Discussants

We study the role of informal collaboration in academic knowledge production using published research papers previously presented and discussed at the NBER Summer Institute. We show that papers that have a discussant are published in highly-ranked journals and are more likely to be published in a top journal. Conditional on having a discussant, the quality of a paper’s journal outlet increases in the discussant’s prolificness and editorial experience. This supports the idea that discussants help reduce information asymmetries that are inherent in the academic publication process. Conversely, using social network analysis we find no evidence that citations accumulate because discussants diffuse information about the paper.

1. Discussants
Michael E. Rose
1
Daniel C. Opolot
2
Co-Pierre Georg
2
1
Max Planck Institute for Innovation and Competition, Munich, Germany
2
University of Cape Town, Cape Town, South Africa
1

2. Science is a social endeavor
ˆWorking with co-authors the norm in scienceWuchty et al. (Science 2007);
Chung et al. (QREF, 2009); Ductor (OBES, 2015)
ˆThere is alsoinformal collaborationLaband and Tollison (JPE, 2002)
ˆEditors
ˆReferees
ˆDoctoral Advisers
ˆColleagues everywhere
Æ)Is informal collaboration part of the academic production function?
2

3. Science is a social endeavor
ˆWorking with co-authors the norm in scienceWuchty et al. (Science 2007);
Chung et al. (QREF, 2009); Ductor (OBES, 2015)
ˆThere is alsoinformal collaborationLaband and Tollison (JPE, 2002)
ˆEditors
ˆReferees
ˆDoctoral Advisers
ˆColleagues everywhere
Æ)Is informal collaboration part of the academic production function?
2

4. Studies on academic collaboration and "paper" as a unit
formal informal
personAzoulay, Gra Zivin and
Wang (QJE, 2010)
Oettl (MS, 2012)
paperDuctor (OBES, 2015) Laband and Tollison (JPE,
2002)(aggregate counts only,
no causal attempt)
3

5. Laboratory: Discussants at NBER Summer Institutes
ˆAnnual, highly prestigious workshop series
ˆWorkshops organized by groups and programs
VSome workshops always havediscussants, the others never do
ˆFocus on Finance for topical homogeneity
4

6. Q: Do Discussants matter?
1. Are papers with discussants at NBER SIs more success-
ful than those with general discussion?
2a. Are papers with more prolic discussants more success-
ful?
2b. Are higher citation counts of discussed papers due to
discussant's diusion eorts?
5

7. A: Discussants do matter
1. Are papers with discussants at NBER SIs more success-
ful than those with general discussion?
Yes
2a. Are papers with more prolic discussants more success-
ful?
Partly
2b. Are higher citation counts of discussed papers due to
discussant's diusion eorts?
No
6

8. Data
ˆ596 published papers presented at 85 workshops of NBER SIs of 12
Finance-related Groups between 2000 and 2009
ˆAuthor and discussants characteristics: Euclidean index of citations,
Experience
ˆCitation count(log) from Scopus, Top Journal Status, Journal Prestige
from SCImago
ˆNeighborhood centrality in two collaboration networks
7

9. 922 presentations in 85 workshops
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 total share
AMRE 5 (6) 5 (6) 10 (12) 83%
AP 8 (8) 5 (8) 8 (9) 8 (10) 6 (9) 9 (9) 9 (9) 7 (9) 11 (12) 8 (8) 79 (91) 87%
CF 9 (10) 4 (7) 7 (9) 9 (11) 10 (13) 13 (14) 10 (11) 11 (13) 16 (18) 11 (12) 100 (118) 85%
EFCE 11 (15) 7 (17) 8 (17) 7 (15) 14 (17) 10 (15) 13 (15) 11 (15) 10 (15) 9 (15) 100 (156) 64%
EFEL 13 (16) 13 (15) 11 (12) 6 (11) 9 (11) 11 (12) 10 (12) 8 (10) 9 (12) 90 (111) 81%
EFFE 11 (14) 10 (12) 9 (12) 8 (12) 10 (12) 9 (12) 7 (12) 9 (12) 6 (12) 8 (12) 87 (122) 71%
IFM 6 (8) 7 (8)* 11 (12) 8 (11) 10 (12) 5 (8) 8 (10) 7 (10) 9 (12) 11 (14) 82 (105) 78%
ME 8 (10) 5 (8) 7 (11) 8 (9) 9 (12) 7 (12) 13 (14) 7 (12) 9 (13) 12 (13) 85 (114) 75%
PERE 4 (5) 6 (6) 4 (6) 1 (6) 6 (9) 7 (10) 5 (10) 7 (10) 4 (11) 8 (9) 52 (82) 63%
RISK 3 (3)1 (2) 2 (3) 11 (12)17 (20) 85%
total 57 (70) 59 (84) 72 (97) 60 (86) 71 (95) 69 (91) 79 (96) 69 (93) 73 (103) 87 (107) 696 (922)
share 81% 70% 74% 70% 75% 76% 82% 74% 71% 81% 75%
Numbers indicate the number of presentations that resulted in a publication (Total number number of
presentations) per individual Workshop. Workshops with asterisk were part of another workshop.
8

10. Are the groups comparable?
ˆSame readability at presentation (= proxy for quality)
t-test
ˆSame average aliation ranking of authors at presentation
t-test
ˆSame average duration (55 min.)
t-test
ˆTopically similar (based on cited journals)
9

11. Groups w/ and w/o discussants topically similarˆSimilarity = cosine of overlap of weighted counts of journals cited in
presented papers
ˆweights obtained fromtdf-vectorization
details
ˆAll papers (in our sample): 0.09
ˆAll top 5 Econ papers: between 0.034 (AER) and 0.07 (Ectma)
EFCE ME IFM EFEL RISK PERE EFFE AMRE AP CF
EFCE 1
ME 0.95 1
IFM 0.8 0.8 1
EFEL 0.7 0.68 0.71 1
RISK 0.47 0.48 0.54 0.87 1
PERE 0.41 0.42 0.39 0.45 0.32 1
EFFE 0.39 0.44 0.42 0.47 0.4 0.21 1
AMRE 0.33 0.35 0.37 0.58 0.56 0.68 0.32 1
AP 0.32 0.31 0.42 0.82 0.88 0.24 0.44 0.57 1
CF 0.29 0.32 0.43 0.8 0.86 0.28 0.35 0.55 0.95 1
10

12. Groups w/ and w/o discussants topically similarˆSimilarity = cosine of overlap of weighted counts of journals cited in
presented papers
ˆweights obtained fromtdf-vectorization
details
ˆAll papers (in our sample): 0.09
ˆAll top 5 Econ papers: between 0.034 (AER) and 0.07 (Ectma)
EFCE ME IFM EFEL RISK PERE EFFE AMRE AP CF
EFCE 1
ME 0.95 1
IFM 0.8 0.8 1
EFEL 0.7 0.68 0.71 1
RISK 0.47 0.48 0.54 0.87 1
PERE 0.41 0.42 0.39 0.45 0.32 1
EFFE 0.39 0.44 0.42 0.47 0.4 0.21 1
AMRE 0.33 0.35 0.37 0.58 0.56 0.68 0.32 1
AP 0.32 0.31 0.42 0.82 0.88 0.24 0.44 0.57 1
CF 0.29 0.32 0.43 0.8 0.86 0.28 0.35 0.55 0.95 1
10

13. Groups w/ and w/o discussants topically similarˆSimilarity = cosine of overlap of weighted counts of journals cited in
presented papers
ˆweights obtained fromtdf-vectorization
details
ˆAll papers (in our sample): 0.09
ˆAll top 5 Econ papers: between 0.034 (AER) and 0.07 (Ectma)
EFCE ME IFM EFEL RISK PERE EFFE AMRE AP CF
EFCE 1
ME 0.95 1
IFM 0.8 0.8 1
EFEL 0.7 0.68 0.71 1
RISK 0.47 0.48 0.54 0.87 1
PERE 0.41 0.42 0.39 0.45 0.32 1
EFFE 0.39 0.44 0.42 0.47 0.4 0.21 1
AMRE 0.33 0.35 0.37 0.58 0.56 0.68 0.32 1
AP 0.32 0.31 0.42 0.82 0.88 0.24 0.44 0.57 1
CF 0.29 0.32 0.43 0.8 0.86 0.28 0.35 0.55 0.95 1
10

14. 1. Are papers with discussants at NBER SIs more successful
than those with general discussion?
10

15. Interesting descriptives
ˆ592 observations (=published in a Journal in 2021 latest)
ˆ50% in top publications (JF, JFE, RFS, AER, JPE, QJE, RESTud, Ecmta)
ˆ60% with discussant; median experience = 11 years
Distribution
ˆ50% published between 2 and 5 years later
DistributionFull summary statistics
11

16. Identication assumptions
A1
A2
A3
sorting)
12

17. Empirical strategy
Top
i
Æ®0Å®1¢Paper
i
Å®2¢Authori,t¡1Å®3¢Group
i
ů¢DiscussioniŲi(1)
SJRi,tÆ®0Å®1¢Paper
i
Å®2¢Authori,t¡1Å®3¢Group
i
ů¢DiscussioniŲi(2)
log(1ÅCitationsi,2022)Æ®0Å®1¢Paper
i
Å®2¢Authori,t¡1Å®3Group
i
Å(3)
Å®4¢SJRi,tů¢DiscussioniÅ°tŲi
13

18. Discussants and Journal Status
Top (Finance) Top (Econ+Finance)
(1) (2) (3) (4)
Discussion 3 .399
¤¤¤
2.939
¤¤¤
1.330
¤¤¤
1.039
¤
(0.000) ( 0.000) ( 0.002) ( 0.075)
Constant ¡5.012
¤¤¤
¡5.264
¤¤¤
¡2.440
¤¤
¡1.393
(0.000) ( 0.000) ( 0.025) ( 0.126)
Paper controls X X X X
Author controls X X X X
Group control X X X X
JEL categories X X
N 593 580 593 593
Pseudo R
2
0.281 0 .464 0 .208 0 .275
AIC 515 .6 386 .2 674 .6 619 .5
Random inference 0 .0333
¤¤
0.0656
¤
0.1174 0 .1458
Note:Logistic regression. Standard errors clustered on NBER group or joint workshop.
14

19. Discussants and journal quality
SJR Avg. citations h-index
(1) (2) (3) (4) (5) (6)
Discussion 2 .182
¤¤¤
1.767
¤
0.968
¤¤¤
0.678
¤¤
33.00
¤¤
17.36
(0.005) ( 0.077) ( 0.002) ( 0.037) ( 0.010) (0.217)
Constant 0 .166 1 .523 0 .723 1 .159
¤
130.5
¤¤¤
156.7
¤¤¤
(0.897) ( 0.280) ( 0.186) ( 0.056) ( 0.000) (0.000)
Paper controls X X X X X X
Author controls X X X X X X
Group control X X X X X X
JEL categories X X X
N 576 576 576 576 576 576
R
2
0.277 0 .321 0 .261 0 .307 0 .162 0 .242
Random inference 0 .0949
¤
0.0656
¤
0.0910
¤
0.0783
¤
0.1321 0 .2133
Note:Logistic regression. Standard errors clustered on NBER group or joint workshop.
log transformation
15

20. Discussants and citation counts
log(1+Total citations)
(1) (2) (3) (4)
Discussion 0 .260
¤
0.108 ¡0.0294 ¡0.0146
(0.056) ( 0.392) ( 0.833) ( 0.889)
Constant 3 .367
¤¤¤
3.417
¤¤¤
3.410
¤¤¤
4.482
¤¤¤
(0.000) ( 0.000) ( 0.000) ( 0.000)
Paper controls X X X X
Author controls X X X X
Group control X X X X
Publication year FE X X X X
JEL categories X X X
Journal control X
Journal FE X
N 596 596 576 596
R
2
0.359 0 .386 0 .427 0 .584
Note:Logistic regression. Standard errors clustered on NBER group or joint workshop.
Negative Binomialsine hyperbolicusFlexible lags 16

21. Recap
ˆDiscussants help bring the paper to better journal on average
ˆ. . . but papers not better cited on average given the journal quality
)Papers get published where they belong
Next:
17

22. 2b. Are higher citation counts of discussed papers due to
discussant's diusion eorts?
2b. Are higher citation counts of discussed papers due to
discussant's diusion eorts?
17

23. Identication assumptions
A4
A5
A6
Survey resultsAss. matching on seniorityAss. matching on characteristicsAss. matching on centrality
18

24. Discussants characteristics
Probe 350 papers with known discussants
ˆEuclidean index of citations in year of discussion
ˆExperience
ˆGender (via genderize.io)
ˆAliation rank
ˆPractitioner (central bank, law rm, government)
ˆEditorial experience (obtained from CVs)
in year of discussion
19

25. Editorial experience matters
...Æ®0Å®1¢Paper
i
Å®2¢Authori,t¡1Å®3Group
i
ů¢DiscussantiųgÅ°tŲi
ˆTop journal: No eect of any characteristic
ˆJournal quality:
ˆEuclidean index (0.00240
¤¤¤
)
ˆaliation rank (¡0.00279
¤¤
)
ˆeditorial experience (1.068
¤¤¤
)
ˆLog Citation counts (w/ journal quality controls): No eect of any
characteristic
20

26. Editorial experience matters
...Æ®0Å®1¢Paper
i
Å®2¢Authori,t¡1Å®3Group
i
ů¢DiscussantiųgÅ°tŲi
ˆTop journal: No eect of any characteristic
ˆJournal quality:
ˆEuclidean index (0.00240
¤¤¤
)
ˆaliation rank (¡0.00279
¤¤
)
ˆeditorial experience (1.068
¤¤¤
)
ˆLog Citation counts (w/ journal quality controls): No eect of any
characteristic
20

27. P(discussant cites "her" paper) < P(other participants cite it)0 5 10 15 20
Years until/since publication
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Citation probability
Citation by
Same authors
Other workshop authors
Own discussant
Other workshop discussants
21

28. Neighborhood Centrality
What's the number of all neighbors of a researcheriwithin a given radius¿,
while discounting for distance with given information quality lter±?
Neighborhood
i(±)Æ
1X
¿Æ1
±
¿¡1
ki¿
!Measures ability to spread information
22

29. Neighborhood centrality in two networks
ˆFormalnetwork linking authors on a paper in 370 Economics journals
a
ˆInformalnetwork (CoFE) linking authors and acknowledged commenters on
papers from 6 Finance journals
b
Rose and Georg (RP 2021)
ˆPapers from 2000-2011, network intuses publications fromt¡1,t,tÅ1
a
Journals ranked at least C by Combes and Linnemer (2010) in the following categories:
General Economics, C, D, E, F, G, H, K, and R
b
JF, RFS, JFE, JFI, JMCB, JBF
23

30. Neighborhood centrality in two networks
ˆFormalnetwork linking authors on a paper in 370 Economics journals
a
ˆInformalnetwork (CoFE) linking authors and acknowledged commenters on
papers from 6 Finance journals
b
Rose and Georg (RP 2021)
ˆPapers from 2000-2011, network intuses publications fromt¡1,t,tÅ1
a
Journals ranked at least C by Combes and Linnemer (2010) in the following categories:
General Economics, C, D, E, F, G, H, K, and R
b
JF, RFS, JFE, JFI, JMCB, JBF
23

31. More central discussants6!higher citation count (informal net-
work)−0.20
−0.15
−0.10
−0.05
0.00
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95
l
l
l
l l l l l l l
24

32. More central discussants6!higher citation count (coauthor net-
work)−0.4
−0.2
0.0
0.2
0.4
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95
l
l
l
l
l l l l l l
25

33. Take-aways of "Discussants"
ˆInformal collaboration widespread, poorly understood
ˆDiscussants = assigned informal collaborators
ˆHaving a discussant increases chances to publish high, though not citation
count
ˆDiscussants with editorial experience help the most, butnotbecause of
diusion eects
Thank you!
26

34. Appendix
26

35. Summary statistics of Journal sample
N Mean Median Std.Dev. Min Max
Total citations 596 170 .50 93.00 263.80 0.00 3418.00
Top publication 596 0 .51 1.00 0 .50 0.00 1 .00
SJR 576 8 .09 6.89 5 .14 0.16 22.54
# of pages 596 29 .62 29.50 11.56 5.00 87.00
# of authors 596 2 .18 2.00 0 .80 1.00 5 .00
Age 596 3 .52 3.00 2 .05¡1.00 14.00
Author total Euclid 596 431 .96 212.83 710.20 0.00 8311.96
Youngest author experience 596 6.89 5.00 6 .68 0.00 54.00
Oldest author experience 596 15.14 14.00 9 .76 0.00 54.00
Discussion 596 0 .63 1.00 0 .48 0.00 1 .00
p
74
2
Å18
2
Å40
2
Å74
2
Å180
2
Æ213.8
back
27

36. Summary statistics of Journal sample
N Mean Median Std.Dev. Min Max
Total citations 596 170 .50 93.00 263.80 0.00 3418.00
Top publication 596 0 .51 1.00 0 .50 0.00 1 .00
SJR 576 8 .09 6.89 5 .14 0.16 22.54
# of pages 596 29 .62 29.50 11.56 5.00 87.00
# of authors 596 2 .18 2.00 0 .80 1.00 5 .00
Age 596 3 .52 3.00 2 .05¡1.00 14.00
Author total Euclid 596 431 .96 212.83 710.20 0.00 8311.96
Youngest author experience 596 6.89 5.00 6 .68 0.00 54.00
Oldest author experience 596 15.14 14.00 9 .76 0.00 54.00
Discussion 596 0 .63 1.00 0 .48 0.00 1 .00
p
74
2
Å18
2
Å40
2
Å74
2
Å180
2
Æ213.8
back
27

37. Distribution of publication lag0 2 4 6 8 10 12 14
Years until publication
back
28

38. Similar readabilityWith Without
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
Gunning fog index
p = 0.87
Paper level
With Without
Gunning fog index
p = 0.96
Workshop averages
Readability linked to future citations; Economics Letters GF index = 16.08
Dowling, Hammami and Zreik (EL, 2018)
back 29

39. Similar average Tilburg aliation rankWith Without
0
20
40
60
80
100
Avg. affiliation rank
p = 0.64
Paper level
With Without
Avg. affiliation rank
p = 0.80
Workshop averages
Note:Rankings for nancial institutions not available.
back 30

40. Similar presentation durationWith Without
0
10
20
30
40
50
60
Duration (in min)
p = 0.41
Paper level
With Without
Duration (in min)
p = 0.32
Workshop averages
back
31

41. Distribution of discussant experience0 5 10 15 20 25 30 35
Discussant experience (in years)
0.0
0.2
0.4
0.6
0.8
1.0
Empirical cumulative density
0
1
2
3
4
5
6
7
8
Frequency of observation (in %)
back
32

42. tdf-vectorization to measure topical similarityˆPaper 1 cites journals A, B, C, D in entire period
ˆPaper 2 cites journals C, D, E, F
ˆPaper 3 cites journals C, F
0
B
B
B
B
B
B
B
B
B
@
1 0 0
1 0 0
1 1 1
1 1 0
0 1 0
0 1 1
1
C
C
C
C
C
C
C
C
C
A
tdf
¡¡¡¡¡¡¡¡!
vectorization
0
B
B
B
B
B
B
B
B
B
@
0.58 0 0
0.58 0 0
0.34 0.37 0.61
0.44 0.48 0
0 0.63 0
0 0.48 0.79
1
C
C
C
C
C
C
C
C
C
A
Cosine similarity between e.g. 1 and 2: 0.34
back
33

43. tdf-vectorization to measure topical similarityˆPaper 1 cites journals A, B, C, D in entire period
ˆPaper 2 cites journals C, D, E, F
ˆPaper 3 cites journals C, F
0
B
B
B
B
B
B
B
B
B
@
1 0 0
1 0 0
1 1 1
1 1 0
0 1 0
0 1 1
1
C
C
C
C
C
C
C
C
C
A
tdf
¡¡¡¡¡¡¡¡!
vectorization
0
B
B
B
B
B
B
B
B
B
@
0.58 0 0
0.58 0 0
0.34 0.37 0.61
0.44 0.48 0
0 0.63 0
0 0.48 0.79
1
C
C
C
C
C
C
C
C
C
A
Cosine similarity between e.g. 1 and 2: 0.34
back
33

44. tdf-vectorization to measure topical similarityˆPaper 1 cites journals A, B, C, D in entire period
ˆPaper 2 cites journals C, D, E, F
ˆPaper 3 cites journals C, F
0
B
B
B
B
B
B
B
B
B
@
1 0 0
1 0 0
1 1 1
1 1 0
0 1 0
0 1 1
1
C
C
C
C
C
C
C
C
C
A
tdf
¡¡¡¡¡¡¡¡!
vectorization
0
B
B
B
B
B
B
B
B
B
@
0.58 0 0
0.58 0 0
0.34 0.37 0.61
0.44 0.48 0
0 0.63 0
0 0.48 0.79
1
C
C
C
C
C
C
C
C
C
A
Cosine similarity between e.g. 1 and 2: 0.34
back
33

45. Discussants and journal quality, log transformation
log(SJR) log(Avg. citations) log(h-index)
(1) (2) (3) (4) (5) (6)
Discussion 0 .279
¤¤
0.198 0 .320
¤¤¤
0.202
¤
0.210
¤¤¤
0.102
(0.035) ( 0.177) ( 0.004) ( 0.068) ( 0.009) ( 0.321)
Constant 0 .899
¤¤¤
1.128
¤¤¤
0.256 0 .402
¤¤
4.848
¤¤¤
5.056
¤¤¤
(0.001) ( 0.000) ( 0.151) ( 0.049) ( 0.000) ( 0.000)
Paper controls X X X X X X
Author controls X X X X X X
Group control X X X X X X
JEL categories X X X
N 576 576 576 576 576 576
R
2
0.181 0 .266 0 .200 0 .264 0 .121 0 .209
Note:Standard errors clustered on NBER group or joint workshop.
back
34

46. Discussants and citation counts, Negative Binomial
(1) (2) (3)
Total citations
Discussion 0 .203 0 .0576 ¡0.0289
(0.121) ( 0.661) ( 0.849)
Constant 3 .829
¤¤¤
3.860
¤¤¤
3.854
¤¤¤
(0.000) ( 0.000) ( 0.000)
Paper controls X X X
Author controls X X X
Group control X X X
Publication year FE X X X
JEL categories X X
Journal control X
N 596 596 576
Pseudo R
2
0.0406 0 .0466 0 .0503
AIC 7033 .5 6989 .6 6756 .1
Note:Standard errors clustered on NBER group or joint workshop.
back 35

47. Discussants and citation counts, inverse hyperbolic sine
(1) (2) (3)
asinh(Total citations)
Discussion 0 .265
¤
0.110 ¡0.0349
(0.059) ( 0.393) ( 0.804)
Constant 4 .008
¤¤¤
4.052
¤¤¤
4.048
¤¤¤
(0.000) ( 0.000) ( 0.000)
Paper controls X X X
Author controls X X X
Group control X X X
Publication year FE X X X
JEL categories X X
Journal control X
N 596 596 576
R
2
0.356 0 .383 0 .426
Note:Standard errors clustered on NBER group or joint workshop.
back
36

48. Discussants and citation counts after varying durationYears since publication
−0.4
−0.2
0.0
0.2
0.4
0.6
base 0 1 2 3 4 5 6 7 8 9 10 11 12 13
l
l
l l l l
l
l l
l
l
l
l
l
l
Note:Standard errors clustered on NBER group or joint workshop.
back
37

49. Workshop organizers look for topical t and ability to presentExcerpts from 15 respondents responsible for 42 workshops:
ˆDiscussants should form "a basis for a lively, productive debate between
authors, discussants, and audience."
ˆDiscussants should not "too close to the author and if possible coming from
a dierent perspective"
ˆThey should have "No fear of authors (i.e., probably don't get a very junior
person to discuss a big shot, unless you know the junior person is fearless)."
ˆThere are "often authors of good papers that were not chosen for
presentation" which qualify as discussant
back
38

50. Do senior authors get senior discussants? 0 5 10 15 20 25 30
Authors' mean experience
0
5
10
15
20
25
30
35
40
Discussant experience
b = 0.13
= 0.11
0 10 20 30 40 50
Authors' max. experience
b = 0.14
= 0.096
AMRE
AP
CF
EFEL
IFM
PERE
RISK
back
39

51. No assortative matching with prolic discussants
Dis. experience Top-30 Univ. Dis. Euclid
neg. bin. logistic OLS
Author max. Euclid ¡0.00005 0 .001 ¡0.054
(0.730) ( 0.172) ( 0.216)
Author max. experience 0 .004 ¡0.023 ¡2.114
(0.590) ( 0.276) ( 0.330)
Author max. experience
2
0.00004 0 .0002 0 .067
¤¤¤
(0.588) ( 0.416) ( 0.004)
# of authors ¡0.031 0 .220 ¡1.173
(0.556) ( 0.148) ( 0.943)
Discussion year FE X X
NBER group FE X X
N 441 401 441
Adjusted R
2
0.206
Akaike Inf. Crit. 3,058.030 526.004
Note:Standard errors clustered on NBER group or joint workshop. Constant not reported.
back
40

52. No assortative matching with central discussants
Dis. neighborhood
coauthor informal
Author max. Euclid ¡0.004 ¡0.019
pÆ0.703 pÆ0.541
Author max. experience ¡0.282 ¡2.481
pÆ0.539 pÆ0.116
Author max. experience
2
0.006 0 .029
¤
pÆ0.194 pÆ0.087
# of authors 4 .141 15 .607
pÆ0.230 pÆ0.188
Discussion year FE X X
NBER group FE X X
N 441 441
Adjusted R
2
0.236 0.387
Note:Standard errors clustered on NBER group or joint workshop.
back
41