This dissertation involves research in two important areas of economics: one is in the field of Econometric methodology while the other deals with a hotly debated topic in Development Economics. First, new methods for estimating panel data models using nonparametric and semiparametric estimation techniques are developed. Then, the effect of a change in household income on individual calorie intake is studied for a rural South Indian data set using the newly developed nonparametric and semiparametric estimators.;The existing literature on panel data econometrics has been mostly confined to linear parametric models, even though, it is well known that misspecification of linear or even non-linear parametric models may lead to inconsistent and inefficient estimates and suboptimal test statistics. The contribution of this dissertation in the panel data literature is that it presents some nonparametric and semiparametric estimators for two of the most popular panel data models, fixed effects and random effects models. The advantage of these new estimators over the existing parametric estimators is that they are robust to misspecification of the functional form since nonparametric modeling is data-based modeling.;Most studies in the literature on poverty and malnutrition have used cross-sectional data to analyze the calorie-income relation which fails to take into account heterogeneity and dynamics. There is, however, a panel data set from the International Crops Research Institute for Semi-Arid Tropics Village Level Studies (ICRISAT VLS) on rural south central India which has been used by some researchers in this area. While they have controlled for heterogeneity given the panel nature of the data, they failed to take care of the functional form problem. One way to take care of the functional form problem is possible through the use of nonparametric and semiparametric methods. This dissertation presents a study of the calorie-income relation for the same ICRISAT data by using the new nonparametric and semiparametric estimators for panel models which takes care of both heterogeneity and functional form misspecification problems.;The empirical findings in this study show that the income elasticity of calorie intake is, in general, small but it is nonzero and statistically significant. The nonparametric and the semiparametric analyses indicate that the calorie response to income change is higher for the relatively poor households in the sample and that the elasticity differs across gender.
展开▼