Table 3:

Effect of Superstar Death on Subfield Entry Rates

	Publication Flows			NIH Funding Flows (Nb. of Awards)
	All Authors	Collaborators Only	Non- Collaborators Only	All Authors	Collaborators Only	Non- Collaborators Only
	(1)	(2)	(3)	(4)	(5)	(6)
After Death	0.051† (0.029)	−0.232** (0.057)	0.082** (0.029)	0.046 (0.035)	−0.265** (0.076)	0.110** (0.033)
Nb. of Investigators	6,260	6,124	6,260	6,215	5,678	6,202
Nb. of Fields	34,218	33,096	34,218	33,912	29,163	33,806
Nb. of Field-Year Obs.	1,259,176	1,217,905	1,259,176	1,049,942	902,873	1,046,678
Log Likelihood	−2,891,116	−729,521	−2,768,257	−1,350,208	−472,329	−1,223,915

Note: Estimates stem from conditional (subfield) fixed effects Poisson specifications. The dependent variable is the total number of publications in a subfield in a particular year (columns 1, 2, and 3), or the total number of NIH grants that acknowledge a publication in a subfield (columns 4, 5, and 6). All models incorporate a full suite of year effects and subfield age effects, as well as a term common to both treated and control subfields that switches from zero to one after the death of the star, to address the concern that age, year and individual fixed effects may not fully account for trends in subfield entry around the time of death. Exponentiating the coefficients and differencing from one yield numbers interpretable as elasticities. For example, the estimates in column (3) imply that treated subfields see an increase in the number of contributions by non-collaborators after the superstar passes away—a statistically significant 100×(exp[0.082]-1)=8.55%. The number of observations varies slightly across columns because the conditional fixed effects specification drops observations corresponding to subfields for which there is no variation in activity over the entire observation period.

Robust standard errors in parentheses, clustered at the level of the star scientist.

^†p < 0.10,

^*p < 0.05,

^**p < 0.01.