Table 6 Instrumental variable approach (2SLS).

From: Consequences of firm-specific stock price crashes on analyst forecasts: Evidence from China

Panel A: Instrumental variable first-stage regression results

 

(1)

 

CRASH

MFHS

0.002**

 

(1.976)

NCSKEW

−0.172***

 

(−12.305)

REGLAG

0.582

 

(1.176)

SIGMA

−1.789***

 

(−5.606)

DUTURN

−2.965**

 

(−2.282)

SIZEPSM

0.076***

 

(8.590)

AGE

−0.015

 

(−0.573)

SALESG

0.022*

 

(1.763)

TANG

−0.025

 

(−1.337)

RM

−5.664***

 

(−2.720)

_cons

−1.697***

 

(−8.869)

N

56833

Pseudo R2

0.009

Year

Yes

Industry

Yes

Panel B: Instrumental variable second-stage regression results

 

(1)

 

FERROR

Fitted CRASH

1.615*

 

(1.696)

POST

0.405*

 

(1.891)

Fitted CRASH*POST

−0.874**

 

(−1.967)

SIZE

−0.457

 

(−1.405)

LEV

−0.687

 

(−0.983)

ROA

−15.933***

 

(−6.736)

INBOARD

0.403

 

(1.003)

INDENP

0.795

 

(0.567)

SHRCR

0.009

 

(0.707)

INST

−0.550*

 

(−1.876)

AUDITFEE

0.062

 

(0.551)

SOE

−0.300

 

(−0.521)

COMMITTEE

0.240

 

(0.929)

BIG4

−0.026

 

(−0.103)

EXP

−0.001

 

(−0.643)

HORIZON

0.060***

 

(12.876)

SPECIFIC

0.000

 

(0.190)

TFOLLOWING

−0.000

 

(−1.407)

BSIZE

−0.011**

 

(−1.979)

TEAM

−0.021**

 

(−2.242)

_cons

7.299

 

(0.992)

N

56833

R2

0.646

Year

Yes

Match

Yes

  1. Table 6 presents the results of the instrumental variable method. Panel A is the results of the first stage of instrumental variables, including the fixed effects of year and industry. Panel B is the result of the second stage of the instrumental variable. The fitted value of CRASH is generated by probit regression in the first stage, and then interacted with POST to estimate the fitted value of CRASH*POST. The regression includes year and matched-pair fixed effects. The sample covers the company-month-analyst data of non-missing values of all variables during the period 2001–2020. Robust t-statistics corrected for firm clustering are reported in parentheses. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively. All variables are defined in Appendix A.