Recently prostate and ovarian cancer has been related to milk consumption. However, existing observational studies based on country level data do not attempt to identify causal effects since they are only based on simple cross-sectional analyses. This paper takes a step toward estimating of causal effects of milk consumption on cancer by applying panel econometric models and by using the within-country variation of the mortality rates and food consumption instead of the between-country variation in a panel of up to 50 countries for 1990 to 2008. Possible methodological problems arising from omitted variables (confounding factors), heterogeneity, and outliers are carefully discussed and a wide range of recent panel econometric estimators are applied. The results indicate fairly well that milk consumption increases both the mortality rate of prostate cancer as well as the mortality rate of ovarian cancer. The estimated effects are also important in quantitative terms, i.e., a reduction in the consumption of milk products can reduce the number of people dying of prostate and ovarian cancer appreciably. Furthermore, the consumption of other animal food products as well as sugar seems to be harmful. For the mortality rate of ovarian cancer we find that total calories intake increases the mortality rate too.
Panel Econometrics GMM Dynamic Panel Data Methods Fixed-Effects Quantile Regression Prostate Cancer Ovarian Cancer Cross-Country Analysis Causal Effect Quantile Regression Bayesian Model Averaging Extreme Bounds Analysis