Radio Listeners Seem to Buy Less Music


Or, More Fun With Correlation and Causality

An article in yesterday’s New York Times highlighted a recent study that found that “very roughly, an hour’s worth of radio listening per person per day, over the course of a year, corresponded with a 0.75 drop in the number of albums purchased per capita in a given city.” The study was authored by Stan Liebowitz, an economics professor at the University of Texas at Dallas, and compared record sales and music radio listening in 100 American cities from 1998 to 2003.

Interesting note: the period being evaluated starts with 1998, two years after the Telecommunications Act of 1996 eliminated the national cap on radio station ownership and increased the number of radio stations that a company could own in a single market.

The Future of Music Coalition has also looked at trends in radio during this period, and come up with some other interesting little tidbits of information. Take, for example, the items listed under the heading Radio Programming in the Wake of Consolidation in their December 2006 report “False Premises, False Promises:”

  • Homogenized Programming: Just fifteen formats make up 76% of commercial programming.
  • Large Station Groups Program Narrowly: Owners who exceed or exactly meet the local ownership cap tend to program heavily in just eight formats.
  • Only Small Station Groups Offer Niche Formats: Niche musical formats like Classical, Jazz, Americana, Bluegrass, New Rock, and Folk, where they exist, are provided almost exclusively by smaller station groups.
  • Small Station Groups Sustain Public-Interest Programming: Children’s programming, religious programming, foreign-language and ethnic-community programming, are also predominantly provided by smaller station groups.
  • Format Overlap Remains Extensive: Radio formats with different names can overlap up to 80% in terms of the songs played on them.
  • Individual Stations Use Highly Similar Playlists: Playlists for commonly owned stations in the same format can overlap up to 97%. For large companies, even the average pairwise overlap usually exceeds 50% [see Figure 3].
  • Network Ownership Is Also Concentrated: The three largest radio companies in terms of station ownership are also the three largest companies in terms of programming-network ownership.

What does that all mean? Basically, it means that the same songs get played over and over and over on the radio, in every major market in the US and a lot of smaller ones, as well. As far as I can tell from a quick search, Professor Liebowitz’ paper isn’t actually available anywhere online, so I can’t say for certain whether he controlled for the increased homogenization of what people were hearing on the radio, but I have my doubts.

Hence our correlation vs. causality question: after all, it’s a little unreasonable to expect radio broadcasts to inspire people to buy albums that aren’t broadcast on the radio…right?

Update: Vincent Clement thoughtfully provided a link to the study in the comments, I’ll take a look through the paper and encourage all interested parties to do the same. There’s nothing quite like primary source data, after all!