            name:  <unnamed>
             log:  C:\Users\isyoo\OneDrive - Singapore Management University\Desktop\
      > Work\JAR AI Aanalysts\JAR Replication Package\Log flies\Askresearchgpt analys
      > es.smcl
        log type:  smcl
       opened on:  19 Jul 2026, 08:56:15
      
     1 . do "C:\Users\isyoo\AppData\Local\Temp\STD407c_000008.tmp"
      
     2 . /*AskResearchGPT Analyses*/
     3 . 
     4 . /*Timeliness*/
     5 . cd "C:\Users\isyoo\OneDrive - Singapore Management University\Desktop\Work\JA
      > R AI Aanalysts"
      C:\Users\isyoo\OneDrive - Singapore Management University\Desktop\Work\JAR AI A
      > analysts
      
     6 . use "askresearchgpt_timeliness_sample", clear
      
     7 . rename *, lower 
      
     8 . keep if year==2024 
      (87,323 observations deleted)
      
     9 . gen post=0
      
    10 . replace post=1 if month>9 & year==2024
      (6,730 real changes made)
      
    11 . drop if (year == 2024 & (month < 4 | month > 11)) | inrange(month, 6, 9)
      (9,939 observations deleted)
      
    12 . gen treat=0
      
    13 . replace treat=1 if brokergvkey==12124
      (1,165 real changes made)
      
    14 . egen firm=group(ticker)
      
    15 . egen firmbroker=group(ticker estimator)
      
    16 . egen broker=group(estimator)
      
    17 . egen yearmonth=group(year month)
      
    18 . keep if lag<5
      (2,798 observations deleted)
      
    19 . drop if lag<0
      (0 observations deleted)
      
    20 . gen eagap=filing_date-anndats_act_l
      
    21 . drop if eagap<0
      (1 observation deleted)
      
    22 . /*Table 1 Panel E*/
    23 . tabstat lag, statistics(n mean sd p5 p25 median p75 p95) columns(statistics)
      
          Variable |         N      Mean        SD        p5       p25       p50     
      >   p75       p95
      -------------+-----------------------------------------------------------------
      ---------------
               lag |      7973  .7062586  1.046346         0         0         0     
      >     1         3
      -------------------------------------------------------------------------------
      ---------------
      
    24 . /*Regression results: Table 9 Column 1*/
    25 . eststo clear
      
    26 . eststo: reghdfe lag post##treat, absorb(firm broker yearmonth) vce(cluster fi
      > rm) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 10 iterations)
      note: 1bn.post is probably collinear with the fixed effects (all partialled-out
      >  values are close to zero; tol = 1.0e-09)
      note: 1bn.treat is probably collinear with the fixed effects (all partialled-ou
      > t values are close to zero; tol = 1.0e-09)
      
      HDFE Linear regression                            Number of obs   =      7,973
      Absorbing 3 HDFE groups                           F(   1,   1727) =       3.37
      Statistics robust to heteroskedasticity           Prob > F        =     0.0666
                                                        R-squared       =     0.4488
                                                        Adj R-squared   =     0.2945
                                                        Within R-sq.    =     0.0006
      Number of clusters (firm)    =      1,728         Root MSE        =     0.8789
      
                                     (Std. err. adjusted for 1,728 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
               lag | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
            1.post |          0  (omitted)
           1.treat |          0  (omitted)
                   |
        post#treat |
              1 1  |  -.1391216   .0757905    -1.84   0.067    -.2877724    .0095291
                   |
             _cons |   .7127671   .0035457   201.02   0.000     .7058128    .7197214
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
              firm |      1728        1728           0    *|
            broker |        14           1          13     |
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est1 stored)
      
    27 . outreg2 using 2.doc, nolabel tstat tdec(2) dec(3) replace
      2.doc
      dir : seeout
      
    28 . noisily: esttab, r2 star(* 0.1 ** 0.05 *** 0.01)  nonote b(%9.3f) keep(1.post
      > #1.treat) 
      
      ----------------------------
                            (1)   
                            lag   
      ----------------------------
      1.post#1.t~t       -0.139*  
                        (-1.84)   
      ----------------------------
      N                    7973   
      R-sq                0.449   
      ----------------------------
      
    29 . /*Figure 3 Panel A*/
    30 . gen zero = 0
      
    31 . gen r_4=0
      
    32 . replace r_4=1 if month==4 & treat==1 & year==2024
      (100 real changes made)
      
    33 . gen r_3=0
      
    34 . replace r_3=1 if month==5 & treat==1 & year==2024
      (296 real changes made)
      
    35 . gen r_2=0
      
    36 . replace r_2=1 if month==10 & treat==1 & year==2024
      (134 real changes made)
      
    37 . gen r_1=0
      
    38 . replace r_1=1 if month==11 & treat==1 & year==2024
      (239 real changes made)
      
    39 . local running "r_4 zero r_2 r_1 post treat"
      
    40 . eststo: reghdfe lag `running', absorb(firm broker yearmonth) vce(cluster firm
      > ) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 10 iterations)
      note: post is probably collinear with the fixed effects (all partialled-out val
      > ues are close to zero; tol = 1.0e-09)
      note: treat is probably collinear with the fixed effects (all partialled-out va
      > lues are close to zero; tol = 1.0e-09)
      note: zero omitted because of collinearity
      
      HDFE Linear regression                            Number of obs   =      7,973
      Absorbing 3 HDFE groups                           F(   3,   1727) =       1.18
      Statistics robust to heteroskedasticity           Prob > F        =     0.3164
                                                        R-squared       =     0.4489
                                                        Adj R-squared   =     0.2943
                                                        Within R-sq.    =     0.0006
      Number of clusters (firm)    =      1,728         Root MSE        =     0.8790
      
                                     (Std. err. adjusted for 1,728 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
               lag | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
               r_4 |   .0257858   .1437825     0.18   0.858    -.2562205     .307792
              zero |          0  (omitted)
               r_2 |  -.1296981   .1052169    -1.23   0.218    -.3360642    .0766679
               r_1 |  -.1339483   .0974321    -1.37   0.169    -.3250457     .057149
              post |          0  (omitted)
             treat |          0  (omitted)
             _cons |   .7121303   .0050992   139.65   0.000      .702129    .7221316
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
              firm |      1728        1728           0    *|
            broker |        14           1          13     |
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est2 stored)
      
    41 . coefplot, level(90) omitted ciopts(recast(rcap)) vertical yline(0, lcolor(bla
      > ck)) xline(2.5) coeflabels(r_4 = "Apr" zero = "May" r_2 = "Oct" r_1 = "Nov") 
      > keep(r_4 zero r_2 r_1)  recast(connected)  lpattern(dash) lcolor(gray) 
      
    42 . 
    43 . /*Accuracy, boldness, new coverage*/
    44 . use "askresearchgpt_sample", clear
      
    45 . rename *, lower 
      
    46 . keep if year==2024 
      (480,795 observations deleted)
      
    47 . gen post=0
      
    48 . replace post=1 if month>9 & year==2024
      (16,364 real changes made)
      
    49 . drop if (year == 2024 & (month < 4 | month > 11)) | inrange(month, 6, 9)
      (37,004 observations deleted)
      
    50 . gen treat=0
      
    51 . replace treat=1 if brokergvkey==12124
      (2,906 real changes made)
      
    52 . egen firm=group(ticker)
      
    53 . egen firmbroker=group(ticker estimator)
      
    54 . egen broker=group(estimator)
      
    55 . egen yearmonth=group(year month)
      
    56 . gen error=abs(value-actual)/prcc_f_1
      (3,653 missing values generated)
      
    57 . gen accu=(error)*(-1)
      (3,653 missing values generated)
      
    58 . replace first=0 if first==.
      (27,236 real changes made)
      
    59 . replace indfirst=0 if indfirst==.
      (27,784 real changes made)
      
    60 . drop if accu==.
      (3,653 observations deleted)
      
    61 . winsor2 accu, cuts(1 99) replace 
      
    62 . /*Table 1 Panel E*/
    63 . tabstat treat post accu bold first indfirst, statistics(n mean sd p5 p25 medi
      > an p75 p95) columns(statistics)
      
          Variable |         N      Mean        SD        p5       p25       p50     
      >   p75       p95
      -------------+-----------------------------------------------------------------
      ---------------
             treat |     24446  .1101612  .3130969         0         0         0     
      >     0         1
              post |     24446  .4871145  .4998442         0         0         0     
      >     1         1
              accu |     24446 -.0287059  .0925965 -.1098968 -.0158594 -.0045941 -.00
      > 13081 -.0001491
              bold |     22527  .7233986  .4473276         0         0         1     
      >     1         1
             first |     24446  .0274073  .1632706         0         0         0     
      >     0         0
          indfirst |     24446  .0115356  .1067849         0         0         0     
      >     0         0
      -------------------------------------------------------------------------------
      ---------------
      
    64 . /*Regression results: Table 9 Columns 2 through 5*/
    65 . eststo clear
      
    66 . eststo: reghdfe accu post##treat, absorb(firmbroker yearmonth) vce(cluster fi
      > rm) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: 1bn.post is probably collinear with the fixed effects (all partialled-out
      >  values are close to zero; tol = 1.0e-09)
      note: 1bn.treat is probably collinear with the fixed effects (all partialled-ou
      > t values are close to zero; tol = 1.0e-09)
      
      HDFE Linear regression                            Number of obs   =     24,446
      Absorbing 2 HDFE groups                           F(   1,   2849) =       3.11
      Statistics robust to heteroskedasticity           Prob > F        =     0.0780
                                                        R-squared       =     0.8879
                                                        Adj R-squared   =     0.8055
                                                        Within R-sq.    =     0.0002
      Number of clusters (firm)    =      2,850         Root MSE        =     0.0408
      
                                     (Std. err. adjusted for 2,850 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
              accu | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
            1.post |          0  (omitted)
           1.treat |          0  (omitted)
                   |
        post#treat |
              1 1  |   .0034277   .0019444     1.76   0.078    -.0003848    .0072403
                   |
             _cons |  -.0288915   .0001053  -274.35   0.000     -.029098    -.028685
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |     10353       10353           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est1 stored)
      
    67 . outreg2 using 2.doc, nolabel tstat tdec(2) dec(3) replace
      2.doc
      dir : seeout
      
    68 . eststo: reghdfe bold post##treat, absorb(firmbroker yearmonth) vce(cluster fi
      > rm) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: 1bn.post is probably collinear with the fixed effects (all partialled-out
      >  values are close to zero; tol = 1.0e-09)
      note: 1bn.treat is probably collinear with the fixed effects (all partialled-ou
      > t values are close to zero; tol = 1.0e-09)
      
      HDFE Linear regression                            Number of obs   =     22,527
      Absorbing 2 HDFE groups                           F(   1,   2785) =       7.84
      Statistics robust to heteroskedasticity           Prob > F        =     0.0051
                                                        R-squared       =     0.4352
                                                        Adj R-squared   =    -0.0113
                                                        Within R-sq.    =     0.0005
      Number of clusters (firm)    =      2,786         Root MSE        =     0.4498
      
                                     (Std. err. adjusted for 2,786 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
              bold | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
            1.post |          0  (omitted)
           1.treat |          0  (omitted)
                   |
        post#treat |
              1 1  |   .0557069   .0198915     2.80   0.005     .0167033    .0947104
                   |
             _cons |   .7201937   .0011444   629.33   0.000     .7179498    .7224376
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |      9943        9943           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est2 stored)
      
    69 . outreg2 using 2.doc, nolabel tstat tdec(2) dec(3) append
      2.doc
      dir : seeout
      
    70 . eststo: reghdfe first post##treat, absorb(firmbroker yearmonth) vce(cluster f
      > irm) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: 1bn.post is probably collinear with the fixed effects (all partialled-out
      >  values are close to zero; tol = 1.0e-09)
      note: 1bn.treat is probably collinear with the fixed effects (all partialled-ou
      > t values are close to zero; tol = 1.0e-09)
      
      HDFE Linear regression                            Number of obs   =     24,446
      Absorbing 2 HDFE groups                           F(   1,   2849) =       9.19
      Statistics robust to heteroskedasticity           Prob > F        =     0.0024
                                                        R-squared       =     0.5688
                                                        Adj R-squared   =     0.2518
                                                        Within R-sq.    =     0.0004
      Number of clusters (firm)    =      2,850         Root MSE        =     0.1412
      
                                     (Std. err. adjusted for 2,850 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
             first | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
            1.post |          0  (omitted)
           1.treat |          0  (omitted)
                   |
        post#treat |
              1 1  |   .0155784   .0051377     3.03   0.002     .0055045    .0256524
                   |
             _cons |   .0265636   .0002783    95.46   0.000      .026018    .0271092
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |     10353       10353           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est3 stored)
      
    71 . outreg2 using 2.doc, nolabel tstat tdec(2) dec(3) append
      2.doc
      dir : seeout
      
    72 . eststo: reghdfe indfirst post##treat, absorb(firmbroker yearmonth) vce(cluste
      > r firm) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: 1bn.post is probably collinear with the fixed effects (all partialled-out
      >  values are close to zero; tol = 1.0e-09)
      note: 1bn.treat is probably collinear with the fixed effects (all partialled-ou
      > t values are close to zero; tol = 1.0e-09)
      
      HDFE Linear regression                            Number of obs   =     24,446
      Absorbing 2 HDFE groups                           F(   1,   2849) =       5.03
      Statistics robust to heteroskedasticity           Prob > F        =     0.0250
                                                        R-squared       =     0.5528
                                                        Adj R-squared   =     0.2241
                                                        Within R-sq.    =     0.0003
      Number of clusters (firm)    =      2,850         Root MSE        =     0.0941
      
                                     (Std. err. adjusted for 2,850 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
          indfirst | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
            1.post |          0  (omitted)
           1.treat |          0  (omitted)
                   |
        post#treat |
              1 1  |   .0083945   .0037438     2.24   0.025     .0010537    .0157354
                   |
             _cons |    .011081   .0002028    54.65   0.000     .0106834    .0114786
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |     10353       10353           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est4 stored)
      
    73 . outreg2 using 2.doc, nolabel tstat tdec(2) dec(3) append
      2.doc
      dir : seeout
      
    74 . noisily: esttab, r2 star(* 0.1 ** 0.05 *** 0.01)  nonote b(%9.3f) keep(1.post
      > #1.treat) 
      
      ----------------------------------------------------------------------------
                            (1)             (2)             (3)             (4)   
                           accu            bold           first        indfirst   
      ----------------------------------------------------------------------------
      1.post#1.t~t        0.003*          0.056***        0.016***        0.008** 
                         (1.76)          (2.80)          (3.03)          (2.24)   
      ----------------------------------------------------------------------------
      N                   24446           22527           24446           24446   
      R-sq                0.888           0.435           0.569           0.553   
      ----------------------------------------------------------------------------
      
    75 . /*Figure 3 Panels B through E*/
    76 . gen zero = 0
      
    77 . gen r_4=0
      
    78 . replace r_4=1 if month==4 & treat==1 & year==2024
      (639 real changes made)
      
    79 . gen r_3=0
      
    80 . replace r_3=1 if month==5 & treat==1 & year==2024
      (730 real changes made)
      
    81 . gen r_2=0
      
    82 . replace r_2=1 if month==10 & treat==1 & year==2024
      (737 real changes made)
      
    83 . gen r_1=0
      
    84 . replace r_1=1 if month==11 & treat==1 & year==2024
      (587 real changes made)
      
    85 . local running "r_4 zero r_2 r_1 post treat"
      
    86 . eststo: reghdfe accu `running', absorb(firmbroker yearmonth) vce(cluster firm
      > ) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: post is probably collinear with the fixed effects (all partialled-out val
      > ues are close to zero; tol = 1.0e-09)
      note: treat is probably collinear with the fixed effects (all partialled-out va
      > lues are close to zero; tol = 1.0e-09)
      note: zero omitted because of collinearity
      
      HDFE Linear regression                            Number of obs   =     24,446
      Absorbing 2 HDFE groups                           F(   3,   2849) =       1.22
      Statistics robust to heteroskedasticity           Prob > F        =     0.3012
                                                        R-squared       =     0.8879
                                                        Adj R-squared   =     0.8055
                                                        Within R-sq.    =     0.0003
      Number of clusters (firm)    =      2,850         Root MSE        =     0.0408
      
                                     (Std. err. adjusted for 2,850 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
              accu | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
               r_4 |   .0016732   .0023293     0.72   0.473     -.002894    .0062404
              zero |          0  (omitted)
               r_2 |   .0042538   .0025444     1.67   0.095    -.0007351    .0092428
               r_1 |    .004213   .0030278     1.39   0.164    -.0017239      .01015
              post |          0  (omitted)
             treat |          0  (omitted)
             _cons |   -.028979    .000177  -163.74   0.000    -.0293261    -.028632
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |     10353       10353           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est5 stored)
      
    87 . coefplot, level(90) omitted ciopts(recast(rcap)) vertical yline(0, lcolor(bla
      > ck)) xline(2.5) coeflabels(r_4 = "Apr" zero = "May" r_2 = "Oct" r_1 = "Nov") 
      > keep(r_4 zero r_2 r_1)  recast(connected)  lpattern(dash) lcolor(gray) 
      
    88 . local running "r_4 zero r_2 r_1 post treat"
      
    89 . eststo: reghdfe bold `running', absorb(firmbroker yearmonth) vce(cluster firm
      > ) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: post is probably collinear with the fixed effects (all partialled-out val
      > ues are close to zero; tol = 1.0e-09)
      note: treat is probably collinear with the fixed effects (all partialled-out va
      > lues are close to zero; tol = 1.0e-09)
      note: zero omitted because of collinearity
      
      HDFE Linear regression                            Number of obs   =     22,527
      Absorbing 2 HDFE groups                           F(   3,   2785) =       3.88
      Statistics robust to heteroskedasticity           Prob > F        =     0.0088
                                                        R-squared       =     0.4354
                                                        Adj R-squared   =    -0.0112
                                                        Within R-sq.    =     0.0008
      Number of clusters (firm)    =      2,786         Root MSE        =     0.4498
      
                                     (Std. err. adjusted for 2,786 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
              bold | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
               r_4 |   .0135831   .0356512     0.38   0.703    -.0563224    .0834885
              zero |          0  (omitted)
               r_2 |      .0916   .0312913     2.93   0.003     .0302435    .1529564
               r_1 |   .0267587   .0283699     0.94   0.346    -.0288695    .0823868
              post |          0  (omitted)
             treat |          0  (omitted)
             _cons |   .7194446   .0020929   343.75   0.000     .7153408    .7235485
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |      9943        9943           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est6 stored)
      
    90 . coefplot, level(90) omitted ciopts(recast(rcap)) vertical yline(0, lcolor(bla
      > ck)) xline(2.5) coeflabels(r_4 = "Apr" zero = "May" r_2 = "Oct" r_1 = "Nov") 
      > keep(r_4 zero r_2 r_1)  recast(connected)  lpattern(dash) lcolor(gray) 
      
    91 . local running "r_4 zero r_2 r_1 post treat"
      
    92 . eststo: reghdfe first `running', absorb(firmbroker yearmonth) vce(cluster fir
      > m) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: post is probably collinear with the fixed effects (all partialled-out val
      > ues are close to zero; tol = 1.0e-09)
      note: treat is probably collinear with the fixed effects (all partialled-out va
      > lues are close to zero; tol = 1.0e-09)
      note: zero omitted because of collinearity
      
      HDFE Linear regression                            Number of obs   =     24,446
      Absorbing 2 HDFE groups                           F(   3,   2849) =       6.73
      Statistics robust to heteroskedasticity           Prob > F        =     0.0002
                                                        R-squared       =     0.5690
                                                        Adj R-squared   =     0.2520
                                                        Within R-sq.    =     0.0009
      Number of clusters (firm)    =      2,850         Root MSE        =     0.1412
      
                                     (Std. err. adjusted for 2,850 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
             first | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
               r_4 |  -.0102897   .0084976    -1.21   0.226    -.0269518    .0063725
              zero |          0  (omitted)
               r_2 |  -.0016954    .009153    -0.19   0.853    -.0196425    .0162518
               r_1 |   .0254461   .0077806     3.27   0.001       .01019    .0407023
              post |          0  (omitted)
             treat |          0  (omitted)
             _cons |   .0271164   .0005821    46.58   0.000     .0259749    .0282579
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |     10353       10353           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est7 stored)
      
    93 . coefplot, level(90) omitted ciopts(recast(rcap)) vertical yline(0, lcolor(bla
      > ck)) xline(2.5) coeflabels(r_4 = "Apr" zero = "May" r_2 = "Oct" r_1 = "Nov") 
      > keep(r_4 zero r_2 r_1)  recast(connected)  lpattern(dash) lcolor(gray) 
      
    94 . local running "r_4 zero r_2 r_1 post treat"
      
    95 . eststo: reghdfe indfirst `running', absorb(firmbroker yearmonth) vce(cluster 
      > firm) keepsingleton
      WARNING: Singleton observations not dropped; statistical significance is biased
      >  (link)
      (MWFE estimator converged in 5 iterations)
      note: post is probably collinear with the fixed effects (all partialled-out val
      > ues are close to zero; tol = 1.0e-09)
      note: treat is probably collinear with the fixed effects (all partialled-out va
      > lues are close to zero; tol = 1.0e-09)
      note: zero omitted because of collinearity
      
      HDFE Linear regression                            Number of obs   =     24,446
      Absorbing 2 HDFE groups                           F(   3,   2849) =       4.84
      Statistics robust to heteroskedasticity           Prob > F        =     0.0023
                                                        R-squared       =     0.5530
                                                        Adj R-squared   =     0.2242
                                                        Within R-sq.    =     0.0006
      Number of clusters (firm)    =      2,850         Root MSE        =     0.0941
      
                                     (Std. err. adjusted for 2,850 clusters in firm)
      ------------------------------------------------------------------------------
                   |               Robust
          indfirst | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
      -------------+----------------------------------------------------------------
               r_4 |  -.0041863   .0061595    -0.68   0.497    -.0162638    .0078913
              zero |          0  (omitted)
               r_2 |  -.0009291   .0065989    -0.14   0.888    -.0138683    .0120101
               r_1 |   .0151764   .0047678     3.18   0.001     .0058278    .0245251
              post |          0  (omitted)
             treat |          0  (omitted)
             _cons |   .0113086   .0003952    28.62   0.000     .0105338    .0120835
      ------------------------------------------------------------------------------
      
      Absorbed degrees of freedom:
      -----------------------------------------------------+
       Absorbed FE | Categories  - Redundant  = Num. Coefs |
      -------------+---------------------------------------|
        firmbroker |     10353       10353           0    *|
         yearmonth |         4           1           3     |
      -----------------------------------------------------+
      * = FE nested within cluster; treated as redundant for DoF computation
      (est8 stored)
      
    96 . coefplot, level(90) omitted ciopts(recast(rcap)) vertical yline(0, lcolor(bla
      > ck)) xline(2.5) coeflabels(r_4 = "Apr" zero = "May" r_2 = "Oct" r_1 = "Nov") 
      > keep(r_4 zero r_2 r_1)  recast(connected)  lpattern(dash) lcolor(gray) 
      
    97 . 
      end of do-file
      
    98 . log close
            name:  <unnamed>
             log:  C:\Users\isyoo\OneDrive - Singapore Management University\Desktop\
      > Work\JAR AI Aanalysts\JAR Replication Package\Log flies\Askresearchgpt analys
      > es.smcl
        log type:  smcl
       closed on:  19 Jul 2026, 08:56:40
      -------------------------------------------------------------------------------
