Docstoc

Corporate Overview Presentation

Document Sample
Corporate Overview Presentation Powered By Docstoc
					An empirical investigation of auditor independence and accounting conservatism: Further evidence from Taiwan market-based incentives
Authors: Yi-Hsing Liao

Outline Presented
 Research motivation

 Research purposes  Research design
Hypotheses development Empirical methodology and variable measurement Sample formation and data resources
Qct. -2001

 Empirical results and analysis

 Summary and concluding remarks

Research motivation
According to the literatures surveyed, whether non-audit services really create a threat to auditor’s independence and reporting quality has been controversial for a long time and not yet come to a conclusion currently. In addition, a number of prior studies find that they are short of significant evidence to support the argument that non-audit services reduce auditor independence. One of the reasons why the auditing literatures can not draw a consistent conclusion on the matter is that they only focus on the impacts of non-audit services on auditor independence but ignore the role of market forces (Craswell et al. 1995; Larcker and Richardson 2004; Ruddock et al. 2006). Therefore, to explore the effect of market forces on the relationship between non-audit services and accounting conservatism is potentially interesting.

Qct. -2001

Research purposes
This paper aims to extend the literatures of nonaudit services research, and further explore the impacts of market forces on the relationship between the provision of non-audit services and accounting conservatism from the viewpoint of corporate governance. I hope my research can provide additional evidence with respect to market forces definitely serve to discipline auditors from compromising their independence.

Qct. -2001

Research design-Hypotheses development
 H1: Ceteris paribus, the auditors’ fee dependence is negatively associated with the level of accounting conservatism. .  H2: Ceteris paribus, the negative effect of auditors’ fee dependence on earnings conservatism is higher for firms with weak quality of board oversight than for firms with strong quality of board oversight .

Qct. -2001

Empirical methodology and variable measurement
Proxies for Auditor Independence:  the ratio of non-audit fees to total fees  abnormal non-audit fees based on the expectation models

 abnormal total fees based on the expectation models
Qct. -2001

Empirical methodology and variable measurement Proxies for Board Characteristics:
 This paper employs four characteristics for board of director to shed light on the independence and monitoring motivation of the board: (i) Percentage of independent directors and supervisors on the board, (ii) Percentage of family directors and supervisors on the board, (iii) CEO duality, and (iv) Board size.  The percentile rank method is applied to the four individual measures of board characteristics-INDEP, FAM, CEO, and BOSIZE-to drive an overall governance score which I denote as CGTS (Milliron 2000; Bushman et al. 2004). For the following analysis, this paper will use the aggregate CGTS measure as a proxy for board characteristics governance.

Qct. -2001

Empirical methodology and variable measurement Measurement of Accounting Conservatism:
 The accrual-based conservatism measure (hereafter ACCRUAL) is computed as follows: Net income before extraordinary items plus depreciation expense less cash flows from operations is deflated by average total assets, and averaged over 3 years centered around year “t”.  Basu’s asymmetric timeliness of earnings measure with the backward-cumulation approach.
Qct. -2001

Empirical methodology and variable measurement
 Empirical model for tests using accrual-based conservatism measures is as follows:

According to H1, as auditors reduces their independence, the level of accounting conservatism will decrease, therefore, I predict that β1 >0. Furthermore, β3 represents the incremental effect of client importance on accounting conservatism for firms with strong board governance. If H2 is valid, then I would predict that β3 <0.

Qct. -2001

Empirical methodology and variable measurement
ACCRUAL: net income before extraordinary items plus depreciation expense less cash flows from operations deflated by average total assets and averaged over 3 years centered around year t; CLIMP: three measures of client importance that focus on the ratio of non-audit fees to total fees, abnormal total fees and abnormal non-audit fees; BOARD: an overall governance scores for board characteristics driven by the percentile rank method; SIZE: the natural logarithm of average total assets; LEV: total long-term liabilities divided by average total assets; SAGR: percentage of annual growth in total sales; INST: ratio of sum of foreign institutional, securities investment trust enterprises’ and securities firms’ ownership to total ordinary shares outstanding; RDAV: research and development costs plus advertising expense divided by sales; CFO: cash flows from operations deflated by average total assets; IMR: the inverse Mill’s ratio obtained from the two-stage Heckman (1978) approach; ε: residual;

Qct. -2001

Empirical methodology and variable measurement
Empirical model for tests using Basu’s asymmetric timeliness of earnings measure is as follows:

Qct. -2001

Under Basu (1997) regression specification, I expect the coefficient for β13 to be significantly negative. According to H2, I predict that the coefficient of β21 will be significantly positive.

Empirical methodology and variable measurement
Et, t-j: income before extraordinary items cumulative from year t-j to year t; Pt-j-1: the market value of equity at the end of the year t-j-1; Rt, t-j: the buy-and-hold return starting 4 months after the end of the fiscal year t-j-1 and ending 4 months after the end of year t; DRt, ,t-j: a dummy variable is equal to one if DR is negative, zero otherwise; MTBt: market-to-book ratio defined as market value of equity divided by book value of equity at the end of current year; LEVt: total long-term liabilities divided by average total assets; CLIMPt: three measures of client importance that focus on the ratio of non-audit fees to total fees, abnormal total fees and abnormal non-audit fees; BOARDt: an overall governance scores for board characteristics driven by the percentile rank method; IMRt: the inverse Mill’s ratio obtained from the two-stage Heckman (1978) approach; ε: residual;

Qct. -2001

Sample formation and data resources
TABLE 2 Sample Selection and Sample Breakdown by Industry from 2002 to 2006
Panel A Sample Selection Number of firm years with complete data for estimated proxy measure for abnormal fees (after the exclusion of Financial Institution) Missing the inverse Mill’s ratio data Missing board characteristics data in the annual report Missing one or more variables related to earnings conservatism measure model Final sample Panel B Sample Breakdown by Industry Industry Chemicals Cement Automobile Glass ceramics Food Textiles Shipping and transportation Information service Paper pulp Trading and consumers' goods Plastics Electron Electronic and cable Electronic machinery Rubber Iron and steel Building material and construction Other Total 412 10 74 58 270

Firm-years 6 5 8 3 7 8 2 1 1 2 11 149 2 20 7 12 9 17

Qct. -2001

Sample formation and data resources
sample period
 the sample period spans five years from 2002 through 2006.

data sources
 Regarding data sources, except for the board characteristics variables which are manually collected from the annual reports, the regression variables required for estimating accounting conservatism and other financial variables of the empirical model are all collected from the TEJ database and the Times Winner database.  The demand for non-audit services may depend upon firm’s economic characteristics. To address this self-selection bias issue, my study uses the two-stage procedure of Heckman (1979) to solve the problem.

Qct. -2001

Empirical results and analysis-Table 5
V ariable Intercept NAFTF RESITF RESINAF BOARD CLIMP* BOARD SIZE LEV SAGR INST RDAV CFO + + + ? ? + ? ? 0.035 (0.005)*** -0.093 (0.007)*** -0.010 (0.000)*** -0.059 (0.005)*** 0.034 (0.008)*** 0.107 (0.001)*** -0.159 (0.003)*** -0.334 (0.000)*** -0.031 (0.001)*** -0.059 (0.043)** 31.16% 13.178 (0.000)*** 270 0.015 (0.049)** -0.057 (0.000)*** -0.009 (0.001)*** -0.058 (0.014)** 0.028 (0.016)** 0.110 (0.003)*** -0.144 (0.003)*** -0.321 (0.000)*** -0.023 (0.006)*** -0.069 (0.000)*** 37.39% 17.063 (0.000)*** 270 Predicted sign Model 1 0.222 (0.000)*** 0.034 (0.033)** Model 2 0.202 (0.000)*** Model 3 0.226 (0.000)*** -0.012 (0.167) 0.005 (0.025)** 0.011 (0.187) -0.011 (0.012)** -0.010 (0.001)*** -0.059 (0.023)** 0.033 (0.008)*** 0.106 (0.010)** -0.156 (0.005)*** -0.331 (0.000)*** -0.031 (0.001)*** -0.007 (0.094)* 30.67% 12.901 (0.000)*** 270

Qct. -2001

IMR

β2+β4 Adjusted-R2 F Statistic (p-value) N

-

Empirical results and analysis-Table 6
V ariable Intercept DR R R*DR MTB MTB*DR MTB*R MTB*R*DR LEV LEV*DR LEV*R LEV*R*DR CLIMP CLIMP*DR CLIMP*R CLIMP*R*DR Predicted sign ? + + ? ? ? ? ? ? + ? ? ? ? ? ? ? ? ? ? Model 1 0.399 (0.101) -2.246 (0.001)*** 0.337 (0.031)** -4.168 (0.009)*** 0.035 (0.384) 0.171 (0.230) -0.013 (0.032)** -0.103 (0.411) -0.374 (0.185) 2.379 (0.005)*** 0.088 (0.506) 5.340 (0.001)*** 0.232 (0.543) 1.534 (0.057)* -0.131 (0.718) 2.225 (0.241) 0.089 (0.498) -0.156 (0.752) 0.113 (0.223) -1.717 (0.106) -0.003 (0.995) 1.838 (0.161) -0.402 (0.328) Model 2 0.513 (0.019)** -2.228 (0.002)*** 0.239 (0.074)* -4.285 (0.003)*** -0.009 (0.805) 0.432 (0.019)** -0.006 (0.279) 0.074 (0.444) -0.259 (0.277) 2.531 (0.008)*** 0.063 (0.609) 6.561 (0.001)*** 0.574 (0.005)*** -1.054 (0.001)*** -0.385 (0.006)*** -1.649 (0.016)** 0.092 (0.186) 0.135 (0.552) 0.054 (0.206) 0.331 (0.530) -0.682 (0.003)*** 1.880 (0.000)*** 0.423 (0.007)*** Model 3 0.406 (0.071)* -1.891 (0.001)*** 0.319 (0.025)** -3.981 (0.005)*** 0.020 (0.598) 0.086 (0.555) -0.013 (0.035)** -0.187 (0.353) -0.263 (0.326) 2.679 (0.004)*** 0.054 (0.703) 6.066 (0.001)*** 0.046 (0.419) 0.415 (0.097)* -0.009 (0.805) 0.825 (0.197) 0.068 (0.308) 0.569 (0.023)** 0.041 (0.324) 1.145 (0.032)** -0.047 (0.498) 0.291 (0.373) 0.005 (0.901)

Qct. -2001

BOARD BOARD*DR BOARD*R BOARD*R*DR CLIMP*BOARD CLIMP*BOARD*DR CLIMP*BOARD*R

Empirical results and analysis-Table 6
Variable CLIMP*BOARD*R*DR IMR IMR*DR IMR*R IMR*R*DR Table 6 (continued) Predicted Model 1 sign 8.194 + (0.000)*** -0.085 ? (0.471) 0.366 ? (0.020)** -0.167 ? (0.060)* 1.085 ? (0.000)*** + 10.419 (0.000)*** 51.08% 10.437 (0.000)*** 245 Model 2 3.886 (0.000)*** -0.134 (0.274) 0.449 (0.046)** -0.109 (0.119) 1.134 (0.000)*** 2.237 (0.016)** 50.65% 10.274 (0.000)*** 245 Model 3 1.347 (0.056)* -0.078 (0.528) 0.324 (0.115) -0.152 (0.094)* 1.031 (0.000)*** 2.171 (0.000)*** 49.47% 9.848 (0.000)*** 245

Qct. -2001

β13+β21 Adjusted-R2 F Statistic (p-value) N

 I rerun the analyses after deleting the outliers and the results do not change.  I repeat all tests reported in Tables 5 and 6 with board governance based on outside directors and supervisors in replace of independent counterparts, all results are not qualitatively affected.  I perform book-to-market ratio tests following Beaver and Ryan (2000) as another conservatism measure.  The paper further investigates the impacts of alternative market forces such as litigation risk and reputation loss on the relationship between client importance and accounting conservatism.  To ensure that the clients of any one audit firm are not driving the results, I form five separate subsamples by deleting in turn the clients of each of the Big 4 firms and non-Big 4 firms at a time.

Robustness tests

Qct. -2001

Summary and concluding remarks
 Empirical results reveal that the impairment of auditor independence is significantly associated with the reduction of conservatism. Especially, I further find that sound board governance mechanisms definitely offset the effect of aggressive accounting caused by impaired auditor independence.  Collectively, these findings are consistent with the view that board governance is the primary determinant of auditor’s attitude toward conservative accounting.
Qct. -2001

Thank you for your attention!

Qct. -2001


				
DOCUMENT INFO