Essays on physical exercise as a behavioral determinant of U.S. protein demand

Abstract

Consumer preference and price sensitivity is a major driver of the U.S. food system. Consumers’ food purchasing decisions have implications not only for individual health outcomes but also have upstream impacts on food retailing and manufacturing decisions, commodity production decisions, and prices experienced across all levels of the supply chain. In this dissertation, I seek to better understand the consumer side of the U.S. food system by evaluating a, thus far, unstudied determinant of food consumption and demand—physical exercise. Specifically, I focus on the relationship between physical exercise and animal protein purchasing and consumption. Nutritional studies and various media outlets have, for decades, discussed this relationship but no research has been conducted to quantify its economic impact. The remainder of this abstract outlines my four dissertation chapters. In Chapter One, I determine the impacts of media information related to physical exercise on U.S. meat demand from 1993 through 2022. Including news media indices that capture publicly available information related to physical exercise and protein consumption in a food-separable Rotterdam demand model, I calculate news media elasticities. I find little to no evidence that news media attention to physical exercise has impacted aggregate demand for beef, pork, or chicken over time. Further, this media attention did not shift U.S. food demand into or out of the aggregate meat complex. This chapter has been accepted for publication in the Journal of Agricultural and Resource Economics. Chapter Two examines the relationship between physical exercise and animal protein consumption (rather than demand). Using micro-level dietary recall and recreation data, I estimate the association of physical exercise with consumption of total protein, beef, pork, poultry, seafood, eggs, and dairy among U.S. adults. I find nonlinear associations, with those who exercise between two and three hours per day having the highest level of total protein consumption. The direction and magnitude of association of physical exercise with disaggregate animal protein consumption differed by protein source. This confirmed what has been observed anecdotally in popular press and social media—not all protein sources are equal in the eyes of those who exercise. This chapter has been accepted for publication in Nutrients. Having assessed physical exercise and its relationship with animal protein consumption, I then examine its relationship with animal protein demand. In Chapter Three, I utilize a novel source of consumer-level data in an AIDS framework to calculate price and expenditure elasticities separately for those who have exercise-related goals and those who do not. Specifically, an “open-ended” choice experiment is used to obtain survey respondents’ purchasing information, which is then used to estimate the AIDS. I find that survey respondents who intentionally consume protein to meet their exercise goals have statistically lower own-price sensitivity for all evaluated protein goods. I further document the difficulties experienced in using open-ended choice experiment data and how future research could validate its use. While analyzing the publicly available survey data used in Chapter Three, I discovered discrepancies in reported prior day food consumption between those who are exercise- and fitness-conscious and those who are not. For instance, eleven percent of survey participants who reported intentionally consuming protein to aid toward some fitness-related goal had also consumed eggs in the prior day. This is compared to eight percent of all other survey participants. I did not pursue this particular phenomenon but did note that eggs and other protein-dense food items (e.g., whey protein powder) were not included in either the survey’s open-ended choice experiment or traditional, “choose one” choice experiment. With a sizable portion of a nationally representative sample indicating that they consume eggs, such omission of this product in the experimental design seemed questionable. These observations, and prior research finding that choice experiment results are sensitive to construction of the choice set, motivated Chapter Four. Chapter Four breaks from the theme of physical exercise but remains under the broader umbrella of consumer behavior and, further, delves more into the issue of experimental design that was broached in Chapter Three. In this chapter, which is intended to improve accuracy of consumer demand estimation, I consider the concept of separability (i.e., separability of goods) in choice experiment research. Specifically, I distribute several surveys that differ only in the food alternatives provided (such as eggs and whey protein powder mentioned above) in a discrete choice experiment and then estimate random parameters logit models of choice separately for each survey variant. Economic measures such as willingness-to-pay and market shares are calculated and compared across surveys. I find that slightly altering survey respondents’ choice sets has substantial impact on mean willingness-to-pay estimates, the distribution of willingness-to-pay, and market shares—even for food alternatives that were available in every survey variant. In sum, the findings from these chapters can be used to inform pricing and marketing strategies of U.S. food manufacturers and retailers, with these strategies having corresponding impacts on both livestock producers and consumers of animal protein. Further, future choice experiment design efforts may consider this work to improve economic conclusions.