I was looking through some MSA literature, and ran across a paper earlier this year by Negash and Shackley (2006) concerning long-distance movement of obsidian in Ethiopia. Here's the conclusion:
Obsidian is an excellent raw material for stone tool production, and because of its source-specific chemistry, it is an ideal candidate for the tracing of its movement from the source to archaeological sites. Our main objective here was to chemically characterize the obsidian artefacts of Porc Epic, and we have determined that at least some of the artefacts were being transported from a distance as far as 250 linear kilometres away. This is in keeping with other previous investigations in East Africa that have demonstrated MSA movement of obsidian over similar distances from sources to archaeological sites (Merrick et al. 1994).
The citation to Merrick and colleagues (1994) references obsidian transport in Kenya and northern Tanzania.
The transport at the present site (Porc Epic) was dated to older than 77,000 years ago. The paper considers trade versus non-contact long-distance movement:
As indicated above, the source of Kone was occupied during the MSA times. Similarities between Kone and Porc Epic artefacts have been surmised on technological and stylistic grounds (Clark and Williamson 1984; Clark 1988). Clark and Williamson (1984) also argue that Porc Epic was a seasonal hunting camp-site, hypothesizing its inhabitants to have been moving from the Afar Rift following the migration of game. We further hypothesize here that such technological similarity, coupled with the artefacts of Porc Epic being derived from Kone, can be taken as an indication that at least part of the obsidian from Porc Epic was transported by the people of Kone, assuming that the sites were contemporaneous. Alternatively, the inhabitants of Porc Epic may have transported the obsidians. Another equally possible explanation is that the obsidians from Kone reached Porc Epic through direct or intermediate trade or contact. Unfortunately, we do not have radiometric dates for Kone, and such inferences have to await further investigation of MSA sites that may be located between Kone and Porc Epic.
Spatial distributions of artifacts make up a sort of second-order data source; it takes a fairly dense sampling of sites to make inferences about movement networks in this way. So a denser sampling of African MSA sites will yield more information about raw material movement. On the other hand, obsidian is a very high-value raw material, with a bonus of being highly noticeable in assemblages, so it may not be quite comparable to flint movement patterns in the European Middle Paleolithic, for example. Hmm....
Negash A, Shackley MS. 2006. Geochemical provenance of obsidian artefacts from the MSA site of Porc Epic, Ethiopia. Archaeometry 48:1-12. DOI link