It's an oft-made argument, at least by lobbyists for a certain industry, that piracy hurts the economy. From an economic perspective, the argument that not spending money on any item hurts the economy is a little weak, given the offsetting effects of personal savings. But the issue is thrown into particular relief by a quote from this study I linked to earlier on software piracy: ' "Teenagers are being tactical spenders," said Dr Bryce. "The money saved lets them spend more on mobile phones, going to the cinema or eating out." ' So if there is a fixed pool of capital among teens, the prime infringers, and that capital is spent on real property with high global opportunity cost (e.g. others are denied the good) instead of on IP (with low global opp. cost), is there a net gain to society?
Simplistic model- say a teenager has a $50 entertainment budget. Games (or the IP item of your choice) are $25 apiece and provide utility according to the function Ug=25-5x (so first item provides 20 utils, second provides 15, etc.). Eating out (avoiding cinema for clarity's sake because that's also IP with low marginal cost) costs $10 with a utility function Ue=50-10x (not sure about the assumption inherent here--that marginal utility is higher for the first meal than for games, but declines more rapidly as you run out of hunger to satiate and friends to dine with--because MU declines pretty darn quickly for games since each takes up so much limited leisure time).
So, looking at some budgetary possibilities:
0g, 5e: 100 utils
1g, 2.5e: 100 utils [not a great example because of the fractions]
2g, 0e: 35 utils
Not a great model at all, but it's good enough for 2am.
So focusing on the first two possibilities, moving from two to one with the additional purchase of a (we're going to assume free) pirated game. Now we have 120 utils within the same budget. So clearly if society were just made of consumers, the picture would be easy.
So we come to the reason I started writing this in the first place. Which is better for producers? The question can basically be boiled down to, "Which item deserves more consumption?" since the marginal cost of the game is essentially zero. Another way of phrasing the question that shifts the whole debate is, "How distortionary is the pirated purchase?" since the method of purchase means that market cues for value cannot be used. This distortionary effect clearly must have a cost, and I'm sure someone's written a paper on it at some point.
The final factor, closely linked with the previous one, is the effect on incentive to invest. This is why the distortionary effect matters.

Anyway, don't think there's anything truly new here, but I'm going to wiki it and keep pondering. I'm way past my level of econ here, so it's posted and if mysterious angels happen to come play and fix all the little gaps and come up with a nice model, that would be great, but I'm going to go to sleep now, which is the proper time and place for dreams....

--Ari