목차
Title page
Contents
Abstract 2
1. Introduction 3
2. Background 10
2.1. NIH Funding and R01 Grants 10
2.2. Where do Funding Delays Come From? 11
2.3. Why Interruptions May Affect Employee Outcomes 13
3. Data 15
3.1. Grant and PI variables (ExPORTER) 16
3.2. Finding Lab Personnel (UMETRICS) 17
3.3. Employment and Earnings Data 19
3.4. Publication History 20
3.5. Demographic Data 20
3.6. Analysis Sample 21
4. Estimation 23
4.1. Stacked Difference-in-Differences 23
4.2. Benchmarking Estimates 25
4.3. Identification 25
5. Results 27
5.1. Sector placement 27
5.2. Presence in the United States 34
5.3. Presence in Science: Publications 37
5.4. Earnings and Job Mobility 39
5.5. Robustness 45
6. Conclusion 48
References 51
Appendix 58
A. Figures and Tables 58
B. Alternative Control Group 67
C. Interviews with PIs of Interrupted R01s 67
D. Estimation 68
E. Data Appendix 72
Table 1. Unique Individuals in Single-R01 Labs by Occupation and Birthplace 21
Table 2. Covariate balance between interrupted and continuously-funded labs 22
Table 3. Effect of Interruptions on Main Employment Outcomes 32
Table 4. Effect of Interruptions on Presence in the 2020 Decennial Census for Nonemployed Personnel from Single-R01 Labs 36
Table 5. Effect of Interruptions on Publishing for Nonemployed Personnel from Single-R01 Labs 38
Table 6. Effect of Interruptions on Arcsinh Earnings 43
Table 7. Effect of Interruptions on Job Changes for Fully-Attached Personnel from Single-R01 Labs 44
Table 8. Main Results, for Personnel from Single-R01 Labs, Controlling for Resubmissions 46
Table 9. Effect of Interruptions by Length of Interruption 48
Figure 1. Each point represents a Fiscal Year (FY) from FY1998 to FY2018, and shows the average date on which NIH grants that FY were funded, plotted... 12
Figure 2. This diagram shows the process of linking R01 grants to personnel and their labor market outcomes, starting with NIH ExPORTER data at the top... 15
Figure 3. This figure shows the distribution of funding delays for renewed R01s expiring in Fiscal Years 2005 to 2018. Panel A (top) shows the proportion of... 18
Figure 4. This figure shows event studies of the effects of interruptions on three mutually exclusive employment outcomes: 1) nonemployed in the US... 28
Figure 5. This figure shows the average probability that a personnel is in one of three mutually exclusive employment categories: 1) nonemployed in the US... 30
Figure 6. This figure shows event studies of the effects interruptions on the arcsinh of total earnings. They are obtained using our modified Callaway and... 41
Table A1. Effect of Interruptionson Being at Original University or Different University for Different Subsamples 60
Table A2. Effect of Interruptions on Being Observed in Subsequent Decennial Censuses 61
Table A3. Effect of Interruptions on Publishing and Presence in 2020 Decennial Census if Absent From Tax Data 62
Table A4. Event Studies for Sector of Employment 63
Table A5. Raw Means: Probability by Sector 64
Table A6. Event Studies for Earnings 65
Table A7. Event Studies for TWFE With an Alternative Control Group 66
Table A8. Example of NIH ExPorter data before aggregation into project periods 76
Figure A1. This figure shows the main results from Tham (2023). The figure shows event-study estimates (with 95% confidence intervals) of the difference... 58
Figure A2. This figure shows the event studies using an alternative control group, employees in labs with interrupted R01s and multiple R01s. The first... 59
Figure A3. Histogram of the logarithm of total federal R&D expenditures for all universities (including UMETRICS universities) in the NSF HERD survey... 74
