Estimating the Distribution of the Body Mass Index
Posted: 14 Jun 2007
Obesity is the most important issue in food markets. It is common knowledge that over the last half-century obesity levels in North America have increased. While means and order statistics (i.e., quartiles) are easily observed, their use ignores important information contained in the complete distribution of body mass indices (BMI) or other measures of obesity. Fully characterizing the shape of the BMI distribution will be an important first step in constructing policies that directly target obesity. An increase in average BMI could result from a location shift of the distribution or, alternatively, from a change in the shape of the distribution. These distinct phenomena will yield different policy prescriptions. Additionally, previous work has ignored the complete conditional distribution of BMIs over different demographic characteristics. For instance, very little is understood regarding the shape of the distribution of BMIs conditional upon different levels of income or education. Our study seeks to understand how the distribution of BMI has changed over time. We (1) develop a methodology to model the complete distribution of BMI and the resulting levels of obesity in the U.S and Canada; (2) examine how the distribution of BMI in the U.S. and in Canada has changed over time; and (3) compare and contrast the distribution of BMI in Canada and the U.S. over time, and relate these changes back to agricultural policy.
Suggested Citation: Suggested Citation