Earnings restatements and duration of supplier and the restating firm

In addition to examining whether suppliers seem to incorporate information from earnings restatement announcements when they make their relationship-specific investment decisions following the restatement, I further examine whether earnings

restatements affect the incentives of suppliers and customers to continue the business relationship following the restatement. If earnings restatements affect business relationship between the restating firms and their suppliers, then the information in earnings restatement will be associated with termination of relationships.

To examine this argument, I use a logistic regression model where the dependent variable equals one if the relationship breaks up in the subsequent year, and zero otherwise. In this model, I treat relationship termination as a discrete variable for each year in the sample. In addition to the proxies for news in earnings restatements variables, which are the variables of interest, I follow Fee et al. (2006) and Raman and Shahrur (2008) and use a set of control variables in my test model setting. In summary, I use the following logistic regression model:

0 1 2 3 4

5 6 7

&

CFC

TERMINATION NEWS DEP R D β ALLIANCE AGE β SIZE β F ε

β β β β

β

= + + + +

+ + + + (3)

where TERMINATION is a dummy variable that takes a value of one for suppliers if the relationship ends in the year subsequent to earnings restatement announcement, and zero otherwise. NEWS is either the abnormal returns of the restating firms (AAER (-1, 1) and AAER (-10, 10)) or the abnormal returns of suppliers (CAR (-1, 1) and

CAR (-10, 10)). DEP is the percentage of supplier sales that are made to the restating

firm.

R&D is a supplier firm’s annual R&D expenditures divided by total assets,

where missing values for R&D are treat as zero. ALLIANCE is a dummy variable that takes a value of one if the firms in a relationship had a formal alliance agreement prior restatement announcement and zero otherwise. AGE is the number of years in which the supplier is listed on Compustat. SIZE is the natural logarithm of supplier total assets. FCFC is a dummy variable that takes a value of one if a supplier’s free cash flow is less than or equal to zero. The free cash flow is a firm’s income before extraordinary items plus depreciation and amortization less capital expenditures.

If information in earnings restatements influences the business relationships between the restating firm and its suppliers, one would observe that the coefficient on

NEWS is negative in the model. This suggests that suppliers are more likely to

terminate business relationship with the restating firms when abnormal returns to either the restating firm or suppliers are more negative. In this section, I examine whether information in earnings restatements affects the incentives of suppliers to continue the business relationship. I test H3 by examining regression results from analyses of the relationship duration.

The results of duration analysis reported in Table 3.7 are based on four models to verify the robustness of our conclusions. I estimate a logistic regression model where the dependent variable equals 1 if the relationship ends in the first year subsequent year, and 0 otherwise. In my logistic regression model I treat relationship termination as a discrete variable for each year in the sample. Consistent hypothesis H3 prediction, in all models I find that the coefficients on NEWS are negative and significant based on the Chi-squared statistic. This indicates that more severity of earnings restatements (more negative abnormal returns to either the restating firms or suppliers) is associated with a significantly larger likelihood of a relationship termination subsequent to restatement announcement year.

[Insert Table 3.7 here]

Finally, I find that higher levels of supplier R&D intensity increase the likelihood of relationship termination, consistent with the findings in prior research (e.g., Fee et al. 2006; Raman and Shahrur 2008). In addition, I find that higher levels of supplier sales dependence are associated with a significantly larger likelihood of a relationship termination subsequent to restatement announcement year. Overall, the findings in this section are consistent with the idea that the information in earnings restatements is associated with the duration of business relationships.

3.5. Summary

This paper examines whether and how earnings restatements convey news relevant to the relationship-specific investment by suppliers. If so, restatement announcements induce suppliers to update their beliefs bout the revenue from such specific assets. I then expect that suppliers following optimal capital budgeting modify their investment decisions subsequent to restatement announcements.

Furthermore, I expect that these investments subsequent to restatement announcements are associated with the news in the restatement.

Following recent studies (e.g., Fee et al. 2006; Kale and Shahrur 2006; Raman and Shahrur 2008), I using R&D intensity of suppliers to proxy for relationship-specific investments by suppliers. The findings show that compared to selected benchmark control firms, suppliers significantly lower their R&D investment in the year after a restatement announcement. This change in suppliers’ R&D investments is significantly related to various proxies for news in the restatements, such as suppliers’ and restating firms’ abnormal returns surrounding the restatement announcement. Finally, I find that the relationships are more likely to terminate when the restatement announcements cause more negative abnormal returns to either suppliers or the restating firms.

Overall, my work indicates that there is an information transfer from restating firms to their suppliers at the restatement announcement involving information about the value of suppliers’ relationship-specific investment investments. This finding suggests that restatements of financial reports have direct implications for suppliers’

relationship-specific investments, and affect the suppliers’ allocation of resources.

Table 3.1 Sample distribution Panel A: Sample distribution by year

Year Restating firms Suppliers

1997

4 5

Industry Restating firms Suppliers

Mining and construction 1 1

Electrical equipment 10 37

transportation 4 8

instruments 5 13

computers 16 68

wholesale 6 26

miscellaneous retail 10 37

restaurant 1 4

services 2 2

Total 70 229

Panel A presents the distribution of restating firms and suppliers by years. Panel B reports the distribution of restating firms and suppliers by industry. Suppliers are identified through SFAS No. 131 disclosures and this information is available on COMPUSTAT Segment file. Any firm that lists the restating firm as a major customer is defined as supplier. The first column shows the number of restating firms that have at least one supplier. The last column reports the number of suppliers of the restating firms at individual firm level. Industries are defined in Beneish et al. (2008).

Table 3.2

Annual changes in suppliers’ R&D investments around the restatement announcement

This table shows descriptive statistics for the annual changes in suppliers’ R&D investments. Panel A displays the unadjusted annual changes in suppliers’ R&D investments. Panel B shows benchmark-adjusted annual changes in R&D investments for suppliers and for benchmark firms that belong to the 4-digit SIC industries which have not had a restatement in our sample period of 1997 to 2002. The Student t and the Wilcoxon statistics test the hypothesis that the mean and median changes in suppliers’ investment are significantly different from zero. p-values are calculated for two-tailed tests of significance.

Table 3.3

Descriptive statistics Panel A: Descriptive statistic for main variables

Variable N Mean Q1 Median Q3 Std. Dev.

D R&

Δ 155 -1.010 -0.010 0.001 0.017 0.310

CAR (-1,1) 226 -0.021 -0.053 -0.009 0.031 0.089 ARRE(-1,1) 226 -0.104 -0.075 -0.041 -0.003 0.138

∆FINANCING 226 0.212

0.009 0.077 0.336 0.246

∆SIZE

220 0.312 -0.087 0.089 2.434 2.549

∆CASH

226 -0.162 -0.391 -0.091 1.98 1.997

∆Q

225 -0.077 -0.109 -0.094 0.357 1.683

Panel B: Pearson correlation coefficients for main variables

CAR(-1,1) ARRE(-1,1) ∆FINANCING ∆SIZE ∆CASH ∆Q

∆R&D 0.187^** 0.162^** -0.060 0.088 -0.165^** -0.232^***

CAR(-1,1)

0.181^*** 0.104 -0.026 -0.080 -0.163^**

ARRE(-1,1)

0.029 -0.054 -0.025 -0.176***

∆FINANCING

-0.085 0.058 -0.051

∆SIZE

-0.123* 0.027

∆CASH

0.189***

This table presents descriptive statistics for main variables and the Pearson correlation coefficients for main variables in this study. ∆R&D = the change in supplier median R&D intensity, as measured by the year -1 to post-restatement median change in R&D intensity for suppliers. It is calculated as median R&D intensity of the three-year +1, +2, +3 after the year 0 of the restatement announcement (defined as period P) minus year -1 R&D intensity for suppliers (defined as period P-1). ∆FINANCING = the scaled change in the sum of equity issues and debt issues divided by total assets of suppliers between periods P-1 and P; ∆CASH = the scaled change in cash from assets-in-place divided by total assets of suppliers between periods P-1 and P. ∆Q = change in assets plus market value of equality minus book value of equity dividend by total assets between periods P-1 and P; ∆SIZE = the scaled change in the natural logarithm of total assets of suppliers between periods P-1 and P.

Table 3.4

Changes in suppliers’ R&D as a function of the news in the restatement Model 1 Model 2 Model 3 Model 4

ARRE (-1,1) ARRE (-10,10) CAR (-1,1) CAR (-10,10) Variable Coef t-value Coef t-value Coef t-value Coef t-value

Intercept

-0.155 -1.89 ^* -0.155 -1.95 ^* -0.040 -0.67 -0.043 -0.71

News

0.158 1.81 ^* 0.154 1.80 ^* 0.240 2.22^** 0.213 2.08 ^**

∆Financing

-0.042 -1.03 -0.014 -0.36 -0.028 -0.71 -0.022 -0.54

∆Size

0.016 2.29 ^** 0.014 2.01 ^** 0.007 1.14 0.007 1.09

∆Cash

-0.048 -2.37 ^** -0.034 -1.71 ^* -0.029 -1.44 -0.029 -1.47

∆Q

-0.003 -0.64 -0.006 -1.13 -0.011 -2.17^** -0.012 -2.29 ^**

Year fixed

effects yes yes yes yes

Adjusted R² 0.157 0.130 0.143 0.134

N 148 148 148 148

Dependent variable is the change in supplier median R&D intensity (∆R&D). ∆R&D = the change in supplier median R&D intensity, as measured by the year -1 to post-restatement median change in R&D intensity for suppliers. It is calculated as median R&D intensity of the three-year +1, +2, +3 after the year 0 of the restatement announcement (defined as period P) minus year -1 R&D intensity for suppliers (defined as period P-1). News is either suppliers’ abnormal returns around the restatement announcement, or restating firms’ abnormal returns around the restatement announcement. ∆Financing

= the scaled change in the sum of equity issues and debt issues divided by total assets of suppliers between periods P-1 and P; ∆Cash = the scaled change in cash from assets-in-place divided by total assets of suppliers between periods P-1 and P. ∆Q = change in assets plus market value of equality minus book value of equity dividend by total assets between periods P-1 and P; ∆Size = the scaled change in the natural logarithm of total assets of suppliers between periods P-1 and P.

Table 3.5

Changes in suppliers’ R&D as a function of the types of earnings restatements

Model 1 Model 2

Variable Coef t-value Coef t-value

Intercept

-0.168 -2.09^* -0.141 -1.70 ^*

REVENUE

-0.032 -1.77^* -0.029 -1.53

FRAUD*REVENUE

-0.078 -2.02 ^**

∆Financing

-0.017 -0.44 -0.012 -0.30

∆Size

0.013 1.99^** 0.015 2.30 ^**

∆Cash

-0.035 -1.77^* -0.032 -1.65 ^*

∆Q

-0.006 -1.26 -0.005 -1.07

Year fixed effects yes yes

Adjusted R² 0.141 0.153

N 148 148

Dependent variable is the change in supplier median R&D intensity (∆R&D). ∆R&D = the change in supplier median R&D intensity, as measured by the year -1 to post-restatement median change in R&D intensity for suppliers. It is calculated as median R&D intensity of the three-year +1, +2, +3 after the year 0 of the restatement announcement (defined as period P) minus year -1 R&D intensity for suppliers (defined as period P-1). REVENUE= 1 if the restatement relates to revenue recognition errors, 0 otherwise; FRAUD= 1 if the restatement relates to accounting fraud, 0 otherwise; FRAUD*

REVENUE= 1 if the restatement relates to revenue recognition errors and accounting fraud;

∆Financing = the scaled change in the sum of equity issues and debt issues divided by total assets of suppliers between periods P-1 and P; ∆Cash = the scaled change in cash from assets-in-place divided by total assets of suppliers between periods P-1 and P. ∆Q = change in assets plus market value of equality minus book value of equity dividend by total assets between periods P-1 and P; ∆Size = the scaled change in the natural logarithm of total assets of suppliers between periods P-1 and P.

Table 3.6

The impact of the economic bond on the relation between changes in suppliers’

relationship-specific investments and news in the restatement

CAR(-1,1) ARRE (-1,1)

(1) (2) (3) (4)

Variable

Coef t-value

Coef t-value Coef t-value Coef t-value Intercept -0.024 -0.41 -0.023 -0.40 -0.145 -1.80 ^* -0.152 -1.84 ^* NEWS 0.425 1.77 ^* 0.594 3.54 ^*** 0.160 1.85 ^* 0.157 1.77^*

DEP -0.085 -1.10 -0.071 -1.03

NEWS*DEP 0.523 1.98 ^** 0.780 1.96 ^**

ALLIANCE 0.012 0.58 0.011 0.49

NEWS*ALLIANCE 0.574 2.68 ^*** 0.023 0.16

∆FINANCING -0.032 -0.80 -0.036 -0.93 -0.054 -1.34 -0.041 -1.02

∆SIZE 0.007 1.15 0.005 0.84 0.016 2.35 ^** 0.015 2.09 ^**

∆CASH -0.028 -1.42 -0.035 -1.74 ^* -0.047 -2.36 ^** -0.051 -2.39 ^**

∆Q -0.010 -2.04 ^** -0.012 -2.41 ^** -0.003 -0.52 -0.003 -0.62 Year fixed effects yes yes yes yes Adjusted R² 0.158 0.187 0.188 0.146

N 148 148 148 148

Dependent variable is the change in supplier median R&D intensity (∆R&D), as measured by the year -1 to post-restatement median change in R&D intensity for suppliers. NEWS is either the abnormal returns of the restating firms (AAER (-1, 1) and AAER (-10, 10)) or the abnormal returns of suppliers (CAR (-1, 1) and CAR (-10, 10)). It is calculated as median R&D intensity of the three-year +1, +2, +3 after the year 0 of the restatement announcement (defined as period P) minus year -1 R&D intensity for suppliers (defined as period P-1). News is either suppliers’ abnormal returns around the restatement announcement, or restating firms’ abnormal returns around the restatement announcement. ∆FINANCING = the scaled change in the sum of equity issues and debt issues divided by total assets of suppliers between periods P-1 and P; ∆CASH = the scaled change in cash from assets-in-place divided by total assets of suppliers between periods P-1 and P. ∆Q = change in assets plus market value of equality minus book value of equity dividend by total assets between periods P-1 and P; ∆SIZE = the scaled change in the natural logarithm of total assets of suppliers between periods P-1 and P. ALLIANCE= a dummy variable that takes a value of one if the firms in a relationship had a formal alliance agreement with the restating firm before earnings restatement and zero otherwise. DEP= supplier sales to restating firm/total sales.

Table 3.7 Duration analysis

Model 1 Model 2 Model 3 Model 4

Variable Coef Chi-

Square Coef Chi-

Square Coef

Chi-Square Coef Chi- Square Intercept 0.400 0.40 0.404 0.41 0.227 0.13 0.228 0.13 NEWS -1.393 3.67 ^* -1.120 3.31 ^* -1.539 4.93 ^** -1.421 5.18 ^**

DEP 3.520 6.79 ^*** 3.434 6.78 ^*** 3.202 5.63 ^** 3.202 5.72 ^**

R&D 0.847 3.75 ^* 0.747 2.91 ^* 0.856 3.83 ^* 0.795 3.35 ^* ALLIANCE 2.520 2.56 1.755 1.40 2.238 2.08 1.683 1.31 AGE -0.004 0.09 -0.006 0.17 -0.003 0.06 -0.004 0.08 FCFC 0.451 2.14 0.398 1.67 0.430 1.93 0.429 1.93 SIZE -0.081 0.76 -0.063 0.45 -0.063 0.46 -0.064 0.47 Log likelihood 21.46 19.85 23.19 19.92

N 226 226 226 226

This table reports results of logistic regression analyses of the relationship termination.

TERMINATION is a dummy variable that takes a value of one for suppliers if the relationship ends in the year subsequent to earnings restatement announcement, and zero otherwise. NEWS is either the abnormal returns of the restating firms (AAER (-1, 1) and AAER (-10, 10)) or the abnormal returns of suppliers (CAR (-1, 1) and CAR (-10, 10)). DEPENDENCE is either the percentage of supplier sales that are made to the restating firm. R&D is a supplier firm’s annual R&D expenditures divided by total assets, where missing values for R&D are treat as zero. ALLIANCE is a dummy variable that takes a value of one if the firms in a relationship had a formal alliance agreement prior restatement announcement and zero otherwise. AGE is the number of years in which the supplier is listed on Compustat. SIZE is the natural logarithm of supplier total assets. FCFC is a dummy variable that takes a value of one if a supplier’s free cash flow is less than or equal to zero. The free cash flow is a firm’s income before extraordinary items plus depreciation and amortization less capital expenditures.

4. Earnings restatements and the efficiency of supply chain capital

在文檔中 The Vertical Information Transfer Effects of Earnings Restatements along the Supply Chain (頁 64-75)