (5.5) 1.5 (0.9) 0.6 19.1 37.6 33.8 8.9 85.8 13.0 1.2 22.2 11.7 66.0 162 Average age at marriage Average number of children ever born Age at first birth Age at second birth Age at higher order birth Average number of moves Percent no school Percent with 1–5 years of schooling Percent with 6–9 years of schooling Percent with 10–12 years of schooling Percent with 13 or more years of schooling Percent residing in a city in 1997 Percent residing in a town in 1997 Percent residing in rural area in 1997 Percent born in a city Percent born in a town Percent born in a rural area Number of women Standard deviations in parentheses Total number of women shown above (1011) excludes respondents with unknown migration status The actual sample contains 1015 ever married women born in rural areas, many have moved to urban areas Though women may be moving primarily for family related reasons, the concentration of migrants in cities suggests that many migrant women and their families are attracted by various modern opportunities available in urban places This urban pull is partly sustained by urban bias, a practice in which governments and private investors tend to favor cities by providing them with better socioeconomic and technological services (Keyfitz 1982; Lipton 1976) ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 215 There are no significant differences in number of children ever born by migration status Likewise, migration status does not seem to affect the process of union formation; the differences in age at marriage by migration status are overall small Apart from number of moves, most variables listed in Table not change much across migration categories This weak association may reflect the cross-sectional, summary nature of these data The dynamic information contained in event history models may reveal other differences It is noteworthy, though, that temporary migrants are younger and intermediate in educational attainment We have evidently captured some persons in the middle of an adjustment process Once enough time passes, many people who are now temporary migrants will register and then be classified as permanent migrants Table presents selected statistics by age and migration groups Comparing within age groups and across migration status, there is some suggestion that migrants are a little later to marry and carry through their first and second birth (We not make comparisons across age due to the variations in the family-building processes) Overall, the effects of migration status are modest Table Median age at marriage and childbearing; ever married Vietnamese women, 1997 Median age Marriage 1st birth 2nd birth 3rd birth Non-migrant Permanent migrant Temporary migrant 21.0 23.0 23.0 23.0 24.0 24.0 26.0 28.0 27.0 28.0 31.0 30.0 All women 22.0 24.0 27.0 29.0 How these traits manifest themselves in the birth history? We now turn to event history analysis predicting the likelihood of having a birth of a given order, as a function of policy period, spatial mobility, and other selected variables Since economic reform and population policy were introduced during the same period, distinguishing their effects on fertility is a daunting task that goes beyond the scope of this paper Because most Vietnamese want to have two children (Goodkind 1995: 98–99), we suggest that the population policy effects will be seen mostly in the higher-order births Economic policy effects, if any, would be more likely to emerge through covariates linked to population redistribution 216 M J WHITE, Y K DJAMBA & D N ANH First marriage to first birth Table contains the results of logit regression models predicting the likelihood of having a first birth since marriage, for each migration status as defined at the time of the survey The data in Table show that the effects of the explanatory variables on the transition to childbearing differ across the three migrant status categories So, while no significant policy period effects were found in the logit regression model predicting the transition to childbearing for permanent migrants, the likelihood of having a first birth increased significantly for temporary migrants during the five years following the introduction of the economic reform program and decreased thereafter The effect of policy period is more pronounced for non-migrants for whom the corresponding regression coefficients are positively and statistically significant for all the post-Doi Moi periods The effect of the disruption hypothesis on first birth is captured by the ‘move in interval’ dummy variable For all three groups of women, the direction of the effect is negative but not significant Such findings not support the idea that spatial mobility reduces childbearing in Vietnam.1 That is, migration does not account for most of fertility decline observed in Vietnam.2 Age is understandably an important predictor of first birth The likelihood of having a first birth since marriage increases with age, but the relationship is not linear The turning point depends on migration status and other factors From the Table coefficients, we estimate that the probability of having a first birth since marriage would start to decline from around age of 29, 32, and 31 years, respectively for non-migrants, permanent migrants, and temporary migrants These calculations show the appreciable age displacement in the childbearing profile of migrants Age at marriage also affects the timing of childbearing The relationship is negative and significant for non-migrants, permanent migrants, and temporary migrants The negative association suggests that those who married at younger ages had their first child sooner after marriage than their counterparts who married at older ages Other variables are also associated with first birth Educational attainment has significant effects on childbearing only for non-migrants, whereas urban residence significantly increases the likelihood of first birth for non-migrants, permanent migrants, and temporary migrants Nativity does influence childbearing, but in an inconsistent manner Among non-migrants only those born in a city had a significant reduction in the risk of first birth, while for temporary migrants only town born respondents had significantly lower risk When all the above factors are controlled, differences in the timing of first birth become visible The regression coefficients for waiting time show a more pronounced duration pattern for permanent migrants Among temporary migrants, the likelihood of first birth is highest during the second year after ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 217 Table Likelihood of having a first birth since marriage, controlling for policy period, migration, and selected background variables: Vietnamese women, 1997 Variable Age at beginning of interval Age square Age at first marriage Policy period 1987–1991 1992–1997 Education 6–9 years 10–12 years 13 or more years Place of birth City Town Place of residence in interval City Town Moved in the interval Waiting time Year one Year two Year three Year four Constant Log likelihood Number of observations Non-migrants Coeff Robust SE Permanent migrants Coeff Robust SE Temporary migrants Coeff Robust SE 2.009∗∗∗ 0.324 −0.035∗∗∗ 0.006 −0.174∗∗∗ 0.045 1.718∗∗∗ 0.258 −0.027∗∗∗ 0.005 −0.276∗∗∗ 0.06 1.676∗∗∗ 0.356 −0.027∗∗∗ 0.007 −0.107 0.045 0.301∗ 0.63∗∗∗ 0.502∗ 0.215 0.143 0.182 0.203 −0.061 0.176 0.221 0.218 0.332 0.974∗∗∗ 0.221 0.955∗∗∗ 0.235 0.229 0.378 0.411 0.296 0.074 0.352 0.336 0.348 0.189 0.031 −0.191 0.349 0.381 0.412 −1.789∗ −1.401 0.739 0.989 −0.033 0.349 0.185 0.191 −0.432 −1.142∗ 0.285 0.521 1.939∗ 1.336 −1.616 0.713 0.993 0.826 0.561∗∗∗ 0.193 0.296 0.301 −0.222 0.219 0.682∗ 0.803∗ −0.531 0.283 0.357 0.368 0.772 0.643 0.956∗ 0.944∗∗∗ −26.04∗∗∗ 1036 4165 0.873 0.683 0.433 0.29 3.699 – 2.74∗∗∗ 1.055 1.51∗∗ −23.845∗∗∗ 350 1504 – 0.745 0.779 0.516 4.724 0.513 1.114 1.359∗∗ 0.926∗∗ −21.02∗∗∗ 820 3705 0.954 0.724 0.432 0.367 2.824 The reference category for policy period is pre-1987, for education it is less than years, for place of birth and residence it is rural, and for waiting time it is five years or more Due to collinearity constraints, dummy variable “year one” is excluded from temporary migrants equation ∗∗∗ p < 0.001; ∗∗ p < 0.01; ∗ p < 0.05 marriage (year constrained to be same as years 5+) and decreases in the following year, before becoming again positively significant in the fourth year.3 First birth to second birth As in Table 3, the models of birth history in Table reveal that the transition from first birth to second birth is not significantly affected by spatial mobility 218 M J WHITE, Y K DJAMBA & D N ANH That is, those who moved during this transitional period of maternity are not statistically different from those who did not The policy period variable is now negatively associated with the reproductive behavior of all respondents, but its effect is only significant for non-migrants and permanent migrants It is noteworthy that this policy effect emerges so strongly amid widespread norms of having two or more children We note further that the period effects differ only modestly across migration status The positive association between age at marriage and second birth suggests that women who married at younger ages spaced their births more than their counterparts who married at older ages, though the relationship is only significant for non-migrants and permanent migrants Overall, these results point to appreciable adjustment of the timing of childbearing We found no evidence that education reduces the likelihood of second birth This weak association between education and fertility, which was also noted in previous work (Luc et al 1993), remains even when we excluded all other variables from the regression models The negative and significant effect of exposure (waiting) time on childbearing in Table indicates the ‘longer’ pattern of birth spacing characterizing the reproductive behavior of current Vietnamese women All other variables are not statistically significant Second birth to higher-order birth The transition to higher-order birth (beyond two births) is analyzed in Table Unlike the previous two tables, age is only statistically significant for non-migrants Although it is positively associated with higher parity, age at marriage is not an important predictor here Instead, place of birth and place of residence are important predictors of higher-order births for non-migrants For such women, being born in a town reduces their chance of higher parity, whereas living in a town increases it We found that policy period is significantly and negatively associated with the chance of higher-order births for permanent migrants and temporary migrants The magnitude of the policy period effect is particularly strong among temporary migrants In the recent period, the magnitude of this effect for temporary migrants is double that for permanent migrants and over triple that for non-migrants Nonetheless, for all migrant categories, the policy period effect is more pronounced in recent years These results suggest that the policy programs – economic reform and two-child policy – introduced in the 1980s reinforced the desire for smaller families Nonetheless, because the population policy was tailored at limiting fertility to two children per woman, it is possible that most of the period effects observed in Table are due to the two-child policy, rather than to economic reform While there is no evidence that migration during the interval contributed to fertility decline, it is ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 219 Table Likelihood of having a second birth after first birth, controlling for policy period, migration, and selected background variables: Vietnamese women, 1997 Variable Non-migrants Coeff Robust SE Permanent migrants Coeff Robust SE Temporary migrants Coeff Robust SE Age at beginning of interval Age square Age at first marriage Policy period 1987–1991 1992–1997 Education 6–9 years 10–12 years 13 or more years Place of birth City Town Place of residence in interval City Town Moved in the interval Waiting time Year one Year two Year three Year four Constant Log likelihood Number of observations −0.066 0.000 0.037∗ 0.128 0.002 0.012 0.284 −0.006∗ 0.043∗∗∗ 0.175 0.003 0.014 0.684∗ −0.013∗ 0.015 0.285 0.005 0.016 −0.499∗∗ −0.659∗∗∗ 0.190 0.165 −0.517∗ −0.347 0.211 0.188 −0.767 −0.557 0.403 0.352 −0.067 −0.526 −0.523 0.230 0.278 0.373 −0.148 −0.157 −0.084 0.320 0.313 0.335 0.306 −0.361 0.294 0.369 0.361 0.451 −0.160 −0.654 0.457 1.672 −0.310 −0.286 0.223 0.230 −0.469 −0.277 0.353 0.490 −0.144 0.054 0.792 0.463 1.677 0.675 −0.288 −0.384 0.042 0.215 0.239 0.420 0.140 −0.049 0.199 0.329 0.441 0.521 −2.368∗∗∗ −0.080 −0.236 −0.148 1.254 768 1646 0.317 0.232 0.221 0.246 1.963 −1.622∗∗∗ −0.212 0.128 −0.516 −4.356 621 1313 0.274 0.224 0.219 0.267 2.545 −2.003∗∗∗ −0.027 0.308 0.141 −9.620∗ 233 554 0.497 0.400 0.395 0.438 4.173 The reference category for policy period is pre-1987, for education it is less than years, for waiting time it is five years or more ∗∗∗ p < 0.001; ∗∗ p < 0.010; ∗ p < 0.05 the migrant groups who are most noticeably constraining their childbearing, net of other factors we control Consistent with most literature on the relationship between education and fertility (Bulatao & Lee 1983; Bledsoe et al 1999), we found that highly educated women are less likely to have higher-order births than less educated ones This pattern is observed among all women, but the relationship is only statistically significant for non-migrants and permanent migrants It is only at these higher-order intervals that we find a significant negative influence of education in our data 220 M J WHITE, Y K DJAMBA & D N ANH Table Likelihood of having a higher-order birth after second birth, controlling for policy period, migration, and selected background variables: Vietnamese women, 1997 Variable Non-migrants Coeff Robust SE Permanent migrants Coeff Robust SE Temporary migrants Coeff Robust SE Age at beginning of interval Age square Age at first marriage Policy period 1987–1991 1992–1997 Education 6–9 years 10–12 years 13 or more years Place of birth City Town Place of residence in interval City Town Moved in the interval Waiting time Year one Year two Year three Year four Constant Log likelihood Number of observations −0.640∗∗∗ 0.009∗∗∗ 0.016 0.178 0.003 0.011 −0.086 0.000 0.017 0.219 0.003 0.010 −0.254 0.004 0.008 0.377 0.005 0.015 −0.169 −0.371 0.212 0.201 −0.489∗ −0.799∗∗∗ 0.232 0.219 −1.117∗∗ −1.543∗∗∗ 0.444 0.435 0.042 −0.134 −1.000∗ 0.187 0.343 0.459 −0.282 −0.520 −0.886∗∗ 0.293 0.295 0.318 −0.576 −0.902 −1.325 0.482 0.529 0.757 0.211 −2.545∗∗∗ 0.476 0.411 −0.333 −0.308 0.269 0.308 0.387 −0.016 0.381 0.610 −0.743 2.811∗∗∗ 1.039 0.478 0.350 0.678 0.210 0.131 −0.869 0.260 0.292 0.609 0.810 0.682 −0.099 0.490 0.546 0.827 −2.253∗∗∗ −1.251∗∗∗ −0.794∗∗ −0.701∗∗ 9.553∗∗∗ 732 2507 0.418 0.282 0.257 0.253 2.958 −2.716∗∗∗ −0.566∗ −1.024∗∗ −0.103 0.715 560 2197 0.598 0.273 0.352 0.244 3.682 −2.611∗ −0.783 −0.021 0.542 2.426 171 675 1.257 0.643 0.579 0.450 6.534 The reference category for policy period is pre-1987, for education it is less than years, for place of birth and residence it is rural, and for waiting time it is five years and more ∗∗∗ p < 0.001; ∗∗ p < 0.01; ∗ p < 0.05 Differential effects of migration Using the results of the sequential logit models in Tables 3–5, we calculated simulated probabilities of childbearing during the first year of exposure separately for non-migrants born and living in rural areas, non-migrants born and living in cities, rural-urban permanent migrants, and rural-urban temporary migrants, controlling for policy period The four models are estimated for all the three transitional stages: transition from marriage to first birth, from first birth to second birth, and from second birth to higher-order birth The simulations were conducted on hypothetical women aged 24 years at time of ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 221 Figure Simulated percent likely to have a first birth, by policy period and migration status survey, married at age 22, with 6–9 years of schooling, and having made no move within the interval Because we defined four migration statuses (two for non-migrants), three policy periods, and three transitional fertility stages, there were 36 simulations in total (4 × × 3) These simulations are developed from coefficient estimates; of course, one must consult the original tables to check on statistical significance The results are summarized in Figures 2–4 (see tabular data in Appendix A2) According to these data, the likelihood of giving birth during the first year of exposure depends on policy period and migration status For example, data in Figure suggest that the likelihood of transition to motherhood has increased continuously for both rural and urban nonmigrants since the introduction of the Doi Moi program, whereas migrants registered some decline More specifically, the information in Figure shows that more nonmigrants are now having their first birth sooner after marriage than ever before The picture is more complex for migrants Both permanent and temporary migrants registered a temporal increase during the first five years of the Doi Moi program, but their likelihood of childbearing declined in the 1992– 1997 years Temporary migrants had the lowest probability of transition to motherhood during all the periods considered here Overall, the results in Figure show childbearing patterns that were earlier for non-migrants and later for migrants As we discussed earlier, the determinants of transition to second birth are difficult to identify We observe a net temporal decline in fertility for all 222 M J WHITE, Y K DJAMBA & D N ANH Figure Simulated percent likely to have a second birth, by policy period and migration status four groups of women: the probability of progressing to second birth is in all cases lower in 1992–1997 than before 1987 (Figure 3) There are, however, enormous variations by migration status Non-migrants whose risk of first birth was higher in recent periods seem to delay their second birth now more than migrants For migrants, the lowest values are observed during the 1987-1991 period, supporting the view that the first five years of the Doi Moi program was a transitional period characterized by the diversification of the market opportunities (Dang 1999) It was a period of uncertainty during which female migrants who already had one child waited longer before deciding on having a second one The relative increase in the likelihood of having a second child during the 1992–1997 period probably reflects the adaptation of rural-urban migrants to the prosperous economy of that time (see Figure 1) The most dramatic pattern emerges in Figure It becomes clear in this figure that government policy periods have played a role in lowering fertility in Vietnam According to the present simulations, women born in rural areas had the highest probability of having a higher-order birth during the period before 1987 In contrast, non-migrants born in cities and living there had a relative lower risk (5%) of childbearing during the same period From 1987 to 1997, the risk of higher-order births declined for all women, but not with the same magnitude Rural non-migrants retained the highest probability of higher-order birth during the last 10 years (1987–1997), followed by urban non-migrants Both permanent and temporary migrants started at the same ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 223 Figure Simulated percent likely to have a higher-order birth, by policy period and migration status level in the pre-reform period, but went through different paths from 19871991, reaching the same probability level in the 1992–1997 period These results suggest that the reproductive behavior of Vietnamese mothers who have already had two children is also affected by their place of residence and their migration status At all points in time migrants bear children at lower rates than non-migrants, including urban natives For all of these groups childbearing rates declined through the policy periods To be sure, both economic reforms and direct population policy are implicated in these shifts Overall, there seems to be an acceleration in the rate of first birth with an attendant decline in the rate of subsequent childbearing By the 1992– 1997 period, migrants are slower than non-migrants to move to first birth, faster to have second birth, and unlikely to have a third child The downturn in migrants’ bearing a second child in the 1987–1991 period may reflect the slower economic growth of the middle 1980s Conclusion Fertility has declined appreciably in Vietnam during the last decades This sustained decline, in the face of apparently modest economic growth and levels of living, has caught the attention of researchers (Nguyen et al 1996; Hoa et al 1996) But the causes of the decline are not yet well documented We identified three policy areas of interest for explaining Vietnamese fertility: the two-child policy, economic reform, and the population redistribution 224 M J WHITE, Y K DJAMBA & D N ANH program The two-child policy and the economic reform program were introduced during the same period, making the statistical separation of their effects on fertility a difficult task We capture their joint effect through a policy period variable Our event history approach analyzed the impact of these changes on birth history separately for non-migrants, permanent migrants, and temporary migrants The results from our event history analysis suggest that these reforms introduced in 1980s – two-child policy and economic reform – have contributed to the decline of fertility Overall, most women have had fewer children since the initiation of economic reform and of the two-child policy More precisely, women are moving more rapidly from marriage to first birth, and they have substantially reduced the rate at which they progress to second and higherorder births As other scholars have noted (Haughton 1997), we would expect some further decline in total fertility in Vietnam in the next decade or so This analysis suggests that policy initiatives may have affected fertility through the free-market reforms, the two-child policy program, and population redistribution deregulation The introduction of private sector structures and new forms of ownership, combined with the progressive elimination of state supported services, may have created a more individualistic view of family and increased the burden of responsibilities facing Vietnamese couples This perception may have motivated couples to delay and even limit their fertility, in order to adapt and to improve their socio-economic conditions in the new free market economy Family planning may have played an additional role in reducing childbearing beyond the new norm of two The loosening of strictures on migration will, we suggest, further serve to move fertility toward replacement Acknowledgments The research reported here has been carried out as part of a larger project on Interrelations Among Migration, Economic Change, Women’s Status, Reproduction, and Health in Countries Undergoing Economic, Political, and Demographic Transitions The project covers four countries and is funded by the United Nations Population Fund (Vietnam, Ethiopia, and Guatemala) and the Andrew Mellon Foundation (South Africa) We are grateful to Alice Goldstein, Sidney Goldstein, Kelley Smith, and two anonymous reviewers for their helpful comments and suggestions, and we thank Hope Moffat for her help with final manuscript preparation ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 225 Notes It is possible that childbearing decreased the chance of spatial mobility for some women, but such assumption cannot be verified from the present analysis As suggested by one referee, we also ran pooled regression models for first, second, and third births, in which migration status was included as a dummy variable Still, there were no significant differential effects of migration status on fertility Nonetheless, because our samples were selected purposively, we must keep migrant categories separately throughout the analysis An identification problem emerged for the temporary migrants model, when all time period dummies were included The results presented in Table constrain the dummy for year one to be zero We also estimated other versions of the equation eliminating all time period dummies and constrained years and to be equal and these gave substantively identical results Appendix A1 Real gross domestic product and total fertility estimates for Vietnam Year GDP Total fertility Year GDP 1980 1981 1982 1983 1984 1985 1986 1987 1988 −3.7 5.1 8.2 7.1 8.4 5.6 3.4 4.0 5.2 4.8 4.7 4.6 4.5 4.4 4.3 4.2 3.9 3.8 1989 1990 1991 1992 1993 1994 1995 1996 1997 7.5 5.1 6.0 8.7 8.1 8.8 9.5 9.8 9.9 Total fertility 3.7 3.6 3.4 3.3 3.2 3.1 3.0 2.9 2.8 226 M J WHITE, Y K DJAMBA & D N ANH Appendix A2 Simulated percent likely to have a birth of a given order during the first year of exposure, by policy period and migration status Non-migrants – Rural Non-migrants – City Permanent migrants Temporary migrants First birth Pre-1987 49 53 52 29 1987–91 56 60 57 40 1992–97 64 68 50 33 Non-migrants – Rural Non-migrants – City Permanent migrants Temporary migrants Second birth Pre-1987 1987–91 11 1992–97 Non-migrants – Rural Non-migrants – City Permanent migrants Temporary migrants Higher-order birth Pre-1987 1987–91 3 1992–97 1 References Allison, P.D (1984), Event History Analysis: Regression for Longitudinal Event Data Beverly Hills/London/New Delhi: Sage Publications Allman, J., Nhan, V.Q., Thang, N.M., San, P.B & Man, V.D (1991), Fertility and family planning in Vietnam, Studies in Family Planning 22(5): 308–317 Barbieri, M., Allman, J., Pham, B.S & Nguyen, M.T (1995), La situation démographique du Viet Nam, Population 50(3): 621–651 Beaujot, R (1991), Population Change in Canada: The Challenges of Policy Adaptation Toronto: McClelland & Stewart Bledsoe, C.H., Casterline, J.B., Johnson-Kuhn, J.A., & Haaga, J.G (1999), Critical Perspectives on Schooling and Fertility in the Developing World Washington, DC: National Academy Press Blossfeld, H.P., Hamerle, A & Mayer, K.U (1989), Event History Analysis Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers Bryant, J (1998), Communism, poverty, and demographic change in North Vietnam, Population and Development Review 24(2): 235–269 Bulatao, R.A & Lee R.D., eds (1983), Determinants of Fertility in Developing Countries New York: Academy Press ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 227 Dang, N.A (1999), Rural-urban transition and the impacts of migration in Vietnam Paper presented at the 1999 annual meeting of the Population Association of America, New York City, 25–27 March 1999 Dang, N.A (1998), Patterns of migration and economic development in Vietnam, Journal of Socio-economic Development 64(2): 35–44 Dang, N.A (1997), Development, government policy and internal labor migration in Vietnam Doctoral dissertation, Department of Sociology, Brown University, Providence, Rhode Island, USA Dang, A., Goldstein, S & McNally, J (1997), Internal migration and development in Vietnam, International Migration Review 31(2): 312–337 Desbarats, J (1987), Population redistribution in Vietnam, Population and Development Review 13(1): 43–76 Djamba, Y., Goldstein, A & Goldstein, S (1999), Permanent and temporary migration in Viet Nam during a period of economic change, Asia-Pacific Population Journal 14(3): 25–48 Do, V.H (1998), Resettlement in Vietnam: its effects on population and production, in Proceedings of the International Seminar on International Migration: Implications for Migration Policy in Vietnam Hanoi: The Population Council Ford, K (1990), Duration of residence in the United States and fertility of US immigrants, International Migration Review 24(1): 34–68 General Statistical Office (GSO) (1999), Viet Nam Demographic and Health Survey 1997 Hanoi, Viet Nam: Statistical Publishing House General Statistical Office (GSO) (1995), Viet Nam Intercensal Demographic Survey 1994: Major Findings Hanoi, Viet Nam: Statistical Publishing House Goldstein, A., White, M & Goldstein, S (1997), Migration, fertility, and state policy in Hubei Province, China, Demography 34(4): 481–491 Goldstein, S & Goldstein, A (1981), The impact of migration on fertility: an ‘own children’ analysis of fertility for Thailand, Population Studies 35(2): 265–284 Goodkind, D.M (1995), Vietnam’s one-or-two-child policy in action, Population and Development Review 21(1): 85–111 Goodkind, D (1994), Abortion in Vietnam: measurements, puzzles, and concerns, Studies in Family Planning 25(6): 1–11 Harvie, C & Hoa, T.V (1997), Vietnam’s Reforms and Economic Growth New York: St Martin’s Press Haughton, J (1997), Falling fertility in Vietnam, Population Studies 51(2): 203–211 Hoa, H.T., Toan, N.V., Johansson, A., Hoa, V.T., Höjer, B & Persson, L.Å (1996), Child spacing and two child policy in practice in rural Vietnam: cross sectional survey, British Medical Journal 313 (7065):1113–1116 Keyfitz, N (1982), Development and elimination of poverty, Economic Development and Cultural Change 30: 649–670 Lipton, M (1976), Why Poor People Stay Poor: Urban Bias in World Development Cambridge, MA: Harvard University Press Luc, N., Thang, N.M., Swenson, I & San, P.B (1993), Selected determinants of fertility in Vietnam: Age at marriage, marriage to first birth interval and age at first birth, Journal of Biosocial Science 25(3): 303–310 National Committee for Population and Family Planning (NCPFP) (1998), Population and Reproductive Health Profile of Vietnam Socialist Republic of Vietnam Ng, E & Nault, F (1997), Fertility among recent immigrant women to Canada, 1991: An examination of the disruption hypothesis, International Migration 35(4): 559–580 228 M J WHITE, Y K DJAMBA & D N ANH Nguyen, V.P., Knodel, J., Mai, V.C & Hoang, X (1996) Fertility and family planning in Vietnam: Evidence from the 1994 inter-censal demographic survey, Studies in Family Planning 27(1): 1–17 Ram, B & George, M.V (1990), Immigrant fertility patterns in Canada, 1961–1986, International Migration 28(4): 413–426 Address for correspondence: Michael J White, Population Studies and Training Center, Brown University, Box 1916, Providence, RI 02912, USA Phone (401)-863-1083; Fax (401)-863-3351; E-mail: Michael_White@brown.edu  ... studies on the implications of the economic reform on fertility in Vietnam have produced contradictory results Some scholars argued ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 209... of first birth is highest during the second year after ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 217 Table Likelihood of having a first birth since marriage, controlling for. .. fertility decline, it is ECONOMIC REFORM AND SPATIAL MOBILITY FOR FERTILITY IN VIETNAM 219 Table Likelihood of having a second birth after first birth, controlling for policy period, migration, and selected