Data and Code for: “What is newsworthy? Theory and evidence"

We introduce a model in which a benevolent news outlet decides whether to report the realization of a state to a consumer, who pays a cost to receive it. A simple statistical rule, called a proper scoring rule, describes when the outlet should be more likely to report the realization. Using data from the US television news, we show that a particular scoring rule successfully predicts many salient features of news reporting. We show how to use this rule as a control variable to discipline tests of reporting bias, and we show that controlling for it matters in our applications.