Beruflich Dokumente
Kultur Dokumente
Trade Liberalization
on Child Labor
Eric EDMONDS and Nina PAVCNIK
Journal of International Economics 65(2) March 2005
Structure of Presentation:
Theory Motivation and Brief Literature
Review
The model, and its estimation strategy
The data
Estimation results
Robustness checks
Conclusion and quick criticisms
3)
The model:
The Data:
Two rounds of the Vietnam Living Standards Survey (VLSS), 1992-93 and 1997-98.
Two panel dimensions to dataset:
115 rural communities visited each round, random sample of children then aged
6-15 years in each round. 4,850 children in Round 1 and 4,703 in Round 2.
3,397 households with children re-visited, restricting survey to children then aged
6-15 years in these households. 4,586 in Round 1 and 4,441 in Round 2.
Child work:
Child labor if more than 7 hours of work during week before (aggregated) in:
Child works outside for pay (whether cash or in kind)?
Child works in agriculture for household.
Child works in household business.
Child works in household production activities, such as collecting water,
wood, building and maintaining house, household chores.
Both participation data, and total hours of work for each category recorded.
Household annual cropland assignments.
Household net production in 1992-93.
Detailed community price survey conducted in community market concurrently with
VLSS.
Price of 1 kg ordinary rice in 1993 and 1998.
Prices deflated with monthly CPI so prices are in 1998 currency units (Dong).
Fig 1: Community Level Rice Price Increases and Declines in Child Labor
Empirical Framework:
Linear probability model for participation in child labor market as a
response to rice price differences:
Notes:
i, j, t denote community I, child j, time t.
Rice prices only vary at the community-year dimension i.
Uses first dimension of panel not limited to the same households in both periods.
Huber-White robust standard errors, clustered at the community-year level.
Why not use Tobit with Honores Trimmed LAD fixed effects?
y:
RP:
M:
X:
Notes:
i, h, j, t denote community I, child j, household h, time t.
Rice prices only vary at the community-year dimension i.
Uses second dimension of panel limited to the same households in both periods.
Huber-White robust standard errors, clustered at the community-year level.
Conclusions:
1)
2)
3)
4)
Quick Criticisms: