Garment Industry in Bangladesh

Heath and Mobarak (2015) Aims

• Examines the ready-made garment industry's growth in Bangladesh and its effects on women's schooling, employment, and childbearing choices.
• Manufacturing expansion is usually related to increased female labor force participation (FLFP).
• There is limited understanding of the welfare effects.

Context

• Research was conducted in 4 subdistricts in Bangladesh
• Garments are Bangladesh's largest export, primarily employing young women.

Theory

• Factory jobs can affect education through several mechanisms:
  • Older children may drop out of school to take jobs.
  • Younger children may attend school if jobs increase the returns to education.
  • Wealth effects occur as parents can now afford to send children to school.

Data

• Baseline surveys were conducted in 2009.
  • This included information on each household member, both current residents and those who had migrated out.
  • Data was collected on schooling (years, timing), marriage, and childbearing history of all offspring in the household.
• Villages were categorized as garment-proximate (44) or non-garment (16).
  • Categorization provided by the Garment Manufacturers Association.
• Start dates and operational status of factories were determined via a 2014 survey of knowledgeable individuals.
• The sample consisted of 40 villages in peri-urban areas outside Dhaka city.

Method

• The study compares schooling, employment, childbearing, and marriage decisions of:
  • Women near garment factories
  • Women far from garment factories
  • Women before the garment factory was nearby
  • Male decisions
• Benchmarking was also conducted against the Female Stipend Program (1994 for girls in 6th grade and above).

Identification Strategy

• Difference-in-differences (Diff-in-Diff) for FLFP (female labor force participation).
• Discrete-time hazard models were used for employment opportunities, age at marriage, and age at first birth.
  • These models account for greater time exposure to factory jobs.
• Controlled for differences between garment-proximate and non-garment villages and for different time trends.
• Assumed exposure to the sector is localized.
• Interviews ruled out reverse causality.
  • Firms locate due to infrastructure and convenience of using buildings already owned.

Regression Models: FLFP

• Difference-in-difference estimates whether female labor force participation is higher in garment-proximate villages due to garment job exposure at critical ages.
• δ = the additional probability of working for a girl exposed to an additional year of garment job, compared to a girl in a garment-proximate village with less exposure.

Regression Models: Childbearing

• Uses a discrete-time hazard model.
• The probability that a girl marries or gives birth in a year depends on years of exposure to garment jobs.
• GarmentVillage x YearsExposure = total years of exposure throughout life
  • Captures labor force and education effects.
• Fixed effects and gender enrollment trends are included.

Regression Models: Educational Attainment

• Examines the effect of GarmentVillage x YearsExposure on years of education for girls and boys.
• δ2 = the effect of the garment industry on girls in a garment village relative to their brothers.
• δ1 + δ2 = the effect on girls in a garment-proximate village compared to girls in a non-garment village.

Regression Models: Enrollment

• The interaction GarmentVillage x PostGarments shows the effect of opening a factory in a garment-proximate village.
• δ1 = effect of factory on a young girl's relative enrolment decision, compared to her brother
• δ1 + δ2 = the enrolment of young girls in a garment-proximate village compared to girls in a non-garment village.
• δ3 = how the enrolment effect relative to brothers varies with age.
• δ3 + δ4 = how enrolment relative to girls in non-garment villages varies with age.

Main Mechanism

• Girls near factories increase schooling and delay marriage and childbirth.
  • Older girls are more likely to be employed outside the house.
  • There is an opportunity cost of marriage and children.
  • Managers prefer educated workers.
  • Better jobs are available for educated workers.
  • Young girls are more likely to go to school due to job opportunities.

Results

• Exposure to garment factory jobs at ages 10-23 increases the probability that a woman works outside the home.
• The hazard of marriage between 12-18 decreases with exposure.
• Girls gain an extra 1.5 years of education relative to males with exposure.
  • Observed even when the mother/older sister never had a factory job.
  • Enrolment increases the strongest in young girls.
• Factory proximity had a small negative effect on school enrolment for 17-18-year-old girls, as they went to work in factories instead.

Female Stipend Program Results

• No evidence that girls who had been in school for at least 6 years were more likely to remain in school after 1994.
  • The interaction was, in fact, negative.
• No evidence of hazard of marriage/childbearing decreasing in 1994 for girls who had been in school for 6 years, relative to girls who hadn't.
  • The coefficient of interaction was close to 0.

Tables

• Tables show factory & FLFP, factory & marriage & childbearing, factory & educational attainment, female stipend school enrolment and female stipend marriage & childbearing

External Validity

• The sample uses woven garment factories (54% women) while the industry mainly consists of knitwear (80% women employed).
  • It is argued that affects girls nationwide in the same way as female jobs are similar in both.

Policy Implications

• Observations showed a larger effect than conditional cash transfers (FSP).
• Investments in education and understanding the demand-side requirements of basic literacy/numeracy are important.
  • Human capital can increase by promoting manufacturing exports.
• Example of increasing gender participation & equality in the workforce

Criticisms: Omitted Variable Bias (OVB)

• There could be OVB (omitted variable bias) e.g., new roads built where factories are located.
  • Effects only on girls' marriage and childbearing decisions, not boys
    • No OVB on marriage market/cost of childbearing
  • Stronger effect on girls' schooling than boys
    • Infrastructure would have to be gender-specific to cause differential