Steve,
I'm in continual contact with local RE agents; I note when properties go on the market, and about how long they've been on. As you might expect, some sell before the sign is in the yard, and others languish for years. Many of the local agents agree with the 6-12 month average estimate. The lazy/shady/dumb ones try to tell me, "Oh, no! I can sell a house in 90 days. Why, just the other day, I ...." These are the ones that have no understanding of "average" stuff. They wouldn't know a mean or a median from their butt in Chapter 7. I estimate inventory from website listings of the various agents, bounced off of Realtor.com (because not all local agents are Realtors). Number of newspaper ads can give you an index into the total number of listings; they don't advertise all of them.
See, I've already got a lot of imprecise data. I see no need to pay for MORE imprecise data. Used to subscribe to MLS for interior stuff, until I kept coming up with sales in public records, and no listing on MLS. That, coupled with lousy DOM figures, caused me to let the MLS subscription go.
Exposure time is historical; it assumes that your value considered that the subject had been exposed to the market for the average exposure time. Market time is forward-looking. It is usually the same as your opinion of exposure time. But suppose you're doing a report on a subject near a plant that has announced its closure in four months. My market time estimate very likely would have to be extended to 12-18 months because of the anticipated rise in supply coupled with reduction in demand as people move to new employment. So I'd have exposure time of 6-12 months, and a market time of 12-18 months. The basis for that estimate would be the exposure time estimate, increased by a factor representing the increase in supply, along with a decrease in demand. Couple that with my own estimate of how long such closures have historically taken to clear. 'Nother argument for geographic competence.
