How much — and why — have product markups increased over the last several decades?1 Since markups serve as indicators of market competitiveness, consumer welfare, and firm behavior, addressing these questions is important for understanding the economy and formulating policies that foster fair and efficient markets. For their paper, “Scalable Demand and Markups,” Philadelphia Fed Economic Advisor and Economist Enghin Atalay, Securities and Exchange Commission Financial Economist Erika Frost, University of Wisconsin–Madison Professor of Economics Alan Sorensen, University of Wisconsin–Madison Assistant Professor of Economics Christopher Sullivan, and Economist Wanjia Zhu introduce new methods to measure markups at an aggregate level, building on approaches developed by microeconomists.
Traditionally, microeconomists estimate consumer demand (using demand curves) for each individual product by analyzing the relationship between the demand for each product and its various characteristics.2 According to this approach, consumers view pairs of products as highly substitutable with one another if the products share characteristics that they care about. Microeconomists need to know which sets of products are close substitutes if they are to identify demand curves for these individual products. However, understanding the relevant product attributes and collecting the necessary data on these attributes is too challenging and time-consuming to do for many distinct markets. For this reason, microeconomists focus almost exclusively on a single market — or, at most, on a handful of markets. But focusing on just one or only a few markets precludes understanding general trends in markups.
To overcome this challenge, Atalay and his coauthors provided a method that emulates the microeconomic approach but can be scaled across many product markets. Instead of relying on products’ characteristics to determine which sets of products are like one another, the authors hypothesized that two products are more likely to be alike if an individual household switched between the two products at different points in time. This approach allowed them to cluster products within a product market into groups of similar products. (In other words, they were able to automate the assignment of products into “nests.”) As a result, they could study consumers’ substitution patterns. Having assigned products into nests, the authors then estimated what is called a “nested logit” demand model.
Once the authors had estimated the demand curves, they could infer firms’ markups for each product that they supplied to the market. Specifically, they analyzed markups for 33,000 products sold between 2006 and 2018, representing 72 distinct product markets. Their sample contains products that are exclusively found in grocery stores, based on a supermarket scanner data set.
The authors found that there was a significant overall upward trend in markups across many product markets between 2006 and 2018, which is consistent with the prior literature. Markups were higher in 51 out of 72 product categories, with the median markup across all products approximately 10 percentage points higher over the sample period.
However, there were significant differences in markups across product markets. For example, low-calorie soft drinks started with relatively low median markups that then doubled over the sample period, whereas light beer had relatively low median markups across all the sample years, and yogurt had higher markups at both the beginning and the end of the sample. Overall, products with greater market shares tended to have higher markups (and increasingly so over the course of the sample period).
In a final step of their analysis, the authors explored why markups have tended to increase. They attribute higher markups, on average and in part, to consumers becoming less price sensitive. (This more insensitive — that is, inelastic — consumer demand may have resulted, they speculate, from factors such as shifts in demographics and firm advertising behavior.) Another key factor explaining the higher markups over the sample period, they find, is that firms had overall lower marginal costs in producing consumer packaged goods. (Markups are the gap between prices and marginal costs, so lower costs tend to imply higher markups.) They find that another potential factor, market concentration, had only a minor impact on markups over the sample period.
In summary, the paper by Enghin Atalay and his coauthors, given its novel approach using foundations in microeconomics and nested logit estimation, complements prior studies that similarly find higher markups over time across a wide range of products. Their study also contributes to the literature by detecting considerable differences in markups both within and across product markets.
- The views expressed here are solely those of the author and do not necessarily reflect the views of the Federal Reserve Bank of Philadelphia or the Federal Reserve System.
- A product’s markup is defined as one minus the ratio of the product’s marginal cost and its price.
- Once they have estimated these demand curves, they can apply an equation describing firms’ profit-maximizing price for each product to obtain estimates of each product’s marginal cost and markup.