Testing the RIAA logic is hard without data

by Matthew Gast

Related link: http://www.blogmaverick.com/entry/1234000523038163/

Mark Cuban writes about a panel at the Consumer Electronics Association show in which the head of the RIAA said it was obvious, according to their own "data," that rampant piracy is the death of the music industry. My economics training has always made me suspicious of that contention. At the end of 2002, I decided to try my own test, but I had to give up for lack of data.

Many years ago, I studied econometrics. When you have a question about economic behavior, you can collect data and run a statistical analysis. It's not definitive proof, but it's as close as you can get in the social sciences. Even if it's not guaranteed to be as accurate a result as in engineering, it's still quite useful.

The objective is to correlate music sales against variables which you might feel are relevant. You're trying to explain CD sales by changes in income (that pesky economic slowdown that started in 2000), price (I buy less of more expensive stuff, as do most people), and yes, the presence of file sharing.

As it turns out, I never completed the project, due to lack of data.

Data on prices and income is collected by the government's bureau of economic statistics, and is easy to come by. No problem there. The stumbling block was that there is no useful data on music sales. The RIAA collects market data, but it is less than useless. They report what they ship to stores and its value at retail prices. It tells you very little about how much music people are actually buying, and even less about what they're paying. Just because the list price is $17 doesn't mean that people are actually forking over that amount. (Just ask the person General Motors who's responsible for rebates on cars. The list price of a Chevy Cavalier keeps rising, but the price you pay is falling because the rebates are getting bigger, too.)

As it turns out, there's a company called SoundScan that collects data from the retail industry at the cash register. How many CDs did people actually buy, and what did they actually pay. Unfortunately, SoundScan data is very expensive, and they do not release anything for research purposes. I tried going into the microfiche archives of Billboard magazine at the public library to compile my own data series on sales at the cash register, but the data wasn't reported reliably in Billboard, and I have yet to find a librarian who could help me track it down. Regrettably, I gave up on the project.

(By the way, the techniques that the RIAA uses to fiddle with the data were written up after I gave up on getting the data -- see this story.)

In the course of trying to assemble the SoundScan data from music industry trade magazines, I found an article on the front page of the December 28, 2002 issue of Billboard, which said that "The music business went from bad to worse in 2002…the past 12 months have played like a repeat of 2001--during which sales dropped 2.8%--only with more severe declines." According to the RIAA data, the total value at list price of music shipments dropped 4%, and the number of unit shipments dropped 10%. Either way you slice it, the SoundScan data shows a much gentler decline than the RIAA would have you believe. It also seems like quite a reasonable decline, given that median household income also dropped by 2% that year. (I would have expected music to act more like a luxury than it does.)

So, the real question for me is: what does the SoundScan data show? Would it substantiate the contention that rampant piracy threatens the livelihood of the music industry? Or does it just show how the RIAA fiddles their data? My gut feel is that a combination of price increases and a general economic slowdown explains the drop in music sales. The SoundScan data is the best way to settle this argument, but it's not available. Coincidence or not?


2005-03-30 21:42:54
Calling Mark Cuban...
If the SoundScan data is so expensive, then perhaps someone with beaucoup bucks could spring for it.

-- Alex (Mavs Fan 4 Life!)

2005-03-31 16:29:43
Calling Mark Cuban...
Except that we've probably alerted them to our intended purpose now. Plus, I bet they have confidentiality agreements to boot. (Or would just for us...)

I don't know how expensive it would be. The main purpose of the SoundScan data is not to track the size of the market for music. The real purpose is for a record label to know how their records are doing in each market, relative to the competition, on a week-by-week basis.