CF · QJE Revise & Resubmit · Task 1

TIGA outcomes under GWAS-coverage controls

The three outcomes in the paper's original column order — (1) breakthrough, (2) gene share, (3) delay — rebuilt on TIGA value labels, under the paper's controls and with each GWAS-coverage control added. One table per specification.

Table 1. Outcomes on the value tier — paper controls only

Value definition (1) Post2000_TIGA_H (2) share_gene_after (3) Delay_To_TIGA_H N
MediumHighMediumHighMediumHigh
n_study +0.299***
(0.044)
+0.763***
(0.037)
−0.232***
(0.048)
−0.247***
(0.034)
−3.932***
(0.617)
−26.974***
(0.630)
5,515
p-value (median) +0.397***
(0.058)
+0.677***
(0.032)
−0.211***
(0.060)
−0.253***
(0.034)
−4.747***
(0.755)
−26.266***
(0.593)
5,515
p-value (max) +0.354***
(0.065)
+0.698***
(0.033)
−0.210***
(0.052)
−0.257***
(0.036)
−4.197***
(0.796)
−26.564***
(0.632)
5,515
meanRankScore +0.325***
(0.052)
+0.706***
(0.039)
−0.232***
(0.056)
−0.244***
(0.034)
−4.200***
(0.660)
−26.701***
(0.706)
5,515
odds ratio (median) +0.652***
(0.099)
+0.700***
(0.034)
−0.325***
(0.059)
−0.232***
(0.040)
−6.543***
(1.117)
−25.494***
(0.562)
5,515
geneNtrait (gene-level) +0.263***
(0.026)
+0.515***
(0.023)
−0.058***
(0.022)
−0.075**
(0.032)
−3.621***
(0.369)
−22.980***
(0.333)
5,515
Disease-class FE · total publications · clustered SEYes
# TIGA-matched traitsNo
# GWAS-hit opportunitiesNo

Table 2. Outcomes on the value tier — paper controls + # TIGA-matched traits

Value definition (1) Post2000_TIGA_H (2) share_gene_after (3) Delay_To_TIGA_H N
MediumHighMediumHighMediumHigh
n_study +0.191***
(0.045)
+0.634***
(0.040)
−0.262***
(0.055)
−0.282***
(0.047)
−2.850***
(0.647)
−25.692***
(0.667)
5,515
p-value (median) +0.227***
(0.058)
+0.489***
(0.037)
−0.241***
(0.066)
−0.286***
(0.046)
−2.903***
(0.751)
−24.231***
(0.635)
5,515
p-value (max) +0.210***
(0.068)
+0.519***
(0.038)
−0.238***
(0.058)
−0.293***
(0.048)
−2.700***
(0.826)
−24.686***
(0.650)
5,515
meanRankScore +0.194***
(0.051)
+0.548***
(0.039)
−0.261***
(0.063)
−0.279***
(0.044)
−2.888***
(0.671)
−25.100***
(0.700)
5,515
odds ratio (median) +0.473***
(0.106)
+0.508***
(0.041)
−0.331***
(0.065)
−0.238***
(0.045)
−4.808***
(1.200)
−23.638***
(0.592)
5,515
geneNtrait (gene-level) +0.250***
(0.026)
+0.480***
(0.025)
−0.055**
(0.022)
−0.068**
(0.031)
−3.441***
(0.372)
−22.479***
(0.356)
5,515
Disease-class FE · total publications · clustered SEYes
# TIGA-matched traitsYes
# GWAS-hit opportunitiesNo

Table 3. Outcomes on the value tier — paper controls + # GWAS-hit opportunities

Value definition (1) Post2000_TIGA_H (2) share_gene_after (3) Delay_To_TIGA_H N
MediumHighMediumHighMediumHigh
n_study +0.286***
(0.043)
+0.649***
(0.048)
−0.227***
(0.047)
−0.203***
(0.034)
−3.748***
(0.596)
−25.364***
(0.747)
5,515
p-value (median) +0.382***
(0.058)
+0.594***
(0.038)
−0.206***
(0.059)
−0.226***
(0.032)
−4.541***
(0.732)
−25.143***
(0.731)
5,515
p-value (max) +0.338***
(0.065)
+0.609***
(0.039)
−0.205***
(0.051)
−0.228***
(0.034)
−3.961***
(0.749)
−25.216***
(0.765)
5,515
meanRankScore +0.310***
(0.052)
+0.594***
(0.044)
−0.228***
(0.055)
−0.209***
(0.033)
−3.986***
(0.640)
−25.061***
(0.804)
5,515
odds ratio (median) +0.594***
(0.096)
+0.614***
(0.036)
−0.303***
(0.054)
−0.200***
(0.039)
−5.743***
(1.055)
−24.312***
(0.589)
5,515
geneNtrait (gene-level) +0.263***
(0.026)
+0.512***
(0.023)
−0.058***
(0.022)
−0.070**
(0.031)
−3.618***
(0.371)
−22.861***
(0.338)
5,515
Disease-class FE · total publications · clustered SEYes
# TIGA-matched traitsNo
# GWAS-hit opportunitiesYes

Notes. OLS; standard errors clustered by disease class in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Coefficients on Medium and High diseases relative to Low; all six full-sample value definitions shown. Outcomes, in the paper's column order: (1) Post2000_TIGA_H = disease has ≥1 H_TIGA gene first discovered after 2000; (2) share_gene_after = distinct genes explored per post-2000 publication (the value-independent gene-share / herding measure — negative = herding); (3) Delay_To_TIGA_H = years from 1980 to the first H_TIGA discovery (censored at 39). Every estimate keeps its sign and significance across all three specifications.