Whether print-on-demand versions of ETDs should be coded as “Type: a” or “Type: t” can be argued either way with both perspectives based on a logical application of existing cataloging instructions and guidelines, all invented without this particular situation in mind. However, I recommend use of “Type: t” coming down on the side of some practical considerations when dealing with materials in this situation.
Apply the “Print on Demand (POD) Reproductions and Photocopies” portion of LC-PCC PS for RDA 1.11, so that you would end up with a record describing the original publication with the addition of “Form: r” and field 533 containing only the text “Print reproduction” in subfield $a. In order words, very much the way we have always described photocopy reproductions under AACR2 except that the resulting reproduction note is now generic in order to stand for the print reproduction as available from any provider. In this case, Proquest, is treated as just another print-on-demand provider to be omitted from field 533. The resulting record is also marked with 040 $e pn.
That would ordinarily mean that an ETD from 2015 would end up as “Type: a” because it existed online, which we consider published by default. It would also mean the 300 $a would say “1 online resource” which is objectionable given that the item in hand is hardcopy and the 533 will contain no physical description. But, if the thesis was dated 2005, did it only exist online? If dated 2000, was it submitted by the student as an unpublished hardcopy manuscript and subsequently digitized? If dated 1975, well, it must have been submitted in hardcopy, subsequently microfilmed by UMI/Proquest, and perhaps made available digitally later; do the interim formats and their publication statuses really matter for the 1975 thesis?
The problem is that it may be easy to tell in some cases whether a thesis only existed online and not so easy in other cases. The traditional thesis title pages used as sources in cataloging are not often going to provide any clues either. Whether a thesis previously existed online only, existed both online and in an unpublished hardcopy version, etc., should not cause catalogers to do a lot of time-consuming research for distinctions that are not all that useful to the end user of the catalog.
You have a hardcopy thesis in hand available on demand. It’s no different than ordering that same thing from UMI back in 1980. Treat it as a reproduction of an original manuscript ignoring the fact that it had been microfilmed and digitized or nowadays was submitted as a digital original because it’s too hard to tell, and when you can tell, it only results in records arbitrarily coded “Type: a” in some cases while coded “Type: t” in others, i.e., inconsistent treatment of what looks like the very same thing.
Based on the situation, this would be recommended:
Date: <the date of the original thesis>
040 $e pn
264 _0 $c <date of the original thesis which is core element in RDA, but not treating it as published which eliminates the need to indicate that the author or the author’s institution or Proquest is the publisher in 264 _1; no use of $a or $b>
300 <the description of the item in terms of pages, not 1 online resource>
533 Print reproduction. <without further details in accord with the LC-PCC policy statement>