The end of the hit (inflation corrected!)

Australian actuary Dermot Baslan kindly corrected my The End of the Hit data for inflation (actuaries are great at this stuff). Then, while I was revisiting the data, I went back and re-ran my queries on the RIAA database to eliminate singles, music DVDs and other non-album fare. Finally, I updated the figures to get the final 2005 numbers.

The revised numbers, with a nice six-degree polynomial regression line to highlight the trend in the lower chart, look like this:

The point? When you express hits as a ratio of total industry sales and then correct for inflation, it's not quite as dramatic--total music sales have been falling along with the hits. But it still ain't pretty. At all.

The updated spreadsheet is here.

Comments

[Tom]

Would it look better if you showed a five degree polynomial which didn't have to end on a massive downward gradient not supported by the data points?

[Will Page]

Chris,

Some loose thoughts...

Can we get more information on how the inflation was stripped out of the data please? I've looked at this for a while and what strikes me is that you've had deflation (at the retail end) of music sales for the past few years, with the price of a new 'hit' CD dropping from a peak of £13 in the late 90's to only £7 today.

This made me think again about how to use delators on the data as its not a consistent trend - price levels are moving up as well as down. Then I thought about what it is you might actually want to deflate - the retail price value of the CD (now falling), the wholesale cost of producing the CD (...who knows whats happening there?) or - better still - the profit margin of the music industry (...okay).

With no clear trend in the direction of price, perhaps it would be better to 'Follow the Money' - that is the only money that matters, profits. If it was possible to strip inflation out of time series data on profits of hits (or majors), then this might be a good 'real terms' illustration of Long Tail effects over time.

Will Page

PS: The very best of luck with the book, cant wait - I've got my advance order in!

[csven]

Right on track.

[Walter Lounsbery]

I violently disagree with fitting variational data like this with high degree polynomials. The data cries out for smoothing, or at worst one of those arbitrary moving averages (which introduces smoothing and lag). What if the latest data is off a bit? Screws up the interpretation. And how did we manage to get curves that appear to end in 2007? Not buying this conclusion, but I really like the blog...