We study changes in markups across 72 product markets from 2006 to 2018. A growing literature has documented a rise in markups over time using a production function approach; we instead employ the standard microeconomic method, which is to estimate demand and then invert firms’ first-order pricing conditions to infer their markups. To make the method scalable, we propose estimating nested logit demand models, using household panel data to automate the assignment of products to nests. Our results indicate an overall upward trend in markups between 2006 and 2018, with considerable heterogeneity across and within product markets. We find that changes in firms’ marginal costs and households’ price sensitivity are the primary drivers of markup increases, with changes in firm ownership playing a much smaller role.