In a provocative post, Donny's Blog calculates that if the RIAA got its way and was able to extract $150,000 per infringement, in one month it would accumulate more money than the annual GDP of France. In the end, some $11,441 billion. This seems a tad more than the $300 million per year the RIAA estimates it is losing because of downloads. To recover that, only 2000 prosecutions would need to be successful.
I enjoy these types of calculations, but it would be remiss of me not to point out the obvious: the fines for those prosecuted are much more than the cost imposed by them on the music industry. This is consistent with the economics of crime and penalties that tells us that to deter an activity when you can't catch everyone you need to set penalties high enough that people who gamble on not getting caught won't do it.
But this calculation does enable us to assess what the probability of being caught might be. Let's suppose that the RIAA targets 2,000 successful prosecutions per year. Using the numbers that got us the $11 trillion cost per month, there are possibly 1 billion downloaded songs per year. Thus, your probability of being caught is 2 a million. To back out the deterring penalty we take this and divide it through the cost of downloading a song legally (that is, $1). That gives us $500,000. [I guess if we could adjust for US versus other downloads it may be less].
So according to this, the RIAA's penalties are too low and not too high as the blog was suggesting. Perhaps that is why downloading still occurs.