Labor and Women's Nutrition: The Impact of Work Effort and Fertility on Nutritional Status in Ghana
利用加纳1987-88年数据，估计3130名女性的营养生产函数，发现工作时间和生育次数显著影响身体质量指数（BMI），其中17%的女性营养不良。
A nutrition production function was estimated for 3130 women aged 18 years or older with an average age of 38 using individual- household- and community-level data from the 1987-88 Ghana Living Standards (GLSS) incorporating control for energy expenditure. The theoretical model underlying the analysis was an extension of Rosenzweig and Schultz (1983) estimating the current nutritional status of the member of the household depending on a set of current nutrition-related inputs conditioned on individual and household characteristics and factors contributing to the current local health environment as well as the net effect of all relevant individual household and community factors. The dependent variable in the nutrition production function was the body mass index (BMI) defined as kilograms of weight per squared meter of height. 63% of the women were rural 41% resided in the moist forest zone and 24% in the semiarid savannah. The average parity was 4.5 children. 10% of the women had primary school and 28% had secondary school education. The normal BMI levels range between 18.5 and 23. Based on this 17% of the women were undernourished in Ghana and under 2% suffered from severe caloric deficiency. 73% of the women worked outside the home and 21% reported working at least two jobs. The model showing a typical nutrition production function excluding variables measuring energy expenditure due to work effort indicated that the caloric coefficient was relatively small and insignificant (p = 0.51). Expenditure on health care had a positive and significant effect on BMI (p = 0.0001). The fertility variable showed a significantly negative parity effect (p = 0.04). The other model included three time allocation variables and showed significance for the caloric coefficient (p = 0.01) and for the impact of recent morbidity (p = 0.03). All three time allocation variables were significant (p-values of 0.01 0.0001 and 0.04 respectively) supporting the hypothesis that an individuals time allocation pattern plays an important role in determining nutritional status.