Thread by @swagitda_

TIL F500 firms won 41% of R&D awards in 1971, but only 6% in 2006. In general, despite increases in science, we aren’t seeing increases in productivity.

This paper explores why, & I’m gonna highlight some interesting things from it in this thread: nber.org/chapters/c1425…

1/14

First, if you wanted a history of corporate labs, this paper is your nerding-out haven.

Turns out antitrust policy mid-century discouraged M&A & led corporations to focus on internal R&D for innovation (vs acquiring it). Maybe renewed antitrust focus will bring them back?

2/14

Another key aspect of the science =/= productivity shift is that universities focus on research, while corporations focus on dev.

This specialization makes it harder to “translate” research findings into practical solutions.

3/14

This chart highlights how corporate research didn’t necessarily disappear, but it did become less public.

Scientific capability no longer improved company valuations, but patenting did, so it was a natural result. And this doesn’t just apply to old companies...

4/14

The FAANGs publish very few research papers relative to their sales (Microsoft does a bit better) 👎

Overall, 73.0% of firms produced fewer publications in 2010 than in 2000. The avg # of pubs per $1mm of R&D fell from 0.726 in 1980-1985 to 0.369 in 2010-2015

5/14

This isn’t true for pharma/biotech, however! IT/software is special in its extreme decrease in its publications relative to patents.

Why? Probably cuz pharma patents are more protected + require proof of efficacy to be approved by regulators + receive more federal funds

6/14

The startup factor is also relevant. Smaller firms are better at generating inventions now, but larger firms can exploit them better.

“Larger firms therefore invest in scientific capability not so much to generate knowledge as to be effective buyers of knowledge”

7/14

Another factor: the interval between invention & commercialization narrowed over time, making it harder to profit from in-house research.

Companies will reduce research when knowledge spills out, because the ROI from research is less.

8/14

Twitter 🖤s blaming Wall Street, so bask in the fact that investors in the 80s pushed large public companies to divest unrelated units & focus on specialization.

But, diversified firms are the best ones to “exploit the unpredictable outcomes of scientific research.” Oops!

9/14

One of the key hypotheses that struck me here is that we largely ignored the social good of large corp R&D. It wasn’t privately profitable anymore, but its decline was strongly correlated with the decline in productivity.

Universities & startups don’t seem to be enough.

10/14

Why aren’t startups & unis enough to increase productivity? B/c large corporations:

1) Focus on systemic & architectural innovations vs autonomous ones
2) Solve specific practical problems
3) Aren’t scrambling for survival
4) Are multi-disciplinary
5) Create spin-offs

11/14

Funnily enough, it’s precisely the messiness of large corporations that leads to the necessary diversity & innovation to provide widespread social benefits.

“Attempts to centralize & direct innovation activity may backfire,” but startups *have* to do that or they’ll die

12/14

They note the increase in in-house machine learning experts & data scientists. But, this won’t lead to long-term research, just solving quick optimization wins.

The test is whether notably new markets are created from these efforts & if outside inventions still reign

13/14

TL;DR for science to increase productivity, we need:
✨ mission-oriented research
✨ systemic innovation
✨ multi-disciplinary focus
✨ expendable resources
✨ antitrust?

Corp labs aren’t coming back, so the Q is: what do we do to translate insights into productivity now?

Fin