Studying the child obesity epidemic with natural experiments

"We utilize clinical records of successive visits by children to pediatric clinics in Indianapolis to estimate the effects on their body mass of environmental changes near their homes. We compare results for fixed-residence children with those for cross-sectional data. Our environmental factors are fast food restaurants, supermarkets, parks, trails, and violent crimes, and 13 types of recreational amenities derived from the interpretation of annual aerial photographs. We looked for responses to these factors changing within buffers of 0.1, 0.25, 0.5, and 1 mile. We found that cross-sectional estimates are quite different from the Fixed Effects estimates of the impacts of amenities locating near a child. In cross section nearby fast food restaurants were associated with higher BMI and supermarkets with lower BMI. These results were reversed in the FE estimates. The recreational amenities that appear to lower children's BMI were fitness areas, kickball diamonds, and volleyball courts. We estimated that locating these amenities near their homes could reduce the weight of an overweight eight-year old boy by 3 to 6 pounds"--National Bureau of Economic Research web site.