Last year I did a bit of test on a niche where its very hard to find expired or auction domains for it - Its a sub nice of adult.
SET ONE: I set up 50x new domains - 35 of them where brought from domain sellers, but they where just never used domains, as opposed to brand new domains or expired ones. Most of them had been indexed before. The other 15 where brand new registered over a 6 week period. All domains where of course on topic. TOTAL COST TO ME: £800
What I did with SET ONE was I acquired manually, many links to them from nice relevant websites, directories etc. The websites on them, where very very money site type sites to be able to pass the manual activations to acquire the listings I need to produce good set of back links to them. After I did the first five manually, I built a bot to build the other 45 out for me to get the listings etc I needed. I then went back and acquired a few more manually from a secondary list, to ensure not all the domains had the same backlink profile of course.
The purpose of this was to build 50 really on topic, totally relevant domains for the PBN links to come out of......
SET TWO: I used 50x PBN domains not at all related to the niche, but could be smartly branded to get some relevance to it. These all had a an average of 60 RD (40 minimum) with TF of 20+ on them. Content was of course wrote to manipulate the domain to be as on relevance as possible. TOTAL COST TO ME: £4000
So identical money sites where built to receive the links. Very similar domain type, same CMS. Same onsite layout etc/ Content was unique for each site of course, but kw's where the same, placement, density, H1. H2 tag usage etc etc all identical.....
The links where placed with identical velocity, same plan of anchor text distribution etc etc. The test was a 1-1 with only difference being the domain types.
HERE WAS THE OUTCOME 6 MONTHS AFTER FIRST LINK PLACED & 2 MONTHS AFTER LAST ONE PLACED
Money site from set 1: Page 1 result 2 for main keyword (Home page), Page 1 result 1 for secondary keyword (Home Page), 70 different inner pages with various more long tail kw's - 48 of them page 1 top 5, 14 page 1 result 6 -10. The rest where page 2.
Money site from set 2: Page 2 result 3 for main keyword (Home page), Page 1 result 8 for secondary keyword (Home Page), 70 different inner pages same sets kws as above - 12 of them on page 1 results 4 to 8, 36 of them page 2 and the rest between page 3 and 4.
My answer to your questions is > I do not think building new pbn domains on a tighter budget is a real problem if you can spare the time to do it. And if you want to increase on topic flow for sure. So yes i think it is possible to do it for sure and does work well for me for sure in these niches. Since this I expanded the network in to 5 nodes of 200 sites per node for various clients sites and its working very very well for me really.