For the past few months I've been hard at work on my new book, The Piracy Crusade. Last week I finished up a chapter in which I review the many complex and intersecting economic factors that contributed to the music industry's "perfect bubble" in the 1990s and its "perfect storm" in the 2000s.
While I was looking at economic data, I stumbled on something bizarre: since 1973, the nominal broad dollar index (NBDI), an instrument used by the Federal Reserve to represent the dollar's strength relative to other currencies, has followed almost exactly the same curve as the figure for global music sales reported in inflation-adjusted dollars.
At first, I thought this may be a helpful explanatory variable (an early draft of the chapter included the above chart). After all, as the US dollar gets stronger, that means that other currencies get weaker. So someone in Europe or Asia would have to pay more, in terms of their local currency, for the same goods, which increases global music spend. But this doesn't really make sense, because even if Francois is paying 11 euro for a CD instead of 10 euro, it still nets out to the same $15 after conversion, and therefor doesn't move the needle on terms of reported revenues. Then there is the point that, if Francois pays 10 euro for a CD regardless of the conversion rate, this would actually mean that revenues go down as the dollar goes up -- which clearly isn't borne out by the data.
Here's the most confusing part: if the NBDI had some kind of effect on music sales, you would expect the former curve to precede, or map onto, the latter. Instead global music sales appear to lead the NBDI by a lag of about three years. It's not just the long climb and the long decline -- the bump in music sales at the end of the 70s is echoed by a similar bump in the NBDI in the mid-80s. Similarly, the bump in music sales in the mid-00s is followed by a bump in the NBDI in the late-00s. This would suggest that, if there's any causal relationship at all, global music sales appear to be influencing, or at least indicating, future trends in the value of the dollar abroad. But music is far too small a market to have such a broad effect, and I can't imagine a conceivable mechanism, so this doesn't make much sense either.
Earlier this morning, I spent half an hour going through these data with a senior researcher at the Federal Reserve (who asked me not to use his name in publication). Though he has a great deal of expertise with the NBDI, and has even published articles about it, he was just as flummoxed as I am. At the end of the call, we were no closer to resolving any of these issues. It's possible, of course, that the apparent correlation is purely coincidental, but the curves are so well fit, over such a long period of time, that a spurious correlation seems unlikely.
If you have any insights, I'd love to hear them...