This is an interesting analysis of data in the publisher's statements. Kudos. And thanks for remembering my identification, months ago, of Marvel comics with newsstand-style barcodes, but without Curtis markings.
You floated an idea that this alternative distributor may have served bookstores, and not newsstands per se. I don't think I have said so publicly, but I have been operating on that assumption for some weeks now. The reason I haven't said so publicly is that I have not yet discovered any evidence supporting that assumption, except for the basic points that comics were known to have been distributed to bookstores, and that comics in bookstores were not marked as direct-sales copies; what I need now is some evidence (even if merely a collector's recollection) that titles in this middle tier of newsstand comics (e.g., Power Pack, Doctor Strange Sorcerer Supreme, The 'Nam) were seen in bookstores during the appropriate timeframe (late '80s–early 90's).
I'm not persuaded that all issues after #39 are direct editions. There are just too many examples of direct-only comics published between 1980 (publication of Superboy Spectacular #1, the first direct-only comic book) and 1993 (when direct-edition Marvels and DCs began bearing barcodes) without any of the usual newsstand trappings (barcode, Curtis symbol, color banding to indicate off-sale week) for me to be believe that Marvel was barcoding direct-only comics before 1993. I still like the notion of a third distribution channel.
Having said that, it does seem clear, with titles distributed via this third channel, that Marvel was not as meticulous about publishing two versions (barcoded and Spidey-head) of each issue as they were with Curtis-distributed titles. I think I pointed out earlier that I collected Doctor Strange Sorcerer Supreme as it was published, and bought each issue at a comics shop. In spite of this, many issues in my collection do have newsstand-style barcodes, and for those issues, I have never found a cover scan on the Internet of a Spidey-head variant. Perhaps the third-channel distributor was less concerned than Curtis was about comics shops returning comics through non-direct channels. Perhaps, with sales booming, comics-shop owners no longer needed the financial edge that returning direct-sales channel comics might have provided.
The facts you have cited regarding news agent returns are puzzling. I have never been able to figure out which issue those "single-issue" sales and return numbers pertain to. Over the years, though, I have seen a number of items stating that final sales numbers of newsstand-distributed comics did not come in until months after a particular issue went on sale. How many months? I don't remember ever seeing a specific number.
I do know, however, that Publisher's Statements are routinely filed on Oct. 1. The statement in PP #37, dated May 1988 (on sale probably Jan. 1988 in comics shops), would have been filed Oct. 1, 1987 (the published statement itself ought to be dated; if I am wrong, please let me know). That still puts filing of the statement more than one year after cover date of the last issue (#25) before Power Pack went direct-only. That's more months than I would have guessed necessary to get final figures, but let's assume (for this discussion) that it took 15 months after cover date for the newsstand distributor to provide final sales figures to publishers. In that case, the Statement might have covered #s 13–24 for the 12-month average figures, and #24 for the single-issue figures.
Power Pack #46, dated May 1989, would have carried the Publisher's Statement dated Oct. 1, 1988. Using the same assumption as in the preceding paragraph, this statement would have covered #s 25–30 (Power Pack went bi-monthly about the time it went direct-only) for the 12-month average figures, and #30 for the single-issue figures. It wouldn't have been unreasonable for Marvel to include only #25 in the news agent sales and returns (averaging in issues not distributed via news agents would have skewed the figures), so the 12-month averages might have been for #25, and the single-issue figures might have been for #30.
Power Pack #56, dated June 1990, would have carried the Statement dated Oct. 1, 1989. Using the same assumptions, this statement would have covered #s 31–38 (reflecting an every-six-weeks publication frequency). Every issue in this time period was published direct-only; zero news agent sales and zero news agent returns are therefore expected and reasonable.
Admittedly, all this depends on a rather extreme assumption of the time periods covered by Publisher's Statements, and if that is proven incorrect, my tentative conclusions here would have no merit. Personally, I prefer to take this as evidence that the single issues referred to in Marvel's Publisher's Statements were much older than previously suspected.