Disintermediation Evidence From a Cleaning Platform

Abstract

We present evidence of disintermediation---a notoriously challenging task---in a European cleaning platform's transaction data between April 2016 and February 2020. In short, we demonstrate that a cleaner's leaving the platform is typically accompanied by unusually many customers' also leaving the platform. This pattern (i) replicates in most cities in our data, (ii) is robust to alternative specifications of the empirical exercise, and (iii) cannot be explained away by bad cleaners driving customers away, cleaners and customers trying out the platform once or twice and then exiting, or customers being unable to find their favorite cleaner on the platform. Further, we project that the platform would enjoy about 24 percent more cleanings if not for disintermediation. Our approach (empirical test plus counterfactual simulation) only requires run-of-the-mill, observational transaction data and can be applied to any similar setting.