The Economics of Processed Food: an Experiment using Continuous Glucose Monitoring

Over the last 40 years worldwide obesity rates have doubled and diabetes rates have quadrupled. Obesity is estimated to be killing at least 2.8 million adults per year and rising (European Association for the study of Obesity, n.d.; World Health Organisation, n.d.). With the current estimated yearly additional cost per obese person of $1723 USD, comprising medical, productivity, transport and human capital costs, it presents a huge public policy issue of how to reduce that cost and increase consumer health (Tsai, Williamson & Glick, 2011). A key factor in the nutritional quality of food, and often non-directly observable characteristic of food is the effect on blood glucose. High blood glucose is correlated with obesity, cardiovascular disease and diabetes, with different meals eliciting different glycaemic responses based on nutritional content and processing technique implemented (e.g. canning) (Jenkins et al, 1981; Food and Agriculturev1 Organization for the United Nations, n.d.). The purpose of this empirical study is to investigate the blood glucose effects and demand stimulating nature of processed food in free feeding conditions. Producers are incentivised to invest in sophisticated product design to stimulate appetite while reducing the costs of production, with the benefit increasing with producer size (Smith, 2011; Smith and Tasnádi, 2014)). The hypothesis driving this study is that due to the nature of processed food products being produced on a global scale processed food producers are more likely to employ appetite stimulating technologies than producers on a smaller scale. Glucostatic theory states that when blood glucose levels fall consumers will experience hunger, stimulating demand (Oxford Index, n.d.). The data used in this study were collected from 12 individuals who wore a continuous glucose monitor system (CGM), the Dexcom G4, over a 7 - 14 day period. Photographic food diaries were collected along with exercise logs during the study period. Postprandial blood glucose response curves were estimated using fixed effect models, controlling for nutritional information and other explanatory variables. Addiction-like dynamic parameters were also estimated. Meals produced at higher levels of economies of scale were found to have a stronger effect on blood glucose even when controlling for nutritional content (including all variables in the nutritional information panel) and meal size. Meals produced at lower and medium levels of economies of scale were not statistically significantly different from one another. This suggests the nutritional informational panel may not have the required information for consumers to infer the likely effects of particular products on blood glucose. To test more directly for addictive properties, meal-specific addiction indices were generated from the CGM time series for two hours following each meal. The addiction indices fall into two categories: dose (i.e., various measures of the magnitude of the post-prandial blood glucose shock) and rate of absorption (i.e., the slope of the response curve, measured at various times). We similarly find that both of these measures are significantly higher for globally-branded fast foods, and that this result is robust to controlling for meal size and nutrient content. The results highlight asymmetric informational issues between producers and consumers in the food industry with respect to postprandial blood glucose. Given the scale of the ongoing obesity epidemic, improvements to existing food labelling requirements could yield large benefits