400 likes | 612 Views
Migrant Opportunity and the Educational Attainment of Youth in Rural China . Alan de Brauw IFPRI John Giles The World Bank April 10, 2008. Growth, Opportunity and School Enrollment. Do economic growth and reduction of barriers to labor mobility lead to more human capital investment?
E N D
Migrant Opportunity and the Educational Attainment of Youth in Rural China Alan de Brauw IFPRI John Giles The World Bank April 10, 2008
Growth, Opportunity and School Enrollment • Do economic growth and reduction of barriers to labor mobility lead to more human capital investment? • Trade-off: continue in school or look for work • Studied in other developing countries • Vietnam: Glewwe and Jacoby (1998, 2004); Edmonds (2004). • India: Kochar (2004).
Our Question and Answer • How does increasing “migrant opportunity” affect high school enrollment in rural China? • “Migrant opportunity” or cost of migrating proxied by size of the village labor force employed as migrants. • Why high school? • Answer: Less likely.
Outline of the Talk • Descriptive Evidence: • Role of networks in migration and job search • Educational attainment of rural migrants in China’s cities • The Primary Data Source (RCRE surveys) • Educational attainment of residents in RCRE villages • Migration behavior from RCRE villages • Econometric Framework • Identification Strategy for the Migrant Network • Discussion/Interpretation of Results • Other Possible Mechanisms
Migrant Networks in China • As elsewhere, migrants in China use networks to find jobs • In an urban survey (2001): • 94% of migrants know someone in city before migrating • 35% have a close family member, 58% an extended family • 56% had arranged a job prior to first migration experience
Human Capital of Rural-Urban Migrants in Urban Areas • Education • 86 % have middle school educational attainment or less (67 % complete middle school). • 14 % have completed high school. • Rural Residents Not Engaged in Farming • 21 % have completed high school. • Migrants Tend to be Young • 20 % left village before 18th birthday • 77 % before 40th birthday
Data: Annual RCRE Household and Village Surveys and 2004 Supplemental Survey • Primary Data Source: Ministry of Agriculture, Research Center on the Rural Economy (RCRE), Household and Village Surveys from 1986 to 2003. • We use data from Anhui, Jiangsu, Henan, and Shanxi • 52 villages in sample, visited annually since 1986 • We use several variables from annual surveys • Number of migrants from village, other village and household level variables
RCRE Supplemental Survey, Summer 2004 • Resurveyed 3999 households • Enumerated educational attainment and work histories • Enumerated all current and past HH residents (over 16,000 individuals) • We have information on all children of household head. • We know educational attainment regardless of residence in village. • Focus for analysis: 3160 individuals who completed middle school between 1986 and 2003.
Share of Age Group with Temporary or Long-Term Migrant Employment
Outline of Theory • Parents are concerned with current and future consumption and the expected future wage of children. • Choice over whether to send a child to high school is influenced by: • Wealth • Credit Markets • Current returns to middle school completion • Expected future benefits from high school graduation • Preferences
Theory (Continued) • Parents enroll children in high school if expected benefits outweigh costs. • Positive effect could be explained by increasing wealth, relaxing credit constraints, or expected returns to education in urban areas. • Negative effect if returns to education are low for migrants or potential migrants.
Enrollment Demand • Migrant opportunity, Mvt, measured as the number of village residents employed as migrants outside the county in each year. • If networks are important, increases are associated with a declining cost of migrating. • Using a linear probability model, we estimate the effect on high school enrollment:
Problem with Simple Estimation • Mreflects factors that influence both the supply and demand of migrants • Sources of Bias: • Positive: if high schools expanded and lowered admissions standards while migrant opportunity also increased. • Negative: negative shock to the local economy makes migration attractive while making it harder to cover high school tuition.
Identification Strategy • Two policy changes • National ID card introduced in 1984 • 1988 Reform of residential registration system • Residents of different counties received IDs at different times • Farmers could not simply move to get ID cards
Identification Strategy • We argue that… • Differences in timing of ID availability affects network quality. • “Cost” of migrating falls as legal long-term migrants are capable of providing referrals. • Networks time to build up. • Non-linear function of years since IDs were issued used to identify the migrant network.
Potential Problems with Our Instrument • Timing of ID Card distribution is not random • Differences in unobservable village characteristics could affect migration network and educational attainment • Demand for migration could have driven ID distribution • Timing of ID distribution could be correlated with trends in educational attainment
Implementing our IV Strategy • We look for observable differences between villages
Implementing our IV Strategy • We look for observable differences between villages • Look for obvious correlation between trends in educational attainment and timing of ID card distribution
Share of Middle School Graduates Entering High School by Timing of ID Card Receipt
Implementing our IV Strategy • We look for observable differences between villages • Look for obvious correlation between trends in educational attainment and timing of ID card distribution • In estimation, we include village dummies to control for time-invariant unobservables. • Test robustness to inclusion of additional time-varying variables that are likely correlated with village level unobservables.
The Reduced Form Effect of Issuing IDs F-Statistic on quartic in years since ID was issued is 3.19.
Heterogeneity in Migration Effect Across Families? • In spite of instrument, perhaps concern that we are “picking up students who might not go to high school anyway.” • Who goes to high school? Children from families who have connections, who can afford it, or who have preferences for education. • Father is “professional” • Father has wage employment experience • Father has at least some high school education
Results: Migrant Opportunity and Family Characteristics (fix this)
If Not School, What Are High School Age Children Doing? • Consider activity choice among teenagers of high school age. • Look for effects of network on migrant and local employment (general equilibrium effects).
Share of Young Pursuing Activities Other than Migrant Employment
Other Mechanisms • A unitary model that does not examine consequences of migration on family composition. • Possibility that our result is not driven by low returns to high school, but by effects of parent migration on child educational attainment.
Years-Since IDs Issued and the Migration of Fathers with 7-12 Year Old Children
Years-Since IDs Issued and the Migration of Fathers with 7-12 Year Old Children
Years-Since IDs Issued and the Migration of Fathers with 13-15 Year Old Children
Years-Since IDs Issued and the Migration of Fathers with 13-15 Year Old Children
Conclusion • Migrant opportunity has a fairly significant, negative effect on educational attainment in rural China • Elasticity at mean size of migrant network: -0.191 • Robust to several extensions of the model • Inclusion of time-varying variables at village level potentially related to unobservables • Inclusion of a range of family characteristics
Discussion • Why? Occupational segregation in cities. • Migrants not employed in jobs that require HS/College education • Likely reinforced by general equilibrium effects and higher wages locally subsequent to depletion of the labor force • Irreversibility of the high school enrollment decision • Implications for long-term inequality within urban areas.