Convex self-control cost and Calorie Information

Calorie labeling was introduced as an important policy tool to fight obesity. However, the relevant academic literature reports conflicting results regarding the consequences of the policy change. Some studies show that calorie labeling has decreased calorie intake (Bollinger et al., 2011; Ellison et al., 2013). However, many other studies (Elbel et al. 2009; Vadiveloo et al., 2011; Finkelstein et al., 2011; Swartz et al., 2011; Bowen et al., 2015) find no significant evidence that calorie labeling reduced calorie consumption. A growing economic literature argues that visceral feelings – such as temptation can burden consumers with self-control costs and can cause non-welfare increasing outcomes (Gul & Pesendorfer, 2001). One of the main predictions of this literature is that the inclusion of a more tempting alternative into the menu may non-linearly increase the self-control cost. For instance, an economic agent may find the temptation to eat a small cake resistible and eventually may prefer not eating any desert in the {no desert, small cake} menu. However, he may change his mind and choose the small cake in the {no desert, small cake, big cake} menu.

We conducted an experiment where we exogenously varied calorie distances between food snacks in menus. In High (Low) menus the calorie distance between alternatives was greater (less or equal to) than 40 calories. We also introduced two treatments: 1) “Homegrown Information” – where subjects guessed the calorie content of food products and made food decision based on their prior knowledge. 2) “Accurate Information” – where subjects chose food products based on the provided accurate calorie information.

We find that subjects chose less-calorie options more frequently when they face Low menus compared to High menus. The convex self-control cost phenomenon is observed primarily in menus where less calorie also meant less sugar in the healthy alternative. We also find that the accurate calorie information helped subjects to choose less-calorie alternatives more frequently in menus where the calorie trade-off was via sugar content. Our findings help to resolve the central contradiction in studies which tried to assess the effect of the provision of the calorie information. We show that the reported conflicting results may stem from the fact that studies do not differentiate sugar-intensive products from fat-intensive alternatives.

We also conducted a post hoc analysis with collected eye-tracking data. With the help of the drift-diffusion model, we estimated the information subjects sampled from food product pictures. It helped us to disintegrate the calorie information from the food information. We find that non-calorie product information helps subjects to choose healthy alternatives more frequently when they receive accurate calorie information. However, this effect is reversed when they choose food products with their homegrown beliefs. Overall, our study contributes to this rapidly growing literature by 1) suggesting possible reasons behind the contradictory results in the literature, 2) testing the convex self-control cost assumption of theoretical self-control literature with empirical findings, and 3) also showing the interaction of the calorie information with other food product cues which has mostly under-investigated in the literature.