Physical Activity, Obesity and Present Bias

Promoting regular physical activity and reducing disparities in physical activity has been and continues to be an important public policy priority because of the many health benefits associated with physical activity.   Physically active people are less likely to be obese and, because they are not obese, are at lower risk for many chronic diseases like diabetes, heart disease and cancer.  Despite all we know about the health benefits of physical activity and numerous campaigns to promote it, many Americans still do not engage in the recommended levels of physical activity to achieve sustainable health benefits.  Physical activity, like many other health behaviors is episodic.  Some people exercise regularly, while others do not.  Some people begin to exercise regularly but then stop and then start again. An improved understanding of these patterns in physical activity behavior may help to design and implement more effective policies aimed at promoting physical activity.  The objective of this research is to develop and empirically test a behavioral model of physical activity and obesity. 

Our model builds on the dynamic inconsistency models of O'Donoghue and Rabin (1999) and Kremer et al (2011). We assume heterogeneous agents of two types:  consistent agents and stochastically present-biased agents as in Kremer et al (2011).  We borrow from O’Donoghue and Rabin (1999) in describing a dynamic-inconsistent utility function where agents are present-biased.  Present-biased agents will procrastinate if a task has immediate costs and delayed rewards. Physical activity can be viewed as a task with immediate costs (in terms of time and effort) and delayed rewards. The reward structure of physical activity generates long-run rewards only when the task is done consistently over time.   The combination of heterogeneous agents with stochastically present bias can explain why some people always participate in physical activity but others participate sporadically.  This model can also explain over-eating and yo-yo diet behavior. Over-eating has immediate rewards but long-term health costs if it is consistently done. A present-biased person tends to over-eat. With stochastically present bias, a person will diet sporadically. 

The critical aspect of empirically testing the model lies in finding a reasonable proxy for present bias.  We use NHANES data to construct measures of present-biasness based on questions about smoking, drinking, and dental care. People who have difficulties quitting smoking, scheduling regular routine dental check-ups, and moderating drinking may have present bias and tend to procrastinate.  The NHANES also asks questions about physical activity and weight.  We use these proxies of present bias in regression models of physical activity participation. We include individual characteristics such as socioeconomic status in the regressions that allow us to evaluate disparities in physical activity.