I've been an industry analyst for the stock photo sector since 1999. I came into the industry because, as a photographers, I was curious about the same question as the original poster: what is the real size of the non-traditional photo buying market?  Since then, I've been doing research, aggregating data, charting trends of related industries that use imagery, and so on. (I have a prior history of economics and business/hi-tech development.)

If you truly want to dive into this subject, my blog has tons of content for you to churn through: http://www.danheller.com/blogs

There are many articles related to this question, but also related to microstock's role in this industry as well. (It's not what you think.)  The various titles are self-descriptive.

Though it would be futile to try to address such a huge topic here, I can summarize it this way:  the total size of the "stock photo licensing market" (where the definition of a "stock sale" is "any financial transaction that involves purchasing a license for using a pre-existing image"--as opposed to paying someone to shoot a photo) is about $20-25B.  The size of the "traditional stock license" market involving stock agencies and "traditional buyers" (ad agencies, news outlets, media companies, corporate) is about $2-3B. (most of that is Getty, followed in sequence by the smaller agencies.)

Approximately 15% of total stock licensing revenue comes from pro photographers. 85% of the rest come from semi-pro's, newbies and amateurs. The distribution of those sales is almost exactly the same as that of any business engaged in internet-based "long tail economics." That is, the amazon.com model, where the "big winners" (such as best-selling books) actually represent a tiny proportion of overall aggregate sales of ones and twos that no one pays attention to.

As for photo sales, there's a pretty consistent distribution across the Us population of people who've "sold an image at some point in the past 5 years for any amount of money." When you aggregate 300 million people, this quite quickly totals large amounts of money. When you start factoring in those incrementally smaller sub-groups that sell increasingly larger and larger revenues--such as $5K/year--then the revenue map starts making more sense.

The reason why most pro and stock photographers don't see this is because long-tail economics is still a new paradigm most people don't understand. ("If you don't see it, it must not exist.")  But this answer is not entirely complete--for photographers, there's a cultural bias rooted in a complicated history that makes their true role in the larger market more difficult to accept.

A longer discussion on that can be found here: http://www.danheller.com/truisms

dan