While Bolivian was predominantly rural in 1976 (38% urban), it became predominantly urban by 2001 (62% urban). Urbanization accelerated in the 1980s and changed the demographic profile of the country. Fertility rates declined, infant and child mortality rates declined and life expectancy increased. Unlike other Latin countries, however, Bolivian urbanization proceeded in a number of cities, rather than in a single hub (Urquiola et al 1999). The cities of La Paz/El Alto, Cochabamba and Santa Cruz today make up two thirds of the total population. Most rural-urban migrants left the rural highlands for better life opportunities in these three cities. Population growth has accelerated in the seven largest cities since 1950. The cities of Santa Cruz and El Alto showed the highest population growth in recent years. A look at recent household data show that most internal migration involves families and individuals moving from rural areas to capital cities (52%), followed by individuals moving from small towns to capital cities (27%).
The story behind rural-to-urban migration accelerates rapidly after two exogenous shocks in the 1980s. The first shock is a nationwide drought in 1982-1983 that affected both the highland
altiplano regions of Oruro and Potosi, as well as the valley regions of Cochabamba and Chuquisaca. While neither census nor household surveys captured the population shifts that followed, there is considerable qualitative literature on the rapid influx of displaced rural-migrants (Sandoval, Albo and Greaves, 1981, 1982, 1983 and 1987), and the growth of the outskirts of the cities of La Paz and Santa Cruz. Sandoval et al, in particular, suggest that rural aymara and quechua migrants were “riding between two worlds”, during this period: rural community life and urban squatter neighborhoods. The “ruralization” of urban life is an important theme of the early 1980s drought wave.
The second shock is an economic growth collapse in 1985-1986, caused by the decline of tin prices and a generalized contraction of the economy, following hyperinflation and adjustment in the late-1980s (see Hausmann, Gray Molina and Rodriguez, forthcoming). The growth collapse affected the heavily populated mining of northern Potosi and southern Oruro The literature of the period describes a relatively rapid process of forced resettlement of as many as 30,000 miner families, aided by compensation policies that provided a lump-sum grant to retired miners of COMIBOL, the state mining enterprise.
We tackle the determinants of migration question in two steps. In the first step, we estimate
probabilistic models for rural-urban migration for the whole sample, including regressors for observed characteristics (age, experience, education, wealth, civil status) In the second step, we
address the self-selection problem, by including the residuals of the selection equation estimate
thus capturing the effects of non-observable characteristics (skill, luck or talent). With this second step, which only includes migrants, we provide a non-biased estimate of the impact of migration over earnings in the place of destination. In the following section we’ll return to this issue focusing more attention on the place premium for migrants.
First, being married reduced the probability of migrating. This makes sense given what we
know about the migration process: typically, heads of household migrate first, make a foothold
and eventually are followed by spouse and children. In some cases, this eventually includes
extended family and friends (see Albo and Sandoval 1983). Second, higher levels of education
predict a higher probability of migration. This also fits in well with what is known from past studies (Andersen 2002). Education levels allow a transition from rural to urban labor markets, from lowpaying jobs to higher paying jobs. Third, the higher the family’s level of wealth, the higher the probability of migrating. In the Bolivian case, the poorest do not migrate. This is indicative of high internal migration costs. Only the better-off can take on the risk and associated costs of migration (see Tannuri-Pianto et al 2005). One important omission in the probit model is the non-statistical significance of indicators that measure the provision of local social services. This would tend to reassert the “pull” factor of urban settings rather than “push” factors from rural communities.
First, earnings increase with schooling. However, the joint effect of schooling and being a
migrant is negative. We hypothesize that this might reflect poor schooling quality in rural or other urban towns, which do not result in higher earnings in capital-city labor markets. The effect of schooling quality is something we return to when estimating place premiums. Second, earnings decrease for women and decrease even further for indigenous women. Both the gender and ethnic biases are reported in other studies on urban earnings in Bolivia. We will also look at this in more detail with the quantile regressions that disaggregate gender and ethnic biases by income level. Finally, being older increases earnings up to a point, and then moves in the other direction. This is also widely documented in the literature on migration in Bolivia. Younger migrants tend to have a better chance of moving up the earnings ladder over time.
The place premium approach allows us to estimate wage differences for otherwise identical
workers in rural and urban areas, and see how much of the difference is based on observable
(education, gender, ethnicity) or unobservable differences (skill, talent, ability). The “place
premium” is the wage difference attributable to geographic place of residence alone, after
controlling for the observable and unobservable effects. It reflects a powerful incentive for
migration, both at the individual and at the household level.
Besides incentives to migrate, high place premiums pose an additional puzzle. If returns to
education are high and increasing, why do we not see many more Bolivian children moving up to
the top of the education ladder? The literature on education returns has focused on two types of
explanation (see Perry et al 2006; Bourguignon et al 2005). First, given that returns to education are lumpy, and diplomas often matter a great deal, education seems attractive only when the long-term investments needed to complete at least a full course of secondary and some tertiary education can be realized. Second, in most countries the high average returns to schooling are not available to everyone; in particular, poor families tend to accrue returns to their investments in higher levels of education that are significantly below the average market return.
This paper has focused on internal migration and its effects over human development. In recent
years, however, academic attention has shifted to external migration, both for its implications at
home and abroad (see Clemens et al 2008; Fajnzylber and Lopez 2008, World Bank 2009).
Unfortunately, there are is no single source of data that tracks both internal and external
migration data for Bolivia. While internal migration data are drawn from census and household
surveys in Bolivia, external migration data are drawn from census data from the US and OECD
countries.
Three characteristics stand out from the comparative data on external migration for Bolivia. The first has to do with the Bolivian migrant profile. Niimi and Ozden (2008) show comparative
information for Bolivia with respect to other Latin American countries, using US historical data on migration. Although the number of Bolivian migrants is not high (about 80,000 captured by the 2000 US Census), the typical Bolivian migrant tends to be have a higher than average degree of education (with respect to non-migrant Bolivians) and tends to be employed in higher skilled jobs (with respect to home). The age and gender composition of Bolivian migrants is mostly average for Latin American migrants to the US.
The second issue is the link between external migration and remittances. Acosta et al (2008)
analyze the effects of migrant remittances over poverty. Levels of remittances have increased
significantly over the past five years, from $ us 159 million in 2003 to an estimated $ us 927 million in 2008 (World Bank 2008). In general terms, the size of remittances is moderate in Latin American terms (between 5% and 10% of GDP), but in absolute terms, remittances are second only to gas exports by volume of foreign currency receipts. Acosta et al estimate, using 2002 household and remittance data, that Bolivian remittances account for a relatively small share of poverty reduction (0.4 percent reduction) and inequality reduction (0.002 of the gini coefficient reduction).
The relatively small impact might be affected by two measurement issues that have affected
comparable studies in Bolivia. The first is sampling. The MECOVI surveys are designed to produce robust estimates of urban and rural income, but tend to underestimate the number of household receiving remittances from abroad. The second problem is with the question that aims to capture remittances, because the wording of the remittances question is not immediately assimilated with transfers from family members by check, wire or cash. Besides measurement problems, the MECOVI surveys suggest remittances are received by middle and upper thirds of the income distribution, rather than the poorest third. This is consistent with the migrant profile which tends to show the poorest individuals in Bolivia do not migrate internally or externally to the extent of the non-poor.
The third issue is the analysis of place premiums for external migration. Clemens et al (2008)
estimate place premiums for a Bolivian workers in the US. A typical Bolivian-born, Bolivian
educated, urban male, formal-sector wage worker with moderate schooling makes 4 times as
much in the US as in Bolivia. The paper adjusts for selectivity and compensation differentials, using a selection model to estimate how migrants’ wage gains depend on their position in the
distribution of unobserved wage determinants both at the origin and at the destination, as well as the relationship between these positions. Following all adjustments, Clemens et al estimate that the wages of a Bolivian worker of equal productivity, willing to move, would be higher by a factor of 2.7 solely by working in the United States. This result builds tacitly upon the rural-to-urban place premiums discussed in this paper. In both cases, the poorest Bolivians do not migrate to the extent predicted by place premiums. Migration costs are likely to be relatively high in both cases and difficult to capture with available household data in both countries.
No comments:
Post a Comment