The music industry likes to complain about sales lost to piracy, but figures that show huge sales declines only tell part of the story. Before we blame this trend on infringement, we have to make several assumptions, including that the demand for music (whether purchased or pirated) has remained steady.
Figures available from the US Census bureau suggest otherwise. Data on “Media Usage and Consumer Spending” abstracted from a study by Veronis Suhler Stevenson show the average number of hours spent listening to music by US residents age 12 and older has declined steadily since 1998 (from 283 to a projected 219 in 2003, a 21% decline). Meanwhile, home video, video games, and consumer Internet have seen dramatic gains. This suggests that people are turning to new forms of entertainment (i.e., the Internet, video games, and DVDs) at the expense of recorded music.
Here’s the data, extracted from the Census Bureau report, on the number of hours Americans spent using various types of media in 1998 and 2003.
| Activity | Hours, 1998 | Hours, 2003 (proj.) | Change (hours) |
| TV | 1551 | 1656 | +105 |
| Radio | 936 | 1014 | +78 |
| Box office | 13 | 13 | 0 |
| Home video | 36 | 96 | +60 |
| Interactive TV | 0 | 3 | +3 |
| Recorded music | 283 | 219 | -64 |
| Video games | 43 | 90 | +47 |
| Consumer Internet | 54 | 174 | +120 |
| Daily newspapers | 185 | 173 | -12 |
| Consumer books | 120 | 106 | -14 |
| Consumer magazines | 125 | 116 | -9 |
| Total | 3347 | 3661 | +314 |
(Source: US Census Bureau, Statistical Abstract of the United States: 2003, p. 720.)
(Note 1: We chose to use 2003 as the ending point, even though the source includes projected 2004 data, on the assumption that the 2003 Statistical Abstract’s projected data would be more trustworthy for 2003 than for 2004. Using 2004 as the endpoint would not materially affect the analysis.)
(Note 2: It is possible that part of the decline in recorded music hours may be an artifact of the study methodology. The table caption states that the data for categories including recorded music were based on “survey research and consumer purchase data”. To the extent that the estimate of music listening hours is based on survey data, it can serve as a possible cause of the drop in music sales. But to the extent that the listening time estimate might be inferred from the drop in sales, it should not be used to explain the sale drop. More methodological details might be available in the VSS report, but that is not available to the public.
However, we think it is unlikely that the listening time estimate is derived entirely from sales data. According to the same Census Bureau report (which cites as its source the same Veronis Suhler Stevenson report), per-capita spending on recorded music fell by only 4% from 1998 to 2003; the RIAA estimated a 15% drop in its total recorded music revenue over the same period. It seems unlikely that a 21% drop in listening time would be inferred entirely from a 4% or 15% spending drop.)
(Note 3: VSS wants $2000 for a copy of their report. We’re not in a position to pay that much. If anybody has a copy of the report and is able to fill us in about their methodology, we’d be grateful.)
[This entry was written by Alex Halderman and Ed Felten. If you cite this, please don’t attribute authorship to Ed alone.]
Leave a Reply