4. RESEARCH RESULTS AND ANALYSIS
4.2 Empirical Results
國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
39
4.2 Empirical Results
4.2.1 Results for VACs and Post-earnings announcement drift (Full Sample)
The results are shown in Table 4, which presents the results for Equation (1) and Equation (2) through Model (1) and Model (2), respectively. Model (1) shows the market reaction around 10-Q filing window (FDt-1~FDt+1) and Model (2) shows those around post-filing drift window (FDt+2~EAt+60). The column of Full sample is about the results of all sample data. Table 4 shows the associations between the control variables and market reaction in Model (1) and Model (2). The insignificant coefficient of SURPRISE× LEVERAGE shows that leverage ratio of firms do not have a relation with market reaction drift after earnings announcement. The significant coefficients of SURPRISE× SGROWTH shows that firms with higher sales growth followed by less stock price drift in the filing window. The main results in Table 4 display insignificant coefficient ofSURPRISE× VAC
andSURPRISE× POST, which means VAC and POST are not related to the association
between earnings surprises and market reactions separately. However, when taking both VAC and POST into consideration, there is significant coefficient ofSURPRISE× VAC_POST in 10-Q filing window (0.285, p<0.1), and insignificant
coefficient in post-filing drift window, which is inconsistent with hypothesis 1 : VAC is negatively associated with the relation between earnings surprises and market reactions on filing date and positively associated with the post-filing announcement drift. That is, VAC firms adopting VACs only evoke the drift of market price in the filing window after VACs, compared to the non-VAC firms. However, the results haven’t involved the effect of accounting choice heterogeneity.‧
Table 4. Results for VACs and Post-earnings announcement drift (Full Sample)
Model (1) (2)
Event Window 10-Q Filing Post-Filing Drift
Dependent variable SARF SARDF
Full Sample Full Sample
Intercept 6.972***
Industry Effect Included Included
N 1349 1349
Adj. R2 0.02 0.02
*significant at 10%, ** significant at 5%, ***significant at 1%, t statistics are in parentheses and are estimated with clustered standard errors as in Petersen (2009). See Appendix for variable definitions.
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
41
4.2.2 Results for VACs and Post-earnings announcement drift (Accounting Choices Heterogeneity)
This study further considers the heterogeneity of accounting choices and categorizes the results into two groups in Table 5. The column of Similar consists of VAC firms adopting similar accounting method as their industry peers (non-VAC firms) after VACs, and the column of Different includes VAC firms adopting different accounting method from their industry peers after VACs. Besides, the column of Full sample is provided for comparison.
Considering the results of Similar column in Table 5, the control variables present some connections with the market reactions. For instance, as mentioned earlier, the negative significant coefficients of SURPRISE× SGROWTH shows that sales growth of firms has a reverse effect on stock price drift in the filing window.
The significant coefficients of SURPRISE× SIZE in the column of Similar in the filing window and insignificant in post-filing drift window, indicate that the earnings surprise of larger firms is followed by more stock price drift in the filing window when VAC firms have similar accounting method as non-VAC firms after VACs.
The significantly coefficient of SURPRISE× VAC (-0.442, p < 0.01) only showing in Model (1) indicates that there is a negative stock return drift in the filing window of VAC firms with similar accounting choices as their industry peers, compared to non-VAC firms. The significantly coefficient of SURPRISE× POST (-0.273, p <
0.01) in Model (1) shows that in the case of VAC firms adopting similar accounting choices as non-VAC firms, this research finds significant market price drift after
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
42
VACs in the filing window, compared to that before VACs. However,
SURPRISE× VAC and SURPRISE× POST do not consider whether the firm-quarter
observation is after VACs or whether the firm has VACs, respectively. The coefficient of SURPRISE× VAC_POST in the column of Similar is significantly positive in the filing window and significantly negative in the post-filing drift window (0.727 and -0.402, p < 0.01 and p < 0.1), showing that when VAC firms adopting similar accounting choices as their industry peers, the association between earnings surprises and market reactions is larger around the filing date but smaller in the post-filing window. Moreover, through the coefficient ofSURPRISE× VAC_POST in the column of Similar compared to that of Full sample, it
seems that the impact of VACs to the filing drift and post-filing drift are much clearer when considering the heterogeneity of accounting choices.Last, focusing on the VACs that are different from their industry’s peers’, the coefficient of SURPRISE× VAC_POST is insignificant in Model (1) (0.081, n.s.) and positively significant in Model (2) (0.393, p < 0.1). Different from the column of
Full Sample and Similar, the results in Different column indicate that when VAC
firms adopt different accounting method from their industry’s peers after VACs, VACs have a positively impact on the association between earnings surprise and market reactions in the post-filing window. Therefore, only when considering the heterogeneity of accounting choices, this study observes that VACs lead to more post-filing drift.‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
43
In summary, the results in full sample indicate that VACs influence the earning information cognizance of external users, which leads to more market reaction in 10-Q filing window. However, it seems that the impact of VACs to post-filing drift doesn’t appear.
Furthermore, after considering the heterogeneity of accounting choices, the empirical results indicate that VAC firms adopting similar accounting choices as non-VAC firms after VACs makes external users easier to incorporate earnings information and reduce their information processing costs, which leads to less delayed price response after earnings announcement, thus, more of earnings-related stock price reaction occurs in the 10-Q filing window and less of earnings-related market reaction appears in the post-filing drift window.
Compared to the data of Similar column, when VAC firms adopting different accounting choices from non-VAC firms after VACs, external users get harder to absorb related earnings information, which results in more delayed price response after earnings announcement. That means earnings-related market reaction defer until post-filing drift window. Therefore, the association between VACs and market reactions is larger in the post-filing window when considering accounting choices heterogeneity.
‧
Table 5. Results for VACs and Post-earnings announcement drift (Accounting Choices Heterogeneity)
*significant at 10%, ** significant at 5%, ***significant at 1%, t statistics are in parentheses and are estimated with clustered standard errors as in Petersen (2009). See Appendix for variable definitions.
Model (1) (2)
Event Window 10-Q Filing Post-Filing Drift
Dependent variable SARF SARDF
Full Sample Similar Different Full Sample Similar Different
Intercept 6.972*** Industry Effect Included Included Included Included Included Included
N 1349 454 711 1349 454 711
Adj. R2 0.02 0.09 0.04 0.02 0.04 0.03
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
45