6/ For example, suppose the number of firms in a consumer’s choice set is a random variable that is negative binomial.

Suppose consumers draw utility shocks from a uniform distribution.

Greater consumer heterogeneity increases the average markup.

5/ The paper generalizes a nice result about extreme value outcomes in Gabaix, Laibson, Li, Li, Resnick and de Vries (2016) by incorporating heterogeneity across agents.

For example, the paper delivers a generalization of a result on markups in Gabaix et al.

4/ How do these new EVDs arise?

For example, suppose the number of options an agent gets is a random variable that is negative binomial.

Suppose options are drawn from a Pareto distribution.

The EVD takes this general form. 👇

3/ The paper presents a new class of extreme value distributions (EVDs) that generalizes the three standard EVDs (Fréchet, Gumbel, Weibull) by incorporating heterogeneity across agents.

Here is an example of a new family of EVDs. 👇

Economic outcomes often depend on the distribution of some maximum value (e.g. the highest valuation, best idea, or lowest cost).

If the average number of options is large, do such outcomes change when some agents have more options than others?

1/ A thread about this paper 🧵