“Saving Soviet Science: The Impact of Grants when R&D Funding Disappears,” I. Ganguli (2010)

Here’s another interesting job market talk: Ina was at Kellogg presenting this paper today. The underlying question concerns whether R&D spending, particularly by governments, is worthwhile for society, or whether people will produce science anyway. There is some earlier evidence that increasing the amount of science grants just increases the salaries of scientists: they’ve invested heavily in a skill that is essentially not transferable to the private sector, so whether they’re paid $50000 or $100000, they’re still going to produce science. But surely at some margin, research scientists will just go become potato salesmen. If we believe that research has large positive externalities, this is particularly harmful.

Ganguli examines a quasi-natural experiment in the Soviet Union. When the USSR dissolved, science funding fell a massive amount. At that time, the Eastern Bloc produced a similar amount of scientific output as the US. In 1993, Soros stepped in and offered scientists who had published 3 times in the last 5 years in good journals a payment of $500 to stay in research another year. This sounds small, but it was similar to a year’s salary for many of these guys. Was such a payment effective in the long run?

There is a nice regression discontinuity here. Some guys had three papers in the past six years, so they did not qualify, and others had three in the past five years, qualifying them. Objectively, these guys are of similar quality – don’t worry, age-career effects are included. At this margin, future publications roughly double for the treated group. Backing out the implication of this is that every $100 Soros spent bought a future scientific publication. Also, among Moscow-based scientists, there was less emigration among the treated group; I don’t really understand what could possibly lead to this result, since presumably it was easier for Moscow-based scientists to leave if they had the money, so perhaps this is just an anomaly in the data.

An interesting experiment, then. The economic content here is not totally clear: the incentives offered are backward looking, so I guess you need some kind of story about how leaving science in 1993 to sell potatoes because you don’t get the grant somehow keeps you from reentering science later, or else degrades your skills. But since science funding was very low for two years before the Soros grant, I’m not sure this story makes sense. In any case, the production of science is a very poorly understood topic – yet a very important one! – so any contribution is a welcome one.

http://www.people.fas.harvard.edu/~iganguli/papers/ganguli_jmp_nov15.pdf (WP)


2 thoughts on ““Saving Soviet Science: The Impact of Grants when R&D Funding Disappears,” I. Ganguli (2010)

  1. Manoel Galdino says:

    Though I had comment here in the past, that’s the first time when I think I understand better the results of a paper than you!
    Maybe because I study in Brazil and we had here conditions more similar to Russia that time.

    Anyaway, my two cents are:

    1. emigration: Emigration is hard because you have to be far from your family, friends, history… In a word, you have to change your identity. So, if you receive money and can stay at home, you stay at home. You will only emigrate if really needed to survive.

    2. About the economic content. Science is (I think) something that demands continous effort. If you have to stop doin science for a year and then come back later you forget a lot of things. If you have to worry about paying the bill at home and not about that scientifc puzzle your out put will decline.


  2. walt says:

    1. I agree: emigration is unlikely without good prospects for employment.

    2. I thought US economists and other academics learned decades ago that if the rewards (e.g., getting past the application stage, tenure) were determined by paper counting, you divide your work into smaller pieces. Of course, specifying the journals helps here.

Comments are closed.

%d bloggers like this: