Peer effects in science: evidence from the dismissal of scientists in Nazi Germany

This paper analyzes peer effects among university scientists. Specifically, it investigates whether the number of peers and their average quality affects the productivity of researchers in physics, chemistry, and mathematics. The usual endogeneity problems related to estimating peer effects are addressed by using the dismissal of scientists by the Nazi government as a source of exogenous variation in the peer group of scientists staying in Germany. Using a newly constructed panel dataset covering the universe of physicists, chemists, and mathematicians at all German universities from 1925 until 1938 I investigate peer effects at the local level and among co-authors. There is no evidence for localized peer effects, as neither department level (e.g. the physics department) nor specialization level (e.g. all theoretical physicists in the department) peers affect a researcher's productivity. Among co-authors, however, there is strong and significant evidence that peer quality affects a researcher's productivity. Loosing a co-author of average quality reduces the productivity of an average scientist by about 13 percent in physics and 16.5 percent in chemistry.