Maybe the guys at Goldman know something I don't (that wouldn't surprise me) but, the increased rating on EMI doesn't make sense. Perhaps it's an after the fact coronation, since, "EMI's stock has soared more than 40 percent since the beginning of the year, bolstered by a report that discounted CD prices pushed UK album sales up 7.6 percent in 2003." If so, it's a sad exhibition of irrational exuberance, since the reasoning behind the 40% stock price increase seems faulty.

Okay, so record companies have perhaps taken off their blinders and recognize lower pricing can increase CD sales. Of course, they'll still complain about the downloading menace, but maybe they'll also fess up that most people and even some downloaders like buying music, they just don't want to spend $20 on CDs. Surely, part of that 7.6% increase is probably pent up demand by purchasers previously discouraged by the $20 price. But the biggest reason this is just a one-time sales bump is that label cost structures prevent any future price reductions. Since it is doubtful EMI will be able to use the same trick next year, I don't think a 40% stock price increase is justified.

This year's positive numbers do not derive from a deep systemic improvement in the industry's economy, but there is a major shift afoot that is not good for business as usual. Let's introduce two other key facts. First, EMI enjoys consistent popularity for it's $15 "Now That's What I Call Music" series of hit compilations. Second, unit sales of digital downloads have exceeded all expectations. The simple explanation for both is that consumers want to pay $1 per song only for the songs they want. That's the straight deal with digital downloads and the implicit deal for purchasers of the Now compilations ($1 per song multiplied by 15 songs).

This $1 per song premise also explains the increase in full length CD sales. For most people, downloading isn't an option (no iPod) and the Now compilations are a bad fit (they don't like all the songs on them). So the option they are left with is buying a 15 song CD with only one song they definitely like (the one released to radio). Let's say they bought a $20 CD and learned they only liked 9 of the other songs. From their perspective, they just paid $2 per song. In fact it's impossible to get down to the $1 per song levelwith $20 CDs, so it's no wonder fewwer and fewer people risked the money. If CDs are priced at $10, however, the chance of a satisfied customer is much higher. If they like 9 other songs, they're acheived the $1 per song nirvana. Anything on top of that is gravy, so I'm not surprised consumers were 7.6% more likely to take the risk.

So what's the problem? Record labels have grown fat on the premise that people will pay $20 for a CD that contains only one song they like. Even now, they only need to guarantee that the one hit song released to radio is good enough to convince people the $10 CD is a good risk, therefore almost all of their production and marketing investment goes into that one song and they essentially bank on getting paid $10 per unit. More and more of the other songs on the album are filler. But, in a world where all consumers can buy just the one song they like, labels will suddenly only get $1 per unit - a 10x drop in revenue. It gets worse when the price point per song drops below $1. That's the source of the implosion.

Now the implosion isn't going to happen overnight, so the industry has some time to adjust, but the changes need to be structural. Albums, if they continue to exist, need to be listen-through experiences: all the tracks need to be worthy and the sum of them needs to be greater than the parts. If an artist has only one song worth selling, they'll need to do so with one tenth the marketing and production they get now or share a large budget with 10 other artists. Finally, labels need to find intelligent ways to increase their hit rate and diversely invest in more bands so they can capture the nine dollars they are missing per mega-hit by cross-selling similar acts. The tough part is: these sorts of adjustments can't happen overnight, and the labels are in a race with their own demise.