Streaming Merger Market Definition

 

One aspect of the antitrust review of the Warner Brothers Discovery (WBD) merger with Netflix (or Paramount) will be what constitutes the relevant market. Eric Fruits provides a nice explanation of the issues over on "Truth on the Market." It essentially boils down to whether a narrow "streaming services" definition is used versus a broader "screen time" definition that includes recreational Internet scrolling and maybe video gaming. The market would be quite concentrated under the former definition and quite a bit less so under the latter. This is an empirical question over the extent to which consumers substitute their time between various screen content. 

I happen to have some experience with time use data from the ATUS from some of my past research. These data are amazing with a quarter million "diary days" covering every day since 2003. Other nice things about these data are that they are publicly available and have consistent definitions over almost a quarter century. Among other activities, these data include time spent watching TV, playing games (mostly video games), and "recreational computer" usage. A major problem with ATUS for screen time measurement is that, because it was setup before smartphones were a thing, there no good way of measuring time spent looking at your smartphone while you are doing something else. The amount of computer time in ATUS is a fraction of time on smartphones reported elsewhere. So I spent the morning seeing how time spent on these three activities related to each other. My quick and dirty analysis indicates that each minute playing games decreases TV time by 4 seconds [P<0.01] while each minute "recreating" with a computer decreases TV time by 8 seconds [P<0.01].* If you confine the sample to just the past 10 years, you get slightly more time diversion. This is evidence suggesting that consumers do substitute between television and other screen time.

Surely the parties, whoever they end up being, will have more granular proprietary data yielding better analyses. 

 

*This analysis includes fixed effects for age category by sex and year by sex and uses ATUS's weights. Interpreting these correlations as causal is problematic. Most of the variation is likely to come from ever better video games and ever more Internet activities (e.g., YouTube, Facebook, TikTok, etc) which would suggest a causal interpretation. But this analysis is merely conditional correlations.