The BMI: Measurement, physician costs and distributional decomposition

Aperçu

Résumé (français)

Veuillez noter que les résumés n'apparaissent que dans la langue de la publication et peuvent ne pas avoir de traduction.

Résumé (anglais)

This thesis comprises three chapters involving the analysis of the body mass index (BMI) in health economics. The first chapter evaluates two correction models that aim to address measurement error in self-reported (SR) BMI in survey data. This chapter is an addition to the literature as it utilizes two separate Canadian datasets to evaluate the transportability of these correction equations both over time and across different datasets. Our results indicate that the older method remains competitive and that when BMI is used as an independent variable, correction may even be unnecessary. The second chapter measures the relationship between long-term physician costs and BMI. The results show that obesity is associated with higher long-term physician costs only at older ages for males, but at all ages for females. We find that accounting for existing health conditions that are often associated with obesity does not explain the increase in long-term physician costs as BMI increases. This indicates that there is an underlying relationship between the two that we could not account for in our econometric models. Finally, the third chapter decomposes the differences in BMI distributions of Canada and the US. The results show that the differences between BMI levels, both over time and across countries, are increasing with BMI; meaning the highest difference is observed at the right tail of the two distributions. In analysis comparing two points in time, these differences are solely due to differences in the returns from attributes and the omitted variables that we cannot account for in our models. In cross-country analysis, there is evidence that the differences observed below the mean can be explained by the differences in characteristics of the two populations. The differences observed above the mean are again due to those in returns and the omitted variables.