Monday, April 15, 2002

Or, Lies, Damned Lies, and Statistics, Part II, wherein Dr. Bonzo shows that the RIAA are a lot of sniveling brats.

First, a disclaimer: IANAENAIASAIHNPOOTV.  (Dr. Bonzo is neither an economist nor a statistician, nor has he ever played either on TV.)  Nevertheless, our faithful reader may find the following to be instructive.

Now, a pair of pictures:

We observe two things:

  1. In the first graphic, the inflation-adjusted average list price of a compact disc declines steadily from 1991 through 1996, and then levels off.  1995-96 is the same time frame in which the a bevy of States' Attorneys General assert that the major record labels colluded together in their price-fixing scheme.
  2. In the second graphic, the average annual number of compact discs purchased per buying "unit" (see definition below) rises very nearly linearly with declining unit cost.  If Dr. Bonzo were an economist, he would know the right microeconomic jargon for this familiar behavior: we buy more of that which is less expensive.  The only deviations from this nearly perfect linear relationship are in the years 1996, 1997, and 2000, and the last is barely significant.  The really noticeable deviations are '96 and '97 -- the years following the inception of the Majors' price-fixing collusion.

Compared to the 1996-97 decline in CD sales, which appears to be strongly correlated with the end of a steady real-dollar decline in CD prices (both effects appearing also to be associated in time with the labels' alleged price-fixing), the relatively small decline in sales in the 2000-01 period, far from presenting dramatic evidence of the costs of "file sharing," appears to us to be a combination of a leveling-off in the number of family/individual "consumer units" and ongoing stability (read: lack of natural decline) in CD prices.

Sorry, we see now that the last paragraph was (as is sometimes our wont) a tad convoluted.  Let us translate:

Earth to RIAA: It's yer own damn fault, so quit yer bitchin'!

Useful definitions from the Bureau of Labor Statistics:

Consumer unit (Consumer Expenditures)
A consumer unit comprises either: (1) all members of a particular household who are related by blood, marriage, adoption, or other legal arrangements; (2) a person living alone or sharing a household with others or living as a roomer in a private home or lodging house or in permanent living quarters in a hotel or motel, but who is financially independent; or (3) two or more persons living together who pool their income to make joint expenditure decisions. Financial independence is determined by the three major expense categories: Housing, food, and other living expenses. To be considered financially independent, at least two of the three major expense categories have to be provided by the respondent.
Consumer Price Index
Consumer Price Index measures a price change for a constant market basket of goods and services from one period to the next within the same city (or in the Nation).

The numbers used above were computed thus:

Annual CD Sales per Consumer Unit: For each year from 1991 through 2000, the number of CDs shipped and sold (as reported by RIAA) was divided by the total number of Consumer Units reported for that year by the Bureau of Labor Statistics (series ID CXUCU000101).

Average CD List Price (1990 Dollars): For each year from 1991 through 2000, the average list price of a CD was computed by dividing the RIAA's reported Total Dollar Value by their reported CDs Shipped.  The resulting dollar amount was then adjusted for inflation by dividing by the ratio between the CPI for that year and the 1990 CPI.

9:48:38 PM    

Or, Lies, Damned Lies, and Statistics, Part I

The Shifted Librarian writes:

Glenn Fleishman pushes back on my post about price fixing and CD sales:

"You forgot to think about constant dollars here. Inflation was low but not non-existent in the 1990s. I found a constant dollar conversion spreadsheet at this site, and the factor from 1992 to 2001 is 1.3. Thus the average price in the chart of $13.00 in 1992 is $16.90 in 2001. The actual current dollar average price is $14.75 in the chart you cite.

Not that I like the RIAA, but let's calibrate before condemning. The $9.99 CD you cite would cost $13.00 today assuming constant costs."

Glenn makes a good point, although I think the original argument still stands that the price of CDs continues to go up when the technology behind it should be making it go down (at best) or stay constant (at worst). Price fixing didn't help that scenario, and now much of the customer base is alienated and dissatisfied with the product, creating a slowdown in sales. Mix all of this up with the current state of the radio industry that's supposed to be helping sell music, and you've got major problems that dwarf file sharing.

Jenny's original post, of course, springboards off of George Scriban's graphic observation of a (roughly) inverse relationship between total annual CD sales and average unit price per CD.  George also pointed out in the same post that a steady increase in per-unit cost (and a levelling-off of demand) began in 1995-96, a time point when (so concluded the FTC and 28 state attorneys general) the major labels colluded in a nice little price-fixing scheme.

What we these data suggest, of course, is that CD sales simply followed the most familiar of microeconomic rules: when price goes up, demand goes down.  The deeper implication is that the RIAA's claims that Napster (and other file-sharing technologies) was to blame for the drop-off in CD sales observed in 1999-2001 are questionable, since there appears to be an equally strong (negative) correllation between the industry's own price-fixing behavior and the sales decline.

As Glenn points out to Jenny, of course, it's not quite fair to correlate the RIAA's sales figures against current dollars, since there was (some) inflation during the '90's.  (On the other hand, it does appear that per-unit costs increased even when corrected for inflation.)  But the problem is more complex than that.  To be really thorough, we'd also have to take into account varying levels of consumer spending (reflecting consumer-confidence levels, variations in real income, and so forth as the economy burned and then tanked), as well as the simple fact that the total number of spenders (and thus spending) increases steadily with time by the sheer fact of population statistics.  (As Stephen J. Gould famously points out, correlation does not imply causation: if it did, then we could blame [say] the overall positive trend in CD sales on the mere fact of his [or my, or Hilary Rosen's] increasing age.)

Just to be a complete pain in the ass, we'll also quibble over the per-unit price figures used by George and Jenny, since (as far as we can tell) they're derived by dividing the RIAA's total-annual-unit-sales numbers into their total-dollar-value numbers, which (according to the RIAA) are calculated based on "suggested list price," rather than on actual retail price.  (Note that having the actual retail price figures available might actually make George's argument stronger, since -- in the pre-price-fixing regime -- the actual retail price would likely actually be lower than the suggested retail price (thanks to aggressive pricing at Target, BestBuy, etc.).

The nice thing about the Internet, of course, is that one can find almost any set of data that one pleases, sometimes even reliable data.  Without further ado, we present the following table of numbers:

Year A B C D E F G
1990 96968 $28,381 $1,422 $31,889 132.5 N/A N/A
1991 97918 $29,614 $1,472 $33,901 136.2 333.3 $4,337.7
1992 100019 $29,846 $1,500 $33,854 140.1 407.5 $5,326.5
1993 100049 $30,692 $1,626 $34,868 143.7 495.4 $6,511.4
1994 102210 $31,731 $1,567 $36,181 147.6 662.1 $8,464.5
1995 103123 $32,264 $1,612 $36,918 151.3 722.9 $9,377.4
1996 104212 $33,797 $1,834 $38,014 156.3 778.9 $9,934.7
1997 105576 $34,819 $1,813 $39,926 158.6 753.1 $9,915.1
1998 107182 $35,535 $1,746 $41,622 161.1 847.0 $11,416.0
1999 108465 $36,995 $1,891 $43,951 165.6 938.9 $12,816.3
2000 109367 $38,045 $1,863 $44,649 171.2 942.5 $13,214.5


A. Total number of Consumer Units, in thousands (BLS series ID CXUCU000101)

B. Average annual spending per Consumer Unit, current dollars (BLS series ID CXUTE000101)

C. Average spending on entertainment per Consumer Unit, current dollars (BLS series ID CXUEN000101)

D. Average money income before taxes per Consumer Unit, current dollars (BLS series ID CXUBI000101)

E. Consumer Price Index (December CPI-W index, seasonally adjusted, 100 = 1982-84 value, BLS series ID CWSR0000SA0)

F. Total number of CDs shipped, in millions, net after returns

G. Dollar value of CDs shipped, in millions of current dollars, net after returns


1.  B-E: U.S. Bureau of Labor Statistics , and

3.  CDs shipped and CD Dollar Value, RIAA Year 2000 Year-End Statistics,

(Updated to include annual income and entertainment spending figures.)

What we (or you) do with these numbers, and what we (or you) conclude, will have to wait: the business of the Inquisition demands our attention!

9:33:05 AM