Recent clinical research has studied weight responses to varying diet composition, but the contribution of changes in macronutrient intake and physical activity to rising population weight remains controversial. Research on the economics of obesity typically assumes a "calories in, calories out" framework, but a weight production model separating caloric intake into carbohydrates, fat, and protein, has not been explored in an economic framework. To estimate the contributions of changes in macronutrient intake and physical activity to changes in population weight, we conducted dynamic time series and structural VAR analyses of U.S. data between 1974 and 2006 and a panel analysis of 164 countries between 2001 and 2010. Findings from all analyses suggest that increases in carbohydrates are most strongly and positively associated with increases in obesity prevalence even when controlling for changes in total caloric intake and occupation-related physical activity. Our structural VAR results suggest that, on the margin, a 1% increase in carbohydrates intake yields a 1.01 point increase in obesity prevalence over 5 years while an equal percent increase in fat intake decreases obesity prevalence by 0.24 points.