I'm very suspicious of these conclusions based on the dataset. Equating number of patents with "productivity" is questionable and leads to a lot of confounders. For example, most patents are filed by large companies like IBM, which filed 2,000 patents on cloud computing in 2018 alone.

Large companies like IBM value patents more than startups and have dedicated teams that help employees make tons of filings. I would bet this is part of why NY-NJ-PA, where IBM is headquartered, does better than you'd expect. And, Boston, which has no very large patent filers, does worse than you'd expect.

The 12% gain in "productivity" from moving maybe more to do with inventors joining companies that encourage them to file more patents than any real increase in their output.