Fabian Waldinger pursued an interesting research agenda so far. Let me explain what’s so fascinating about his work. But in order to do so, I first need to describe why (good) empirical research is actually such a hard task.
A university, or academia in general, is a complex organization. This complexity poses a challenge for empirical work on research productivity. On top of measurement problems, the question of causality arises. Let’s say you observe a positive relationship between teaching and research (apparently an old debate). But how can you know which drives which? Or whether another factor like a professor’s intrinsic motivation, which you can hardly measure, is the cause of the relationship. A simple correlation won’t tell you. And in a system like academia there are plenty of factors, which could secretly be at play.
To avoid these issues you can exploit what econometricians call a quasi-experiment. An intervention to the complex system under study, that is independent of all these other factors. Much like a real experiment, with the exception that you usually have no control over the quasi-experiment and can only hope to analyze something insightful after it happened.
Germany’s history in the 20th century offers two (unfortunate) points of departure. With arduous manual effort, Waldinger collected data on the dismissal of mostly Jewish scientists in German universities after the Nazi’s seizure of power (“Gesetz zur Wiederherstellung des Berufsbeamtentums”). These dismissals were not based on research productivity but solely driven by antisemitic and political motivations. In fact, as the author shows, the dismissed scientists were on average more productive than their peers. For example, eleven Nobel laureates are included in the list of dismissed scientists. Albert Einstein is only one of them.
On the basis of this unique data set, Waldinger has already addressed several empirical questions in economics of science. In his paper “Bombs, Brains, and Science” (Waldinger, 2013) he adds other interesting data on the impact of Allied bombing during World War II. He wants to know to which degree bombing attacks until 1945 destroyed a university’s infrastructure (buildings, labs, etc.). At this point, the huge effort of data collection becomes apparent. A substantial amount of information originates from face-to-face interviews with university archivists.
In the same way as the Nazi’s racist policies for human capital, the bombing attacks serve as an exogenous shock to physical capital. Ultimately, it allows to determine which of the two production factors is more important for the research productivity of universities. The results are very clear-cut. In the short run, bombing damages show a negative effect on research outputs. However, this effect is not very persistent. Already by 1961, universities that experienced a lot of destruction have restored their productivity, compared to universities which were bombed less. By 1970, they even turn out to be slightly more productive. Presumably because the reconstruction equipped them with state-of-the-art equipment.
In contrast, the dismissal of Jewish scientists caused a substantial decline in research output. The effect is particularly strong if a department lost its most productive researchers. Star scientists, like Einstein, cannot be replaced easily. Something that we knew already from Azoulay et al. (2010). What is striking, however, is the persistence of the effect. The drop in productivity is still visible in 1980, the last period that the data covers. Departments which lost one of their colleagues in the 1930s are still suffering from it 50 years later. Waldinger estimates the total loss of scientific publications to be around 34% in the disciplines of physics, chemistry, and mathematics.
So here you have the answer why Göttingen is not the world’s leading centre in mathematics anymore. Waldinger’s results convince in their clarity. You have to invest in people, not so much into equipment. Unfortunately, though, building new laboratories is faster than educating new star scientists.
A caveat about the generalizability of the analysis, however, should be born in mind. Nowadays, shocks to physical capital might have a much more persistent effect on research productivity than they had in the 1940s. Otto Hahn’s discovery of nuclear fission in December 1938 was done with the help of an experimental setup installed on an ordinary office desk. By contrast, the network of particle accelerators provided by the European Organization for Nuclear Research (CERN), that nowadays enables nuclear physicists to carry out frontier research, is much more costly to replace.
Azoulay, P., Graff Zivin, J. S., and J. Wang (2010): “Superstar Extinction,” The Quarterly Journal of Economics, 125(2), 549-589.
Waldinger, F. (2013): “Bombs, Brains, and Science,” Working Paper.