Political Incentives to Suppress Negative Information: Evidence from

Political Incentives to Suppress Negative Information:
Evidence from Chinese Listed Firms
Internet Appendix
Joseph D. Piotroski
Stanford University
T.J. Wong
The Chinese University of Hong Kong
Tianyu Zhang
The Chinese University of Hong Kong
November 2014
Table 4A
Influence of political events on the incentive to suppress negative financial information
Inclusion of both National Congresses and Political Promotions Event Indicators in same
estimation (Footnote 15)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. Political is an indicator variable equal to one if the firm-year relates to a specific
political event; Post-Political is an indicator variable equal to one in the year immediately following the event. In
the first set of estimations, Political is an indicator variable equal to one for the years that a National Congress of the
Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following each National Congress event, zero otherwise. In the second set of estimations, Political is an
indicator variable equal to one for the year preceding and corresponding to a provincial level political promotion
event, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following the
provincial-level promotion event, zero otherwise. In the last set of estimations, indicators variables for both political
events are included. All other variables are defined in Table 1 and Appendix A. T-statistics derived using clustered
standard errors by province are presented in parentheses. Models include annual and provincial fixed effects [Year
and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the estimated coefficient
is significantly different than zero at the one, five and ten percent level (two-tailed test), respectively. N=12,723
National Congress
(1)
(2)
Provincial-Level Promotion
(1)
(2)
Both Political Events
(1)
(2)
-0.241***
(-8.37)
0.054**
(2.25)
-0.043***
(-7.31)
-0.019
(-1.68)
-9.432***
(-4.95)
0.927***
(7.36)
-0.416***
(-4.12)
-0.275***
(-8.90)
-0.037***
(-3.58)
0.078***
(10.39)
-
-0.228***
(-7.61)
0.058**
(2.34)
-0.041***
(-6.70)
-0.019
(-1.68)
-9.554***
(-4.94)
0.924***
(7.36)
-0.434***
(-4.19)
-0.271***
(-8.70)
-0.038***
(-3.70)
0.077***
(10.28)
-0.030*
(-2.00)
-0.058**
(-2.10)
0.122***
(3.75)
-0.055***
(-6.46)
-0.013
(-1.09)
-18.744***
(-7.89)
0.397***
(2.79)
-0.158
(-0.87)
-0.296***
(-10.88)
0.020
(1.62)
0.076***
(8.42)
-
-0.058**
(-2.12)
0.121***
(3.75)
-0.052***
(-5.77)
-0.013
(-1.08)
-19.006***
(-7.84)
0.396***
(2.78)
-0.176
(-0.96)
-0.291***
(-10.67)
0.019
(1.53)
0.075***
(8.15)
-0.037**
(-2.26)
-0.171***
(-3.76)
0.064**
(2.26)
-0.061**
(-2.42)
0.118***
(3.64)
-0.051***
(-6.21)
-0.007
(-0.54)
-15.769***
(-8.42)
0.346**
(2.71)
-0.008
(-0.05)
-0.310***
(-10.74)
0.009
(0.81)
0.068***
(7.58)
-
-0.169***
(-3.57)
0.062**
(2.18)
-0.062**
(-2.46)
0.118***
(3.66)
-0.047***
(-5.38)
-0.007
(-0.56)
-16.049***
(-8.39)
0.338**
(2.66)
-0.019
(-0.14)
-0.304***
(-10.47)
0.008
(0.71)
0.066***
(7.33)
-0.043***
(-2.82)
Year Fixed Effects
Provincial Fixed
Effects
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Adjusted R-squared
0.174
0.174
0.125
0.126
0.180
0.181
Congresst
Post-Congresst
Promotiont
Post-Promotiont
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
SOEt
Table 4B
Influence of political events on the incentive to suppress negative financial information
Inclusion of Central Government-owned Firms in Pooled Sample (Footnote 5)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. Political is an indicator variable equal to one if the firm-year relates to a specific
political event; Post-Political is an indicator variable equal to one in the year immediately following the event. In
the first set of estimations, Political is an indicator variable equal to one for the years that a National Congress of the
Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following each National Congress event, zero otherwise. In the second set of estimations, Political is an
indicator variable equal to one for the year preceding and corresponding to a provincial level political promotion
event, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following the
provincial-level promotion event, zero otherwise. All other variables are defined in Table 1 and Appendix A. Tstatistics derived using clustered standard errors by province are presented in parentheses. Models include annual
and provincial fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,*
indicate that the estimated coefficient is significantly different than zero at the one, five and ten percent level (twotailed test), respectively. N=15,450
Political Event:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
National Congress
(2)
-0.240***
(-7.54)
0.049*
(2.02)
-0.045***
(-8.34)
-0.011
(-1.05)
-9.115***
(-4.50)
0.984***
(7.44)
-0.471***
(-4.74)
-0.258***
(-8.71)
-0.038***
(-4.14)
0.082***
(11.70)
-0.231***
(-6.86)
0.051**
(2.06)
-0.043***
(-8.02)
-0.011
(-1.03)
-9.151***
(-4.49)
0.980***
(7.46)
-0.483***
(-4.72)
-0.254***
(-8.46)
-0.039***
(-4.29)
0.081***
(11.65)
-0.023
(-1.57)
-0.062**
(-2.39)
0.126***
(3.75)
-0.054***
(-7.73)
-0.008
(-0.70)
-18.631***
(-7.69)
0.412***
(2.82)
-0.106
(-0.63)
-0.282***
(-10.78)
0.019*
(1.87)
0.083***
(10.43)
-0.062**
(-2.40)
0.125***
(3.76)
-0.051***
(-7.16)
-0.007
(-0.67)
-18.753***
(-7.66)
0.408***
(2.80)
-0.119
(-0.70)
-0.278***
(-10.48)
0.018*
(1.79)
0.082***
(10.19)
-0.030*
(-1.89)
Included
Included
Included
Included
Included
Included
Included
Included
0.173
0.173
0.126
0.126
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
(1)
(2)
(1)
Table 4C
Influence of political events on the incentive to suppress negative financial information
Use of Duvol (“down-to-up” volatility) as alternative measure of stock price crashes
(Section 6.1)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Duvoli,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Duvol is the firm’s “down-to-up” volatility, measured as the log of the ratio of the standard
deviation of residual returns on down days to the log of the standard deviation of residual returns on up days.
Political is an indicator variable equal to one if the firm-year relates to a specific political event; Post-Political is an
indicator variable equal to one in the year immediately following the event. In the first set of estimations, Political
is an indicator variable equal to one for the years that a National Congress of the Chinese Communist Party was
held, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following each National
Congress event, zero otherwise. In the second set of estimations, Political is an indicator variable equal to one for
the year preceding and corresponding to a provincial level political promotion event, zero otherwise; Post-Political
is an indicator variable equal to one for the year directly following the provincial-level promotion event, zero
otherwise. All other variables are defined in Table 1 and Appendix A. T-statistics derived using clustered standard
errors by province are presented in parentheses. Models include annual and provincial fixed effects [Year and
Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the estimated coefficient is
significantly different than zero at the one, five and ten percent level (two-tailed test), respectively. N=12,723
Political Event:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
National Congress
(2)
-0.148***
(-11.28)
0.053***
(4.05)
-0.014***
(-4.89)
-0.007*
(-1.71)
-2.660***
(-3.04)
0.239***
(4.24)
-0.151***
(-3.01)
-0.168***
(-14.30)
-0.040***
(-10.97)
0.018***
(5.78)
-0.138***
(-9.80)
0.057***
(4.16)
-0.012***
(-3.96)
-0.007
(-1.69)
-2.772***
(-3.11)
0.238***
(4.24)
-0.158***
(-3.13)
-0.165***
(-13.79)
-0.041***
(-11.12)
0.017***
(5.49)
-0.020***
(-3.22)
-0.028**
(-2.38)
0.045***
(3.50)
-0.025***
(-6.45)
-0.006
(-1.28)
-4.613***
(-4.23)
-0.045
(-0.69)
-0.062
(-0.77)
-0.174***
(-17.55)
-0.018***
(-4.08)
0.019***
(5.17)
-0.028**
(-2.40)
0.045***
(3.50)
-0.023***
(-5.81)
-0.006
(-1.27)
-4.737***
(-4.26)
-0.046
(-0.70)
-0.070
(-0.87)
-0.172***
(-16.98)
-0.019***
(-4.24)
0.018***
(4.90)
-0.017**
(-2.75)
Included
Included
Included
Included
Included
Included
Included
Included
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
(1)
(2)
(1)
Table 4D
Influence of political events on the incentive to suppress negative financial information
Use of Fraction (Percentage of weeks with large negative stock price drop) as alternative
measure of stock price crashes (Section 6.1)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Fractioni,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Fraction is the percentage of weeks the firm experienced a large negative stock price drop in
year t. A large negative stock price drop is measured as a negative return more than 1.96 standard deviations below
the sample mean or in excess of -20%, respectively. Political is an indicator variable equal to one if the firm-year
relates to a specific political event; Post-Political is an indicator variable equal to one in the year immediately
following the event. In the first set of estimations, Political is an indicator variable equal to one for the years that a
National Congress of the Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable
equal to one for the year directly following each National Congress event, zero otherwise. In the second set of
estimations, Political is an indicator variable equal to one for the year preceding and corresponding to a provincial
level political promotion event, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following the provincial-level promotion event, zero otherwise. All other variables are defined in Table 1
and Appendix A. T-statistics derived using clustered standard errors by province are presented in parentheses.
Models include annual and provincial fixed effects [Year and Province, respectively; coefficients not reported]. The
superscripts ***,**,* indicate that the estimated coefficient is significantly different than zero at the one, five and ten
percent level (two-tailed test), respectively. N=12,723
Political Event:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
National Congress
> 1.96 Std. Dev.
> 20% Price
Price Decline
Decline
Provincial-Level Political Promotion
> 1.96 Std. Dev.
> 20% Price
Price Decline
Decline
-0.005***
(-7.37)
0.005***
(7.29)
-0.001***
(-5.61)
-0.001**
(-2.58)
-0.267***
(-8.25)
0.006
(1.54)
-0.014***
(-3.99)
-0.001
(-0.79)
-0.004***
(-14.52)
-0.000
(-0.16)
-0.001
(-1.41)
-0.001***
(-5.15)
-0.000
(-0.75)
-0.000
(-0.37)
-0.000
(-1.30)
0.305***
(14.65)
0.004*
(1.91)
-0.007***
(-5.95)
-0.002***
(-6.96)
-0.000
(-1.18)
0.000
(1.37)
0.000
(0.94)
-0.001
(-1.45)
0.001**
(2.46)
-0.001***
(-6.57)
-0.001**
(-2.56)
-0.253***
(-6.60)
-0.003
(-0.77)
-0.007*
(-1.91)
-0.000
(-0.07)
-0.004***
(-14.51)
0.000
(0.03)
-0.001
(-1.36)
-0.000**
(-2.10)
0.000*
(1.77)
0.000
(0.36)
-0.000
(-1.35)
0.332***
(14.25)
0.002
(0.85)
-0.004***
(-3.69)
-0.002***
(-6.86)
-0.000
(-1.25)
0.000
(1.21)
0.000
(1.13)
Included
Included
Included
Included
Included
Included
Included
Included
0.114
0.140
0.108
0.143
Table 4E
Influence of political events on the incentive to suppress negative financial information
Robustness using non-parametric measure of skewness, defined as the ratio of the
difference between the firm’s 95th percentile and median return realizations to the
difference between median and 5th percentile realization in year t. (Section 6.1)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
NP Skewi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable NP Skew is the measured as the ratio of the difference in the firm’s 95 th percentile and
median excess return realization to the difference in the median and 5 th percentile realization in year t. As defined,
NP Skew is decreasing in negative skewness. Political is an indicator variable equal to one if the firm-year relates to
a specific political event; Post-Political is an indicator variable equal to one in the year immediately following the
event. In the first set of estimations, Political is an indicator variable equal to one for the years that a National
Congress of the Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to
one for the year directly following each National Congress event, zero otherwise. In the second set of estimations,
Political is an indicator variable equal to one for the year preceding and corresponding to a provincial level political
promotion event, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following
the provincial-level promotion event, zero otherwise. All other variables are defined in Table 1 and Appendix A. Tstatistics derived using clustered standard errors by province are presented in parentheses. Models include annual
and provincial fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,*
indicate that the estimated coefficient is significantly different than zero at the one, five and ten percent level (twotailed test), respectively. N=12,723
Political Event:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
National Congress
(2)
0.326***
(14.49)
-0.028***
(-3.41)
0.032***
(8.95)
0.013**
(2.23)
4.243***
(6.96)
0.053
(0.80)
0.095*
(1.77)
0.202***
(18.76)
0.077***
(16.25)
0.003
(0.95)
0.315***
(13.51)
-0.032***
(-3.82)
0.030***
(8.33)
0.013**
(2.20)
4.363***
(6.96)
0.055
(0.82)
0.103*
(1.90)
0.199***
(18.43)
0.078***
(16.50)
0.004
(1.23)
0.021***
(3.18)
0.018*
(1.87)
-0.031***
(-2.96)
0.040***
(9.78)
0.012*
(1.96)
5.983***
(7.89)
0.230***
(3.39)
0.072
(0.91)
0.207***
(19.44)
0.061***
(13.14)
0.002
(0.57)
0.018*
(1.89)
-0.031***
(-2.96)
0.039***
(9.27)
0.012*
(1.94)
6.123***
(7.83)
0.231***
(3.40)
0.082
(1.02)
0.205***
(19.26)
0.062***
(13.48)
0.003
(0.82)
0.020***
(2.93)
Included
Included
Included
Included
Included
Included
Included
Included
0.295
0.296
0.271
0.272
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
(1)
(2)
(1)
Table 4F
Influence of political events on the incentive to suppress negative financial information
Robustness Tests (outlined in Section 6.5)
Specification:
Political Event:
Politicalt
Post-Politicalt
Without Control
Variables except Year
and Province Effects
National
Political
Congress Promotion
-0.139***
(-4.40)
0.158***
(5.86)
-0.071**
(-2.45)
0.139***
(3.98)
LogSizet
Growtht
Sigmat
Using Lagged
Control Variables
National
Political
Congress
Promotion
Incremental Control for
Accounting Performance
(ROA)
National
Political
Congress
Promotion
-0.092**
(-2.23)
0.169***
(4.80
-0.031***
(-4.78)
-0.022*
(-1.97)
-0.058**
(-2.13)
0.121***
(3.81)
-0.050***
(-6.28)
-0.009
(-0.81)
-0.098**
(-2.47)
0.190***
(5.62)
-0.011
(-1.65)
-0.001
(-0.10)
-0.057**
(-2.11)
0.121***
(3.80)
-0.038***
(-4.50)
0.016
(1.33)
-8.974***
(-5.36)
-16.225***
(-8.50)
-12.956***
(-6.94)
0.999***
(8.21)
-0.147
(-1.25)
-0.269***
(-8.55)
-0.029***
(-3.22)
0.076***
(10.92)
-0.044***
(-2.81)
-0.394***
(-3.92)
-21.564***
(-9.48)
0.488***
(3.56)
-0.238
(-1.45)
-0.287***
(-10.62)
0.035***
(2.93)
0.086***
(9.83)
-0.042**
(-2.58)
-0.680***
(-6.84)
Turnovert
Turnovert-1
0.150
(1.15)
-0.228***
(-7.79)
Betat
-0.027
(-0.16)
-0.279***
(-11.14)
Returnt
Returnt-1
0.064***
(9.04)
-0.041**
(-2.60)
SOEt
0.068***
(7.93)
-0.038**
(-2.31)
ROA
Without Controlling for
Firm Risk Variables (Beta
and Sigma)
National
Political
Congress
Promotion
-0.105**
(-2.72)
0.145***
(4.82)
-0.016**
(-2.08)
-0.003
(-0.29)
-0.064**
(-2.27)
0.132***
(3.85)
-0.042***
(-4.52)
0.005
(0.49)
0.188*
(1.95)
-0.335**
(-2.54)
-0.802***
(-6.37)
-0.086
(-0.51)
-0.063***
(-6.07)
0.055***
(8.33)
-0.044***
(-2.86)
-0.018
(-1.69)
0.052***
(6.62)
-0.035**
(-2.28)
Market-to-Book
Including Market-toBook in lieu of Sales
Growth
National
Political
Congress
Promotion
-0.095**
(-2.32)
0.213***
(6.42)
-0.004
(-0.51)
-0.061**
(-2.20)
0.124***
(3.89)
-0.041***
(-4.13)
14.460***
(-7.69)
1.047***
(8.86)
-0.147
(-1.18)
-0.150***
(-4.43)
-0.061***
(-6.86)
0.056***
(7.92)
-0.027
(-1.68)
23.359***
(-10.28)
0.514***
(3.67)
-0.272
(-1.49)
-0.185***
(-5.80)
0.004
(0.38)
0.066***
(7.46)
-0.024
(-1.45)
0.017***
(7.68)
0.016***
(6.85)
Year Fix. Effect
Prov. Fix. Effect
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Adj. R-squared
0.163
0.0987
0.177
0.125
0.182
0.129
0.169
0.107
0.187
0.130
***,**,*
indicate that the estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test), respectively. N =12,723.
Table 4G
Influence of political events on the incentive to suppress negative financial information
Use of firm fixed effects (section 6.5)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. Political is an indicator variable equal to one if the firm-year relates to a specific
political event; Post-Political is an indicator variable equal to one in the year immediately following the event. In
the first set of estimations, Political is an indicator variable equal to one for the years that a National Congress of the
Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following each National Congress event, zero otherwise. In the second set of estimations, Political is an
indicator variable equal to one for the year preceding and corresponding to a provincial level political promotion
event, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following the
provincial-level promotion event, zero otherwise. In the last set of estimations, indicators variables for both political
events are included. All other variables are defined in Table 1 and Appendix A. T-statistics derived using clustered
standard errors by province are presented in parentheses. Models include annual and provincial fixed effects [Year
and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the estimated coefficient
is significantly different than zero at the one, five and ten percent level (two-tailed test), respectively. N=12,723
National Congress
(1)
(2)
Congresst
Provincial-Level Promotion
(1)
(2)
Both Political Events
(1)
(2)
-0.258***
(-6.15)
0.058**
(2.19)
-0.046***
(-4.03)
-0.010
(-0.82)
-13.533***
(-7.47)
1.249***
(7.71)
-0.284**
(-2.38)
-0.268***
(-7.88)
-0.050***
(-4.64)
0.059***
(6.65)
-
-0.258***
(-6.15)
0.058**
(2.19)
-0.046***
(-4.03)
-0.010
(-0.82)
-13.533***
(-7.47)
1.249***
(7.71)
-0.284**
(-2.38)
-0.268***
(-7.88)
-0.050***
(-4.64)
0.059***
(6.65)
0.045
(1.61)
-0.050*
(-1.95)
0.116***
(3.52)
-0.007
(-0.44)
-0.003
(-0.21)
-28.123***
(-12.50)
0.727***
(4.39)
-0.350
(-1.68)
-0.299***
(-9.25)
0.006
(0.50)
0.054***
(5.35)
-
-0.050*
(-1.92)
0.117***
(3.56)
-0.008
(-0.49)
-0.002
(-0.18)
-28.115***
(-12.61)
0.735***
(4.44)
-0.350
(-1.68)
-0.302***
(-9.35)
0.006
(0.54)
0.055***
(5.44)
0.067**
(2.11)
-0.180***
(-4.42)
0.062**
(2.16)
-0.063**
(-2.59)
0.114***
(3.52)
-0.018
(-1.07)
0.001
(0.04)
-21.345***
(-12.24)
0.569***
(3.37)
0.102
(0.70)
-0.317***
(-9.49)
-0.006
(-0.58)
0.045***
(4.30)
-
-0.185***
(-4.66)
0.063**
(2.17)
-0.063**
(-2.58)
0.114***
(3.53)
-0.018
(-1.09)
0.001
(0.06)
-21.302***
(-12.23)
0.572***
(3.38)
0.102
(0.70)
-0.318***
(-9.48)
-0.006
(-0.57)
0.045***
(4.31)
0.023
(0.82)
Year Fixed Effects
Firm Fixed Effects
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Adjusted R-squared
0.177
0.177
0.129
0.129
0.120
0.120
Post-Congresst
Promotiont
Post-Promotiont
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
SOEt
Table 4H
Influence of political events on the incentive to suppress negative financial information
Using logged returns to estimate equation 1 and construct Ncskew (Section 6.5)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. In this table, Ncskew is estimated from residuals based on estimation of equation (1)
using logged returns. Political is an indicator variable equal to one if the firm-year relates to a specific political
event; Post-Political is an indicator variable equal to one in the year immediately following the event. In the first set
of estimations, Political is an indicator variable equal to one for the years that a National Congress of the Chinese
Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year directly
following each National Congress event, zero otherwise. In the second set of estimations, Political is an indicator
variable equal to one for the year preceding and corresponding to a provincial level political promotion event, zero
otherwise; Post-Political is an indicator variable equal to one for the year directly following the provincial-level
promotion event, zero otherwise. All other variables are defined in Table 1 and Appendix A. T-statistics derived
using clustered standard errors by provinces are presented in parentheses. Models include annual and provincial
fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the
estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test),
respectively. N=12,723
Political Event:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
National Congress
(2)
-0.159***
(-4.61)
0.126***
(4.00)
-0.009
(-1.52)
-0.000
(-0.04)
2.427
(1.63)
0.548***
(5.10)
-0.326***
(-2.90)
-0.243***
(-9.37)
-0.078***
(-14.01)
0.022***
(5.65)
-0.139***
(-3.72)
0.134***
(4.05)
-0.007
(-1.00)
-0.001
(-0.06)
2.147
(1.39)
0.546***
(5.11)
-0.342***
(-3.03)
-0.238***
(-9.13)
-0.078***
(-14.10)
0.022***
(5.52)
-0.037**
(-2.47)
-0.062**
(-2.39)
0.119***
(3.79)
-0.037***
(-5.00)
0.002
(0.18)
-1.705
(-0.99)
-0.247**
(-2.05)
-0.037
(-0.23)
-0.257***
(-11.27)
-0.023***
(-3.98)
0.020***
(5.34)
-0.062**
(-2.40)
0.118***
(3.80)
-0.034***
(-4.48)
0.002
(0.17)
-2.003
(-1.13)
-0.247*
(-2.04)
-0.055
(-0.35)
-0.253***
(-11.08)
-0.023***
(-3.97)
0.020***
(5.25)
-0.033**
(-2.19)
Included
Included
Included
Included
Included
Included
Included
Included
0.189
0.190
0.130
0.131
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
(1)
(2)
(1)
Table 4I
Influence of political events on the incentive to suppress negative financial information
Using standard errors clustered by province and political event period (footnote 11).
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. Political is an indicator variable equal to one if the firm-year relates to a specific
political event; Post-Political is an indicator variable equal to one in the year immediately following the event. In
the first set of estimations, Political is an indicator variable equal to one for the years that a National Congress of the
Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following each National Congress event, zero otherwise. In the second set of estimations, Political is an
indicator variable equal to one for the year preceding and corresponding to a provincial level political promotion
event, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following the
provincial-level promotion event, zero otherwise. All other variables are defined in Table 1 and Appendix A. Tstatistics derived using clustered standard errors by province and corresponding National Congress or political
promotion event period are presented in parentheses. This clustering procedure generates 93 and 91 clusters for our
National Congress and our political promotion estimations, respectively. Models include annual and provincial
fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the
estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test),
respectively. N=12,723
Political Event:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
National Congress
(2)
-0.241***
(-8.34)
0.054**
(2.25)
-0.043***
(-4.72)
-0.019
(-1.46)
-9.432***
(-4.82)
0.927***
(6.92)
-0.416***
(-3.40)
-0.275***
(-8.17)
-0.037***
(-3.22)
0.078***
(7.77)
-0.228***
(-7.08)
0.058**
(2.30)
-0.041***
(-4.29)
-0.019
(-1.46)
-9.554***
(-4.85)
0.924***
(6.88)
-0.434***
(-3.38)
-0.271***
(-7.94)
-0.038***
(-3.35)
0.077***
(7.61)
-0.030*
(-1.94)
-0.058*
(-1.98)
0.122***
(3.98)
-0.055***
(-4.35)
-0.013
(-1.12)
-18.744***
(-8.21)
0.397**
(2.23)
-0.158
(-1.00)
-0.296***
(-9.60)
0.020*
(1.74)
0.076***
(6.03)
-0.058*
(-1.98)
0.121***
(3.97)
-0.052***
(-3.84)
-0.013
(-1.12)
-19.006***
(-8.20)
0.396**
(2.22)
-0.176
(-1.10)
-0.291***
(-9.32)
0.019
(1.65)
0.075***
(5.83)
-0.037**
(-2.10)
Included
Included
Included
Included
Included
Included
Included
Included
0.174
0.174
0.125
0.126
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
(1)
(2)
(1)
Table 8A
Political incentives to suppress negative information conditional upon the presence of a
foreign stock exchange listing
Hong-Kong Listed Stocks + Top Quartile of Mainland-listed Firms by Size (Footnote 19)
The following panels present select coefficients from pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t +2Politicali,t*Hong Kongi,t + 3Post-Politicali,t
+ 4Post-Politicali,t*Hong Kongi,t + 5Hong Kongi,t-1 + 6LogSizei,t + 7Growthi,t + 8Sigmai,t + 9Turnoveri,t
+ 10Turnoveri,t-1 + 11Betai,t + 12Returni,t + 13Returni,t-1 + i,
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. Political is an indicator variable equal to one if the firm-year relates to a specific
political event; Post-Political is an indicator variable equal to one in the year immediately following the event. In
the first set of estimations, Political is an indicator variable equal to one for the years that a National Congress of the
Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following each National Congress event, zero otherwise. In the second set of estimations, Political is an
indicator variable equal to one for the year preceding and corresponding to a provincial level political promotion
event, zero otherwise; Post-Political is an indicator variable equal to one for the year directly following the
provincial-level promotion event, zero otherwise. For each event, the model is estimated using three samples: (1) a
reduced form of the empirical model is estimated using a small sample of Hong Kong-listed Chinese firms (first
column), (2) the full model is estimated using a pooled sample of Hong Kong-listed Chinese firms and (3) the full
model is estimated using a pooled sample of Hong Kong-listed Chinese firms and the top quartile of our sample of
mainland listed Chinese firms by firm size (third column). In the second and third estimations, Hong Kong is an
indicator variable equal to one if the firm-year relates to a Chinese firm with a Hong Kong cross-listing, zero
otherwise. All other variables are defined in Table 1 and Appendix A. T-statistics derived using clustered standard
errors by province are presented in parentheses. Models include annual and provincial fixed effects [Year and
Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the estimated coefficient is
significantly different than zero at the one, five and ten percent level (two-tailed test), respectively.
Political Event:
Fraction:
Politicalt
Hong Kong
Sample
National Congress
Hong Kong Hong Kong
+ Mainland
+ Top ¼
Sample
Mainland
Provincial-Level Political Promotion
Hong Kong Hong Kong + Hong Kong
Sample
Mainland
+ Top ¼
Sample
Mainland
-0.855***
(-3.04)
-0.352
(-1.22)
-0.241***
(-7.52)
0.027
(0.27)
0.050**
(2.53)
-0.360***
(-5.15)
0.141
(1.39)
0.066
(1.43)
-0.275***
(-3.27)
0.290*
(1.91)
-0.121***
(-4.21)
-0.183*
(-1.95)
0.125***
(3.99)
0.025
(0.86)
-0.095
(-0.92)
0.080**
(2.61)
-
0.220**
(2.45)
-0.064
(-0.98)
0.200**
(2.04)
-0.063
(-1.08)
-
0.158*
(1.84)
-0.026
(-0.37)
0.192*
(1.92)
-0.032
(-0.45)
Control Variables
Year Fixed Effects
Provincial Fixed Effects
Included
Included
Excluded
Included
Included
Included
Included
Included
Included
Included
Included
Excluded
Included
Included
Included
Included
Included
Included
Adjusted R-squared
Number of Observations
0.219
273
0.174
12,723
0.168
3,886
0.088
273
0.125
12,723
0.137
3,886
Politicalt*Hong Kongi,t
Post-Politicalt
Post-Politicalt*Hong
Kongi,t
Hong Kongi,t
Table 11A
Frequency of firm-specific news articles around Political Events: Short window / Monthly Data analyses (Footnote 24)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Log(1+Articlesi,t) or Log(1+Articlesi,t/Sizei,t) =  + Year + Province + 1Pre-Political[-6,-4] i,t + 2Pre-Political[-3,0] i,t + 3Post-Political[1,3] i,t
+ 4Post-Political[4,6] i,t + 5Logsizei,t +6Growthi,t + 7Leveragei,t-1 + 8Returni,t + 9SOEi,t + i,t
The dependent variable, Articlesi,t, is one of two measures capturing the number of news articles published about firm i in year t in either official or professional
newspapers. Log(1+Articlesi,t) is a logarithmic transformation of the number of articles published. Log(1+Articlesi,t/Sizei,t) is a logarithmic transformation of the
number of articles published scaled by the market value of the firm (in RMB billions). Pre-Political is an indicator variable equal to one if the firm-month
precedes a specific political event over the indicated range; zero otherwise. Post-Political is an indicator variable equal to one if the firm-month follows the
political event over the indicated range, zero otherwise. In the first set of estimations, Pre-Political and Post-Political relate to the months preceding and
following the month that the National Congress of the CCP was held. In the second set of estimations, Pre-Political and Post-Political relate to the months
surrounding a provincial level political promotion event. For each political event, the first (second) set of estimations reflects articles published in official
(professional/financial) newspapers. All other variables are defined in Table 1 and Appendix A. T-statistics derived using clustered standard errors by province
are presented in parentheses. Models include annual and provincial fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts
***,**,*
indicate that the estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test), respectively. N=195,259
Political Event
Type of Newspaper:
Dependent Variable:
National Congress
Official Newspaper
Professional Newspaper
Articlesi,t
Articlesi,t/Sizei,t
Articlesi,t
Articlesi,t/Sizei,t
Provincial-level Political Promotion
Official Newspaper
Professional Newspaper
Articlesi,t
Articlesi,t/Sizei,t
Articlesi,t
Articlesi,t/Sizei,t
Pre-Political [-6,-4]
-0.039***
(-4.90)
-0.015**
(-2.00)
0.020***
(2.92)
-0.014*
(-1.95)
0.311***
(23.69)
0.005
(0.31)
0.213***
(7.50)
-0.001**
(-2.36)
-0.039***
(-4.90)
-0.031***
(-6.86)
-0.013***
(-3.02)
0.007
(1.63)
-0.014***
(-2.75)
-0.021***
(-4.44)
0.004
(0.43)
0.047***
(2.91)
-0.001***
(-3.25)
-0.031***
(-6.86)
-0.115***
(-9.08)
0.007
(0.58)
0.077***
(6.52)
0.037***
(3.22)
0.633***
(45.65)
-0.038*
(-1.71)
0.226***
(5.25)
-0.006***
(-9.62)
-0.115***
(-9.08)
-0.103***
(-10.52)
-0.025***
(-2.62)
0.053***
(5.59)
0.033***
(3.28)
-0.098***
(-11.06)
-0.024
(-1.41)
0.086**
(2.40)
-0.004***
(-9.42)
-0.103***
(-10.52)
0.013*
(1.80)
-0.001
(-0.23)
0.000
(0.04)
0.001
(0.17)
0.311***
(23.70)
0.005
(0.31)
0.213***
(7.50)
-0.001***
(-3.00)
0.013*
(1.80)
0.001
(0.20)
-0.001
(-0.29)
0.003
(0.88)
0.003
(0.85)
-0.021***
(-4.44)
0.004
(0.43)
0.047***
(2.91)
-0.001***
(-3.78)
0.001
(0.20)
0.005
(0.40)
0.002
(0.18)
-0.001
(-0.13)
0.006
(0.52)
0.633***
(45.65)
-0.038*
(-1.70)
0.226***
(5.25)
-0.006***
(-10.58)
0.005
(0.41)
-0.003
(-0.36)
-0.006
(-0.79)
0.003
(0.31)
0.008
(0.92)
-0.098***
(-11.05)
-0.024
(-1.41)
0.086**
(2.40)
-0.005***
(-10.53)
-0.003
(-0.36)
Included
0.242
Included
0.0691
Included
0.301
Included
0.141
Included
0.242
Included
0.0689
Included
0.300
Included
0.141
Pre-Political [-3,-0]
Post-Political [1,3]
Post-Political [4,6]
LogSizet
Growtht
Leveraget
Returnt
SOE
Year & Province FE
Adjusted R-squared
Table A
Influence of political events on the incentive to provide “good news”: Stock Price Jumps
(Section 6.2)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Fraction_Jumpsi,t =  + Year + Province + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Growthi,t
+ 5Sigmai,t + 6Turnoveri,t + 7Turnoveri,t-1 + 8Betai,t + 9Returni,t + 10Returni,t-1 + 10SOEi,t + i,t
The dependent variable Fraction_Jumps is the percentage of weeks the firm experienced a large positive stock price
increase, or “jump,” in year t. A large positive stock price jump is measured as a positive return more than 1.96
standard deviations above the sample mean or more than 20%. Political is an indicator variable equal to one if the
firm-year relates to a specific political event; Post-Political is an indicator variable equal to one in the year
immediately following the event. In the first set of estimations, Political is an indicator variable equal to one for the
years that a National Congress of the Chinese Communist Party was held, zero otherwise; Post-Political is an
indicator variable equal to one for the year directly following each National Congress event, zero otherwise. In the
second set of estimations, Political is an indicator variable equal to one for the year preceding and corresponding to
a provincial level political promotion event, zero otherwise; Post-Political is an indicator variable equal to one for
the year directly following the provincial-level promotion event, zero otherwise. All other variables are defined in
Table 1 and Appendix A. T-statistics derived using clustered standard errors by province are presented in
parentheses. Models include annual and provincial fixed effects [Year and Province, respectively; coefficients not
reported]. The superscripts ***,**,* indicate that the estimated coefficient is significantly different than zero at the one,
five and ten percent level (two-tailed test), respectively. N=12,723
Political Event:
Fraction:
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
National Congress
> 1.96 Std. Dev.
> 20% Price
Price Gain
Gain
Provincial-Level Political Promotion
> 1.96 Std. Dev.
> 20% Price
Price Gain
Gain
-0.000
(-0.11)
-0.012***
(-5.22)
0.001***
(3.03)
0.000
(0.85)
0.098*
(2.01)
0.005
(1.36)
-0.002
(-0.41)
0.002***
(3.09)
0.003***
(8.27)
-0.000
(-0.28)
0.001
(1.39)
0.001
(0.56)
0.005***
(3.61)
0.000
(1.14)
0.000
(0.25)
0.817***
(33.36)
0.010***
(2.96)
-0.006**
(-2.69)
-0.003***
(-4.89)
0.003***
(12.34)
0.000
(0.35)
0.001***
(4.06)
0.000
(0.25)
0.000
(0.23)
0.001***
(3.01)
0.000
(0.84)
0.098*
(2.02)
0.005
(1.37)
-0.002
(-0.42)
0.002***
(3.08)
0.003***
(8.27)
-0.000
(-0.28)
0.001
(1.39)
-0.000
(-0.78)
-0.000
(-0.29)
0.000
(1.15)
0.000
(0.25)
0.817***
(33.29)
0.010***
(2.95)
-0.006**
(-2.65)
-0.003***
(-4.89)
0.003***
(12.32)
0.000
(0.35)
0.001***
(4.06)
Included
Included
Included
Included
Included
Included
Included
Included
0.0640
0.327
0.0639
0.327
Table B
Influence of political events on the incentive to defer release of “good news”: Stock Price
Jumps before, during and after political event periods (Section 6.2)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Fraction_Jumpsi,t =  + Year + Province + 1Pre-Political + 2Politicali,t + 3Post-Politicali,t + 4Logsizei,t
+ 5Growthi,t + 6Sigmai,t + 7Turnoveri,t + 8Turnoveri,t-1 + 9Betai,t + 10Returni,t +  11Returni,t-1 + 12SOEi,t + i,t
The dependent variable Fraction_Jumps is the percentage of weeks the firm experienced a large positive stock price
increase, or “jump,” in year t. A large positive stock price jump is measured as a positive return more than 1.96
standard deviations above the sample mean or more than 20%. Political is an indicator variable equal to one if the
firm-year relates to a specific political event; Post-Political is an indicator variable equal to one in the year
immediately following the event; Pre-Political is an indicator variable equal to one in the year directly preceding the
measurement of the political event. In the first set of estimations, Political is an indicator variable equal to one for
the years that a National Congress of the Chinese Communist Party was held, zero otherwise; Post-Political is an
indicator variable equal to one for the year directly following each National Congress event, zero otherwise. In the
second set of estimations, Political is an indicator variable equal to one for the year preceding and corresponding to
a provincial level political promotion event, zero otherwise; Post-Political is an indicator variable equal to one for
the year directly following the provincial-level promotion event, zero otherwise. All other variables are defined in
Table 1 and Appendix A. T-statistics derived using clustered standard errors by province are presented in
parentheses. Models include annual and provincial fixed effects [Year and Province, respectively; coefficients not
reported]. The superscripts ***,**,* indicate that the estimated coefficient is significantly different than zero at the one,
five and ten percent level (two-tailed test), respectively. N=12,723
Political Event:
Fraction:
Pre-Politicalt
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
National Congress
> 1.96 Std. Dev.
> 20% Price
Price Gain
Gain
Provincial-Level Political Promotion
> 1.96 Std. Dev.
> 20% Price
Price Gain
Gain
0.008***
(7.30)
-0.003**
(-2.56)
-0.006***
(-5.86)
0.000
(1.06)
0.000
(0.74)
0.036
(0.71)
-0.002
(-0.65)
0.005
(1.23)
0.003***
(4.07)
0.003***
(8.40)
-0.000
(-0.20)
0.001**
(2.11)
0.003***
(3.78)
-0.000
(-0.59)
0.004***
(11.05)
0.000*
(1.78)
0.000
(0.31)
0.817***
(32.95)
0.010***
(3.13)
-0.006**
(-2.49)
-0.003***
(-5.03)
0.003***
(12.47)
0.000
(0.33)
0.001***
(4.09)
-0.000
(-0.34)
0.000
(0.59)
-0.001
(-1.60)
0.001***
(3.00)
0.000
(0.83)
0.097*
(2.00)
0.005
(1.33)
-0.002
(-0.39)
0.002***
(3.07)
0.003***
(8.27)
-0.000
(-0.29)
0.001
(1.38)
-0.000
(-0.69)
-0.000
(-0.67)
-0.000
(-0.34)
0.000
(1.15)
0.000
(0.26)
0.817***
(33.37)
0.010***
(2.96)
-0.006**
(-2.65)
-0.003***
(-4.87)
0.003***
(12.31)
0.000
(0.36)
0.001***
(4.06)
Included
Included
Included
Included
Included
Included
Included
Included
0.0567
0.327
0.0640
0.327
Table C
Influence of political events on the incentive to defer release of “good news”: Evidence from
Ncskew before, during and after political event periods (Section 6.2)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Pre-Political + 2Politicali,t + 3Post-Politicali,t + 4Logsizei,t + 5Growthi,t
+ 6Sigmai,t + 7Turnoveri,t + 8Turnoveri,t-1 + 9Betai,t + 10Returni,t + 11Returni,t-1 + 12SOEi,t + i,t
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. A large negative stock price drop is measured as a negative return more than 1.96
standard deviations above the sample mean. Political is an indicator variable equal to one if the firm-year relates to
a specific political event; Post-Political is an indicator variable equal to one in the year immediately following the
event; Pre-Political is an indicator variable equal to one in the year directly preceding the measurement of the
political event. In the first set of estimations, Political is an indicator variable equal to one for the years that a
National Congress of the Chinese Communist Party was held, zero otherwise; Post-Political is an indicator variable
equal to one for the year directly following each National Congress event, zero otherwise. In the second set of
estimations, Political is an indicator variable equal to one for the year preceding and corresponding to a provincial
level political promotion event, zero otherwise; Post-Political is an indicator variable equal to one for the year
directly following the provincial-level promotion event, zero otherwise. All other variables are defined in Table 1
and Appendix A. T-statistics derived using clustered standard errors by province are presented in parentheses.
Models include annual and provincial fixed effects [Year and Province, respectively; coefficients not reported]. The
superscripts ***,**,* indicate that the estimated coefficient is significantly different than zero at the one, five and ten
percent level (two-tailed test), respectively. N=12,723
Political Event:
Pre-Political
Politicalt
Post-Politicalt
LogSizet
Growtht
Sigmat
Turnovert
Turnovert-1
Betat
Returnt
Returnt-1
National Congress
(2)
-0.807***
(-12.09)
-0.241***
(-8.37)
0.054**
(2.25)
-0.043***
(-7.31)
-0.019
(-1.68)
-9.432***
(-4.95)
0.927***
(7.36)
-0.416***
(-4.12)
-0.275***
(-8.90)
-0.037***
(-3.58)
0.078***
(10.39)
-0.794***
(-11.59)
-0.228***
(-7.61)
0.058**
(2.34)
-0.041***
(-6.71)
-0.019
(-1.68)
-9.554***
(-4.94)
0.924***
(7.36)
-0.434***
(-4.19)
-0.271***
(-8.70)
-0.038***
(-3.70)
0.077***
(10.28)
-0.030*
(-2.00)
-0.143***
(-3.75)
0.009
(0.35)
0.093**
(2.12)
-0.056***
(-6.86)
-0.013
(-1.00)
-18.814***
(-8.17)
0.410***
(2.76)
-0.170
(-0.95)
-0.298***
(-10.59)
0.023*
(2.03)
0.076***
(8.10)
-0.143***
(-3.76)
0.009
(0.35)
0.093**
(2.12)
-0.052***
(-6.03)
-0.013
(-0.99)
-19.082***
(-8.14)
0.409***
(2.75)
-0.188
(-1.04)
-0.293***
(-10.38)
0.021*
(1.93)
0.075***
(7.80)
-0.038**
(-2.31)
Included
Included
Included
Included
Included
Included
Included
Included
0.174
0.174
0.126
0.126
SOEt
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
(1)
(2)
(1)
Table D
Relation between political events and real economic activity (Section 6.3)
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
ROAi,t or Investmenti,t =  + Year + 1Province + 1Politicali,t + 1Post-Politicali,t + 2Logsizei,t + 3Growthi,t
+ 4Sigmai,t + 5Turnoveri,t + 6Turnoveri,t-1 + 7Betai,t + 8Returni,t + 9Returni,t-1 + i,t
The dependent variables ROAi,t and Investmenti,t is the firm’s return on assets realization (defined as net income
scaled by beginning assets) and investment rate (defined as capital expenditures scaled by beginning assets),
respectively, in year t. Political is an indicator variable equal to one if the firm-year relates to a specific political
event; Post-Political is an indicator variable equal to one in the year immediately following the event. In the first set
of estimations, Political is an indicator variable equal to one for the years that a National Congress of the Chinese
Communist Party was held, zero otherwise; Post-Political is an indicator variable equal to one for the year directly
following each National Congress event, zero otherwise. In the second set of estimations, Political is an indicator
variable equal to one for the year preceding and corresponding to a provincial level political promotion event, zero
otherwise; Post-Political is an indicator variable equal to one for the year directly following the provincial-level
promotion event, zero otherwise. All other variables are defined in Table 1 and Appendix A. T-statistics derived
using clustered standard errors by province are presented in parentheses. Models include annual and provincial
fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the
estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test),
respectively.
Panel A: Return on assets
National Congress
Without controlling
With controlling
SOE
SOE
Politicalt
Post-Politicalt
Control Variables
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
Without controlling
With controlling
SOE
SOE
-0.001
(-0.28)
-0.021***
(-6.95)
-0.001
(-0.18)
-0.022***
(-7.02)
-0.003
(-0.88)
0.000
(0.09)
-0.003
(-0.86)
0.000
(0.06)
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
0.362
0.365
0.354
0.358
Panel B: Investment Rate
National Congress
Without controlling
With controlling
SOE
SOE
Politicalt
Post-Politicalt
Control Variables
Year Fixed Effects
Provincial Fixed Effects
Adjusted R-squared
Provincial-Level Political Promotion
Without controlling
With controlling
SOE
SOE
-0.016**
(-2.43)
-0.005
(-1.11)
-0.016**
(-2.48)
-0.004
(-1.06)
-0.006***
(-2.92)
0.003
(1.59)
-0.006***
(-2.94)
0.003
(1.60)
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
0.099
0.099
0.100
0.100
Table E1
Influence of capital market forces on incentives to suppress negative financial information
(Section 5.3)
The following panels present select coefficients from pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t +2Politicali,t*Attributei,t + 3Post-Politicali,t
+ 4Post-Politicali,t*Attributei,t + 5Attributei,t-1 + 6LogSizei,t + 7Growthi,t + 8Sigmai,t + 9Turnoveri,t
+ 10Turnoveri,t-1 + 11Betai,t + 12Returni,t + 13Returni,t-1 +  14SOEi,t + i,
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. In these estimations, Political is an indicator variable equal to one for the year of a
National Congress meeting (first set of estimations) or the year preceding and corresponding to a provincial-level
political promotion event (second set of estimations), zero otherwise; Post-Political is an indicator variable equal to
one for the year directly following the specific political event, zero otherwise. Attribute captures one of two proxies
for capital market incentives for transparency: (1) the relative importance of equity markets at the provincial level
and (2) whether the firm issued B-shares. In the first estimation, Attribute is an indicator variable equal to one if the
total market capitalization of all listed firms in the province, scaled by provincial GDP, is in the top half of the
sample, zero otherwise. In the second estimation, Attribute is an indicator variable equal to one if the firm also
issued a class of B-share securities. All other variables are defined in Table 1 and Appendix A. T-statistics derived
using clustered standard errors by province are presented in parentheses. Models include annual and provincial
fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the
estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test),
respectively. N=12,723
Political Event:
Market-based Incentives:
Politicalt
Politicalt*Attribute
Post-Politicalt
Post-Politicalt*Attribute
Attribute
Control Variables
Year Fixed Effects
Provincial Fixed Effect
Adjusted R-squared
National Congress
Market Development
B-Shares
Provincial-Level Political Promotion
Market Development
B-Shares
-0.243***
(-7.54)
0.057
(0.51)
0.036*
(1.76)
0.411***
(3.37)
-0.123
(-0.92)
-0.058*
(-1.90)
-0.006
(-0.06)
0.116***
(3.38)
0.109
(0.72)
-0.045
(-0.40)
-0.220***
(-6.43)
-0.089
(-1.47)
0.054***
(2.73)
0.001
(0.02)
-0.042
(-1.54)
-0.060**
(-2.22)
0.029
(0.80)
0.118***
(3.92)
0.045
(0.63)
-0.066*
(-1.74)
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
0.174
0.125
0.174
0.126
Table E2
Influence of capital market forces on incentives to suppress negative financial information
(Section 5.3)
The following panels present select coefficients from pooled, cross-sectional estimations of the following model:
Ncskewi,t =  + Year + Province + 1Politicali,t +2Politicali,t*Attributei,t + 3Post-Politicali,t
+ 4Post-Politicali,t*Attributei,t + 5Attributei,t-1 + 6LogSizei,t + 7Growthi,t + 8Sigmai,t + 9Turnoveri,t
+ 10Turnoveri,t-1 + 11Betai,t + 12Returni,t + 13Returni,t-1 +  14SOEi,t + i,
The dependent variable Ncskew is the firm’s third moment of excess daily stock returns scaled by its cubed standard
deviation times minus one. In these estimations, Political is an indicator variable equal to one for the year of a
National Congress meeting (first set of estimations) or the year preceding and corresponding to a provincial-level
political promotion event (second set of estimations), zero otherwise; Post-Political is an indicator variable equal to
one for the year directly following the specific political event, zero otherwise. Attribute captures one of two proxies
for capital market incentives for transparency: (1) the relative importance of equity markets at the provincial level
and (2) whether the firm issued B-shares. In the first estimation, Attribute is an indicator variable equal to one if the
total market capitalization of all listed firms in the province, scaled by provincial GDP, is in the top half of the
sample, zero otherwise. In the second estimation, Attribute is an indicator variable equal to one if the firm also
issued a class of B-share securities. All other variables are defined in Table 1 and Appendix A. T-statistics derived
using clustered standard errors by province are presented in parentheses. Models include annual and provincial
fixed effects [Year and Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the
estimated coefficient is significantly different than zero at the one, five and ten percent level (two-tailed test),
respectively. N=12,723
Political Event:
Market-based Incentives:
Politicalt
Politicalt*Attribute
Post-Politicalt
Post-Politicalt*Attribute
Attribute
Control Variables
Year Fixed Effects
Provincial Fixed Effect
Adjusted R-squared
National Congress
Market Development
B-Shares
Provincial-Level Political Promotion
Market Development
B-Shares
-0.173***
(-4.64)
0.014
(0.13)
0.048
(0.95)
0.277
(0.85)
-0.125
(-1.33)
-0.125***
(-3.24)
-0.174**
(-2.27)
0.066
(1.23)
-0.005
(-0.10)
-0.104***
(-4.73)
-0.088**
(-2.40)
0.183
(0.87)
0.122***
(3.07)
0.137
(0.58)
-0.307
(-1.42)
-0.082**
(-2.71)
0.070
(0.91)
0.127***
(3.63)
0.072
(0.80)
-0.169***
(-2.91)
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
Included
0.105
0.106
0.0760
0.0770
Table F
Relation between political events and annual realized returns
The following panels present coefficients from various pooled, cross-sectional estimations of the following model:
Market-adjusted Returni,t =  + Year + 1Politicali,t + 2Post-Politicali,t + 3Logsizei,t + 4Market-to-booki,t +
5Market-adjusted Returni,t-1 + i,t
The dependent variable Market-adjusted Returni,t is the annual market-adjusted return of firm i in year t. Political is
an indicator variable equal to one if the firm-year relates to a specific political event; Post-Political is an indicator
variable equal to one in the year immediately following the event. In the first set of estimations, Political is an
indicator variable equal to one for the years that a National Congress of the Chinese Communist Party was held,
zero otherwise; Post-Political is an indicator variable equal to one for the year directly following each National
Congress event, zero otherwise. In the second set of estimations, Political is an indicator variable equal to one for
the year preceding and corresponding to a provincial level political promotion event, zero otherwise; Post-Political
is an indicator variable equal to one for the year directly following the provincial-level promotion event, zero
otherwise. All other variables are defined in Table 1 and Appendix A. T-statistics derived using clustered standard
errors by province are presented in parentheses. Models include annual and provincial fixed effects [Year and
Province, respectively; coefficients not reported]. The superscripts ***,**,* indicate that the estimated coefficient is
significantly different than zero at the one, five and ten percent level (two-tailed test), respectively. N=12,723
Political Event:
Dependent Variable
Politicalt
National Congress
Market-adjusted
Log(1+MarketReturn
Adjusted Return)
Provincial-Level Political Promotion
Market-adjusted
Log(1+MarketReturn
Adjusted Return)
-0.044
(-0.49)
-0.025
(-0.29)
0.148***
(10.40)
0.042***
(21.68)
-0.140***
(-9.99)
0.098
(0.73)
0.197
(1.50)
0.122***
(11.24)
0.030***
(19.59)
-0.112***
(-8.45)
-0.007
(-0.38)
0.007
(0.58)
0.146***
(11.70)
0.041***
(19.39)
-0.149***
(-11.96)
-0.011
(-0.57)
-0.007
(-0.54)
0.134***
(14.11)
0.032***
(19.18)
-0.130***
(-11.96)
Year Fixed Effects
Included
Included
Included
Included
Adjusted R-squared
0.237
0.215
0.261
0.256
Post-Politicalt
LogSizet
Market-to-bookt
Returnt-1