This study scrutinizes spatial econometric models and specifications of crop yield response functions to provide a robust evaluation of empirical alternatives available to researchers. We specify 14 competing panel regression models of crop yield response to weather and site characteristics. Using county corn yields in the US, this study implements in-sample, out-of-sample, and bootstrapped out-of-sample prediction performance comparisons. Descriptive propositions and empirical results demonstrate the importance of spatial correlation and empirically support the fixed effects model with spatially dependent error structures. This study also emphasizes the importance of extensive model specification testing and evaluation of selection criteria for prediction.

This study examined the relationship between the total fertility rate and under-five child sex ratio to understand the role of fertility in the phenomenon of missing girls in India. Using data from the last four decennial censuses for the fifteen major states of India and their districts, covering more than 90% of the population of India, the study showed that there was a major decline in the female to male child sex ratio from 1981 to 2011 in most of the major Indian states and their districts. The panel regression model showed that the total fertility rate was significantly associated with the under-five child sex ratio at the district level for the 30-year period from 1981 to 2011 in India, even after controlling for other factors and any other unobserved heterogeneity. This indicates that areas of India with the highest fertility had the higher female to male child sex ratio, while low-fertility districts had a more male-biased sex ratio.