VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… ·...

27
1 The Nonlinear Association between Abnormal Audit Fees and Audit Quality: Evidence from Hong Kong DBA Candidate : Alan S. K. Ho DBA VIVA presentation 28 th July 2008 Panel: Chairman: Prof. Bin SRINIDHI Supervisor: Prof. Jeong-Bon KIM External Examiners: Dr. Charles CHEN J P Prof. FIRTH Michael Arthur Version 1.0 DBA VIVA Presentation 28 July 2008 2 Outline 1. Introduction 2. Background 3. Operationalisation 4. Empirical procedures 5. Results 6. Conclusions 7. Q&A

Transcript of VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… ·...

Page 1: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

1

The Nonlinear Associationbetween

Abnormal Audit Fees and Audit Quality:Evidence from Hong Kong

DBA Candidate : Alan S. K. Ho

DBA VIVA presentation 28th July 2008

Panel:

Chairman: Prof. Bin SRINIDHI

Supervisor: Prof. Jeong-Bon KIM

External Examiners:

Dr. Charles CHEN J P

Prof. FIRTH Michael Arthur

Version 1.0 DBA VIVA Presentation 28 July 2008 2

Outline

1. Introduction2. Background3. Operationalisation4. Empirical procedures5. Results6. Conclusions7. Q&A

Page 2: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

2

Version 1.0 DBA VIVA Presentation 28 July 2008 3

1. Introduction

Version 1.0 DBA VIVA Presentation 28 July 2008 4

The research questions

Do audit fees create economic bond between auditor and its client?

If so, how does this affect the audit quality?

Page 3: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

3

Version 1.0 DBA VIVA Presentation 28 July 2008 5

The research gapResults of prior studies are mixed and inconclusive. E.G., A survey on 25 most recent studies: 12 with no association, 2 with +ve association, 5 with -ve association, 6 with different associations in different divisions.

Why?Incorrect measure for economic bond (use size of fees and various fee ratios) - the concept of super profitsNonlinearity of abnormal fees – The effects of +ve and –ve abnormal fees are cancelling each other out.The measure of the consequence of lower audit quality (e.g. discretionary accruals) is not powerful enough.

Version 1.0 DBA VIVA Presentation 28 July 2008 6

This study will ….Use positive abnormal audit fees (POSAAF) to measure the economic bond.Use accrual quality and C-Score to measure the audit quality.Demonstrate the nonlinear association between abnormal audit fees and audit quality.

Page 4: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

4

Version 1.0 DBA VIVA Presentation 28 July 2008 7

2. Background

Version 1.0 DBA VIVA Presentation 28 July 2008 8

Audit quality & Independence

DeAngelo (1981a) defines audit quality as the market assessed joint probability that a given auditor will both discover a breach in the client’s financial statements and report such breach.

Competence

Independence

Audit Quality

Page 5: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

5

Version 1.0 DBA VIVA Presentation 28 July 2008 9

The Conceptual Model

Independence

Audit

Quality

Training and

Standards Earnings Quality

Economic Bond

Competence

Abnormal Audit Fees

Version 1.0 DBA VIVA Presentation 28 July 2008 10

3. Operationalisation

Page 6: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

6

Version 1.0 DBA VIVA Presentation 28 July 2008 11

How should the economic bond be measured ?

DeAngelo (1981) suggests that Client-Specific Quasi Rent (CSQR) will create the economic bond. CSQR is the “super profits” arising from the client relationship.Therefore, large fees may not create an economic bond as long as they do not generate “super profits”.

Fees = Normal Fees + Abnormal Fees

Normal Fees = ƒ ( size, complexity, risk) [Simunic 1980]

The Positive Abnormal fees = “super profits”.

Version 1.0 DBA VIVA Presentation 28 July 2008 12

Abnormal audit fees estimationTAF = NAF + AAF ………. (1)

TAF = α0 + α1 LNTA + α2 INVREC + α3 EMPLOY + α4 ISSUE+ α5 FOREIGN + α6 EXORD + α7 LOSS + α8 LEVE+ α9 ROA+ α10 CURRENT + α11 Big4+ α12 PB + α13 AUDITSPEC + Industry dummy + Yearend dummy +HSI dummy+ H share dummy + Red Chip dummy + Year dummy +ξ ………..(2)

Keys:

TAF = natural log of fees paid to auditors for their financial statement audits (i.e., audit fees) in million Hong Kong dollars; LNTA = natural log of total assets in million Hong Kong dollars; INVREC = inventory and receivables divided by total assets; EMPLOY = square root of the number of employees + 1; ISSUE = 1 if there is issue of equity or debt in the current year, 0 otherwise; FOREIGN = 1 if the firm pays any foreign income tax, 0 otherwise; EXORD = 1 if the firm reports any extraordinary gains or losses, 0 otherwise; LOSS =1 if the firm reported a loss during the year, 0 otherwise; LEVE = leverage (total liabilities divided by total assets); ROA = return on assets (income before extraordinary items divided by total assets); CURRENT = current assets divided by current liabilities; BIG4 = 1 if the auditor is one of Big 4, 0 otherwise; PB =Price to Book ratio; AUDITSPEC = 1 if the auditor is the largest market shares holder within the client industry, 0 otherwise; Industry dummies = the industry classification of Worldscope; Yearend dummy =1 if the financial year is ended in 31 December, 0 otherwise; HSI dummy =1 if the company is a composite stock of Hang Seng Index, 0 otherwise; H Share dummy =1 if the company is listed as a H share, 0 otherwise; Red Chip dummy =1 if the company is listed as Red Chip, 0 otherwise; Year dummy =Financial year of the company

(Sources: Carcello et al., 2002,, Chaney, Jeter, & Shivakumar, 2004, Choi et al., 2005, Choi et al., 2006, Craswell et al., 1995, DeFond, Francis & Wong, 2000, DeFond et al., 2002, Francis & Stokes, 1986, Gist, 1992, Gul, Chen, & Tsui, 2003, Gul, Srinidhi & Shieh, 2002, Gul & Tsui, 1998, Larker & Richardson, 2004, Pong & Whittington, 1994, Sankaraguruswamy & Whisenant, 2005, Seetharaman, Gul, & Lynn, 2002, Simunic, 1980, Tsui, Jaggi & Gul, 2001, Whisenant, Sankaraguruswamy, & Raghunandan, 2003.For detail specifications of their models please refer to Appendix 4)

Page 7: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

7

Version 1.0 DBA VIVA Presentation 28 July 2008 13

Dechow & Dichev (2002) accrual quality (“DDAQ”)

Accruals = Informative (private information signalling) + non-informative (earnings management).non-informative accruals (are less likely to be realized as future cash flows ) = poor earnings quality.Dechow and Dichev (2002) measure the accrual quality by the residuals (or its standard deviation) of the accruals and cash flows regression. The higher the value indicates lower accrual quality (i.e. lower EQ).

Version 1.0 DBA VIVA Presentation 28 July 2008 14

Accrual quality estimationTCAj,t = b0 + b1 CFOj,t-1 +b2 CFOj,t + b3 CFOj,t+1 +b4 ChgRevj,t

+ b5 PPEj,t + ε j,t ……… (3)

Keys:

TCAj,t = For company j in year t, net income minus cashflow from operations

CFOj,t-1 = For company j in year t-1, cashflow from operations

CFOj,t = For company j in year t, cashflow from operations

CFOj,t+1 = For company j in year t+1, cashflow from operations

ChgRevj,t = For company j in year t, the change of revenue from period t-1 to t

PPEj,t = For company j in year t, the gross value of property, plant and equipment

Source: Dechow and Dichev, 2002

Page 8: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

8

Version 1.0 DBA VIVA Presentation 28 July 2008 15

C-Score and conditional conservatism

Conditional and unconditional conservatism – (Ball and Shivakumar, 2005) “Unconditional Conservatism” is effectively a “known”bias and has little effect on EQ Conditional conservatism is firm-year specific – (Ryan, 2006).Khan and Watts (2007) propose C-Score, a firm-year specific conservatism measure, which is a linear function of firm and year specific attributes which are leverage, firm size, and price to book ratio

Version 1.0 DBA VIVA Presentation 28 July 2008 16

C-Score estimationa) Yearly λ coefficients estimationX j, t = β 1,t + β 2,t D j,t + R j,t (μ 1,t + μ 2,t Size + μ 3,t PB + μ 4,t Lev)

+ D j, t * R j,t (λ 1,t +λ 2,t Size +λ 3,t PB +λ 4,t Lev) + ε j,t ……… (4)

b) Firm year specific C-Score estimationC-Score j, t = λ 1,t +λ 2,t Size j,t +λ 3,t PB j,t +λ 4,t Lev j,t ……. (4A)

Keys:

X j, t = net income before extraordinary items, scaled by the beginning of year market value of equity

D j,t = 1 if R ≦ 0, zero otherwise

R j,t = Annual returns beginning from the fifth month after the fiscal year end

Size = Natural log of the end of year market value equity

PB = Price to book ratio

Lev = Total liabilities divided by beginning of year market value of equity

Source: Khan and Watts (2007)

Page 9: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

9

Version 1.0 DBA VIVA Presentation 28 July 2008 17

The Research Model

IndependenceAudit

QualityEconomic

Bond

Positive abnormal audit fees

(POSAAF)

Earnings Quality

Operational

Measures

Constructs

Accrual Quality and

Conservatism

Control Variables

Firm and Auditor

CharacteristicsIndependent

Variables

Dependent variable

Version 1.0 DBA VIVA Presentation 28 July 2008 18

The HypothesesPositive Abnormal Audit Fees

H1 (null): The positive abnormal audit fees (POSAAF) do not have a relation with earnings management, ceteris paribus.H1 (alternative): The positive abnormal audit fees (POSAAF) have a positive relation with earnings management, ceteris paribus.H2 (null): The positive abnormal audit fees (POSAAF) do not have a relation with accounting conservatism, ceteris paribus.

Negative Abnormal Audit FeesH3 (null): The negative abnormal audit fees (NEGAAF) do not havea relation with earnings management, ceteris paribus.H3 (alternative): The negative abnormal audit fees (NEGAAF) havea negative relation with earnings management, ceteris paribus.H4 (null): The negative abnormal audit fees (NEGAAF) do not havea relation accounting conservatism, ceteris paribus.

Page 10: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

10

Version 1.0 DBA VIVA Presentation 28 July 2008 19

4. Empirical procedure

Version 1.0 DBA VIVA Presentation 28 July 2008 20

Data and Sample

Based on the market capitalization as of 31 December 2006, the top 350 Hong Kong companies (the “sample”) are selected from the Worldscope database. The sample has a total market capitalization of HK$ 14.68 trillion which is 98.30% of the population market capitalization. The sample is roughly 38% of the population in terms of number of companies.Fy1996 – FY2006All data are Winsorized at top and bottom 1%

Page 11: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

11

Version 1.0 DBA VIVA Presentation 28 July 2008 21

Number of observations

890With Fees + DDAQ783With Fees + C-Score

1,776With C-Score2,076With DDAQ

1,129With abnormal audit fees metrics (Fees)

3,850Full panel (11 yr x 350 firms)

Number of observations

Version 1.0 DBA VIVA Presentation 28 July 2008 22

Abnormal audit fees and DDAQ relation estimation

DDAQ j, t = α0 + α1 SIZEj,t + α2 OPCYCLEj,t +α3 SALESVj

+ α4 LOSSDUMMYj,t +α5 ABFeesj,t +ε .......... (5)

Keys:

DDAQj,t = Accrual quality, absolute value of the residual from equation (3) for company j on year t

ABFeesj,t = abnormal audit fee metrics, POSAAF and NEGAAF are residuals from equation (2) for company j on year t scaled by natural log of total audit fees

SIZEj,t = Natural log of the total assets of company j at the end of year t

OPCYCLEj,t = Sum of the account receivables days and inventory days of company j on year t

LOSSDUMMYj,t =1 if net income < 0, otherwise = 0 for company j on year t

SALESVj=Standard deviation of (sales/ total assets) from 1996 to 2006 of company j

Source: Dechow and Dichev (2002)

Page 12: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

12

Version 1.0 DBA VIVA Presentation 28 July 2008 23

Abnormal audit fees and DDAQ Section A

controlsSection BPOSAAF

Section CNEGAAF

Section DAAF

Variable Expsign

Coefficient(t value)

Coefficient(t value)

Coefficient(t value)

Coefficient(t value)

(Constant) N/A -3137.85 -4532.69 -3077.81 -3191.20(-19.37***) (-8.80***) (-10.35***) (-13.30***)

SIZE (-) 425.66 558.21 410.80 426.14(24.50***) (11.02***) (13.08***) (17.19***)

OPCYCLE (+) 0.01 -0.27 -0.02 -0.05(0.13) (-1.47) (-0.17) (-0.60)

SALESV (+) 397.36 78.82 290.39 214.19(3.46***) (0.33) (1.70*) (1.50)

LOSSDUMMY (+) 601.89 737.46 487.72 612.53(8.07***) (4.86***) (4.19***) (6.44***)

ABFees 623.65 26.83 56.93(3.79***) (0.86) (-1.52)

n 1638 428 462 890

Adj R2 0.2800 0.3040 0.3030 0.2930

Keys:

T statistics are in parentheses,

*** = 1% sig,

** = 5% sig,

* = 1% sig

Version 1.0 DBA VIVA Presentation 28 July 2008 24

Abnormal audit fees and C-Score relation estimation

C-Scoret = α0 +α1 C-Scoret-1 + α2 DDAQt + α3 VORt + α4 Age +α5 Cyclet

+α6 ABFeest +ε

Sources: Khan and Watts (2007) and Chi et al. (2007)

Keys:

C-Scoret = C-Score of year t

C-Scoret-1 = C-Score of year t-1

DDAQt = Accrual quality measure of Dechow and Dichev of year t

VORt = Standard deviation of the firm level monthly return for year t

Age = Number of year that the firm is listed during the period 1996 to 2005.

Cyclet = Depreciation expenses at year t divided by the gross fixed assets at the beginning of the year

ABFeest = AAFt - Abnormal audit fees at year t; or

POSAAFt = Positive abnormal audit fees at year t; or

NEGAAFt = Negative abnormal audit fees at year t

Page 13: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

13

Version 1.0 DBA VIVA Presentation 28 July 2008 25

Abnormal audit fees and C-Score Variable

ExpSign

Section Acontrols

Section BPOSAAF

Section CNEGAAF

Section DAAF

Coeffieicent(t value)

Coeffieicent(t value)

Coeffieicent(t value)

Coeffieicent(t value)

(Constant) 0.505 0.441 0.106 0.294(2.144**) (0.843) (0.261) (0.914)

C-score t-1 (+) 0.045 0.165 0.032 0.098(3.735***) (4.303***) (0.952) (3.894***)

AQ t (-) 0.000 0.000 0.000 0.000(-7.517***) (-2.593***) (-3.6***) (-4.471***)

VOR t (+) -0.046 -0.053 -0.044 -0.055(-3.06***) (-1.373) (-1.118) (-1.989**)

Age (-) -0.011 -0.026 -0.005 -0.014(-0.47) (-0.504) (-0.128) (-0.429)

Cycle t (-) -1.901 -3.007 -0.494 -1.916(-1.805) (-1.415) (-0.205) (-1.22)

ABFees t 0.260 0.133 0.141(2.042**) (3.106***) (3.227***)

n 1563 374 409 783

Adj R2 0.069 0.128 0.075 0.095

Keys:

T statistics are in parentheses,

*** = 1% sig,

** = 5% sig,

* = 1% sig

Version 1.0 DBA VIVA Presentation 28 July 2008 26

5. Results

Page 14: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

14

Version 1.0 DBA VIVA Presentation 28 July 2008 27

DDAQ results

POSAAF is negatively related to accrual quality (+ve with DDAQ)

NEGAAF and AAF have no significant relation with DDAQ

Srinidhi and Gul (2006) find no significant relation between unexpected audit fees (AAF) and accrual quality, using US data from 2000 to 2001- Similar results

Version 1.0 DBA VIVA Presentation 28 July 2008 28

C-Score resultsBoth POSAAF and NEGAAF have positive relations with C-Score. And the coefficients of POSAAF is twice as NEGAAF nonlinearity.

Gul et al. (2002) using Hong Kong data from 1990 to 1997 find that total audit fee has a negative relation with conservatism. Repeat the test on total audit fees, the coefficient is -0.167 (t = - 4.897, p-value = 0.000) Similar resultsKrishnan (2004) using US data from 2000 to 2001 finds that greater conservatism for high-fee clients (POSAAF) than low-fee clients. Similar resultsJones et al. (2006) using US data from 2000 to 2004 find that firms who pay high unexpected fees (similar to POSAAF in this study) have higher conservatism in their financial statements. Similar results

Page 15: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

15

Version 1.0 DBA VIVA Presentation 28 July 2008 29

Nonlinearity

Yes

Yes

Nonlinearity

Coefficient of POSAAF is twice as NEGAAF

+VE+VEC-Score

Insig.+VE DDAQ

RemarksNEGAAFPOSAAF

Version 1.0 DBA VIVA Presentation 28 July 2008 30

Hypotheses verification

Hypothesis Status Agree with Prediction?

H1 (null): The positive abnormal audit fees (POSAAF) do not have a relation with earnings management, ceteris paribus.

Rejected Yes

H1 (alternative): The positive abnormal audit fees (POSAAF) have a positive relation with earnings management, ceteris paribus.

Accepted Yes

H2 (null): The positive abnormal audit fees (POSAAF) do not have a relation with accounting conservatism, ceteris paribus.

Rejected Yes

H3 (null): The negative abnormal audit fees (NEGAAF) do not have a relation with earnings management, ceteris paribus.

Accepted No

H3 (alternative): The negative abnormal audit fees (NEGAAF) have a negative relation with earnings management, ceteris paribus.

Rejected No

H4 (null): The negative abnormal audit fees (NEGAAF) do not have a relation accounting conservatism, ceteris paribus.

Rejected Yes

Page 16: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

16

Version 1.0 DBA VIVA Presentation 28 July 2008 31

Additional TestsAutocorrelationLowballingSOX effectsC-Score on different groups of firmsDropping firms with marginal POSAAF and NEGAAFNonaudit fees and abnormal audit fees…… Results are substantially the same as the main test!

Version 1.0 DBA VIVA Presentation 28 July 2008 32

6. Conclusions

Page 17: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

17

Version 1.0 DBA VIVA Presentation 28 July 2008 33

Concluding remarksThis study shows association between abnormal audit fees and audit quality and the nonlinearity of abnormal audit fee metrics. But we cannot infer POSAAF must cause lower audit quality. (Higher audit effort will also cause POSAAF).POSAAF may not be the only source of economic bond. It will be a challenge for researchers to find a more comprehensive model to capture factors affecting the economic bond.A funny paradox --- When the association of POSAAF and audit quality is more exposed, this will make such association weaker due to the shifting of channels of benefits delivery.

Version 1.0 DBA VIVA Presentation 28 July 2008 34

Contributions of this studyi) Provide empirical evidence that earnings management is positively associated with POSAAF; ii) Provide empirical evidence that conservatism is positively associated with POSAAF; iii) Provide empirical evidence on the nonlinear nature of abnormal audit fees; iv) Provide additional evidence on the research on accounting conservatism using the latest measure of conservatism (C-Score); and v) Provide additional empirical evidence from Hong Kong on the research of audit quality.

Page 18: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

18

Version 1.0 DBA VIVA Presentation 28 July 2008 35

LimitationsAudit quality can be expressed in many contexts. Using accrual quality and conservatism as measures for audit quality may omit other important contexts.Although the accrual quality measure of Dechow and Dichev (2002)and the C-Score of Khan and Watts (2007) are well accepted as good measures, one cannot rule out the possibility of measurement errors.Even though the audit fee model of this study has an adjusted R2 of 0.733, there is still a possibility that important variables are omitted from the model. For example, nonaudit service fees, and various corporate governance variables are totally ignored from the measuring of abnormal audit fees. Hence, the POSAAF and NEGAAF may be estimated incorrectly. The model determining abnormal audit fees only takes in variables that relate to the supply of audit services and assuming all demand variables are uniform for all firms. It may not be appropriate to assume that the demand variables are uniform for all firms. The small sample size and the sampling method may create unknown bias in the results.

Version 1.0 DBA VIVA Presentation 28 July 2008 36

7. Q&A

Page 19: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

19

Version 1.0 DBA VIVA Presentation 28 July 2008 37

Supporting slides

Version 1.0 DBA VIVA Presentation 28 July 2008 38

What is independence ?“the conditional probability of reporting a discovered breach”(DeAngelo, 1981a);“the ability to resist client pressure” (Knapp, 1985);“an attitude/state of mind” (Moizer, 1994);“A function of character, with the attributes of integrity and trustworthiness“ (Magill and Previts, 1991);“An absence of interests that create un acceptable risk of bias” (AICPA, 1997); and“Freedom from those pressures and other factors that compromise, or can reasonably be expected to compromise, an auditor’s ability to make unbiased audit decision” (ISB, 2000)In summary, they all relate to Objectivity and Integrity(Dunmore and Falk, 2001)

Page 20: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

20

Version 1.0 DBA VIVA Presentation 28 July 2008 39

The performance of auditorsIn the US, the total audit fees for 5500 largest publicly listed firms in 2003 was US$ 3.4 billion which represents 0.04% of sales and 0.03% of the market capitalization of these firms, while outright audit failure is infrequent, far less than 1% annually (Francis,2004).

27% of bankruptcies are preceded by going concern audit reports and 47% of bankruptcies are preceded by unqualified audit reports. This indicates that 73% of the bankrupted companies are issued with ‘false negative’ reports (Carcello and Palmrose,1994).

During 1990 to 1994, 1003 going concern audit reports were issued while only 143 among these were finally bankrupted. This indicated that 85% of the going concern reports are ‘false positive’ (Francis and Krishnan, 2002)

Version 1.0 DBA VIVA Presentation 28 July 2008 40

Interaction of abnormal fees metrics and audit quality

Demand of high audit effort

High audit fee

(POSAAF)High Audit Quality

Creation of Economic Bond

High audit fee

(POSAAF)Low Audit Quality

Low audit fee

(NEGAAF)Low Audit Quality

Demand of low audit effort

Page 21: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

21

Version 1.0 DBA VIVA Presentation 28 July 2008 41

Explanations for the positive relation between C-Score and POSAAF

First, it may be due to managerial bonding, which caused higher conservatism and demand of additional audit effort. Second, C-Score measures the asymmetric timeliness of responses to good and bad news. It does not indicate that a high level of conservatism is associated with high earnings quality. In fact, high level of conservatism is a sign of weak earnings quality (Schipper and Vincent, 2003) ! Third, one should not assume that opportunistic earnings management activities will always be making the financial statements less conservative.

Version 1.0 DBA VIVA Presentation 28 July 2008 42

7 attributes of Earnings Quality (“EQ”) There are many definitions for EQ. Schipper and Vincent (2003) suggest EQ as the extent to which reported earnings faithfully represent the Hicksian income [1] . Since there is no uniform definition for EQ, there are numerous measures of it. At times some of these measures are contradictory. In order to get a better picture, it is advisable to look at a number of measures (Abdelghany, 2005).Francis et al. (2004) summarize 7 attributes for EQ. They are: accrual quality, persistence, predictability, smoothness, value relevancy, timeliness, and conservatism. The first 4 are accounting based attributes and the last 3 are market based attributes

[1] Hicksian income is defined as the amount that can be paid out from a firm during a period and leaving the firm equally well off at the beginning and the end of the period (Hicks 1939, 176).

Page 22: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

22

Version 1.0 DBA VIVA Presentation 28 July 2008 43

POSAAF estimationVariable Exp

signs Ceofficients t p-value

(Constant) N/A -2.988 -19.240 0.000LNTA (+) 0.399 21.652 0.000INVREC (+) 0.553 5.142 0.000EMPLOY (+) 0.004 11.761 0.000ISSUE (+) 0.028 0.646 0.518FOREIGN (+) 0.311 8.262 0.000EXORD (+) 0.115 3.429 0.001LOSS (+) 0.254 3.784 0.000LEVE (+) 0.198 1.937 0.053ROA (-) 0.002 1.000 0.317CURRENT (-) -0.021 -2.983 0.003Big4 (+) -0.052 -1.049 0.295PB (+) 0.013 2.171 0.030AUDITSPEC (+) 0.100 2.957 0.003Yearenddummy N/A -0.144 -3.801 0.000

n 1,129

Adjusted R2 0.733

Version 1.0 DBA VIVA Presentation 28 July 2008 44

POSAAF and NEGAAF by year

Average POSAAF & NEGAAF by year

0.0000

0.2000

0.4000

0.6000

0.8000

1.0000

1.2000

1.4000

1.6000

1.8000

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

POSAAF NEGAAF

Page 23: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

23

Version 1.0 DBA VIVA Presentation 28 July 2008 45

Mean DDAQ by year

Mean DDAQ By year

0.00200.00400.00600.00800.00

1000.001200.001400.001600.001800.00

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

DD

AQ

Mean DDAQ

Version 1.0 DBA VIVA Presentation 28 July 2008 46

Mean C-Score by yearMean C-Score by year

-1.50

-1.00

-0.50

0.00

0.50

1.00

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Mean C-Score

Page 24: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

24

Version 1.0 DBA VIVA Presentation 28 July 2008 47

Additional test - Autocorrelation

NEGAAF coefficient t p-value Durbin-Watson

OLS 26.830 0.860 0.390 1.615Prais-Winsten 15.816 0.507 0.612 2.008

POSAAF coefficient t p-value Durbin-Watson

OLS 0.260 2.042 0.042 1.231Prais-Winsten 0.212 1.851 0.065 2.221

NEGAAF coefficient t p-value Durbin-Watson

OLS 0.133 3.106 0.002 1.240Prais-Winsten 0.132 3.095 0.002 1.999

AAF coefficient t p-value Durbin-Watson

OLS 0.141 3.227 0.001 1.135Prais-Winsten -0.001 -0.483 0.629 0.715

DDAQ

C-Score

Version 1.0 DBA VIVA Presentation 28 July 2008 48

Additional test - LowballingFor DDAQ regressions

Coefficient t p-value n Adj R2

POSAAF 623.65 3.79 0.000 428 0.304POSAAF 3-year-mean 532.39 1.93 0.055 210 0.312

NEGAAF 26.83 0.86 0.390 462 0.303NEGAAF 3-year-mean 24.41 0.38 0.708 224 0.379

AAF 56.93 4.52 0.128 890 0.293AAF 3-year-mean 73.63 1.15 0.252 434 0.342

For C-Score regressions

POSAAF 0.26 2.04 0.042 374 0.128POSAAF 3-year-mean 0.49 3.22 0.002 195 0.263

NEGAAF 0.13 3.11 0.002 409 0.075NEGAAF 3-year-mean 0.14 1.81 0.072 224 0.125

AAF 0.14 3.23 0.001 783 0.095AAF 3-year-mean 0.17 2.57 0.011 419 0.165

Page 25: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

25

Version 1.0 DBA VIVA Presentation 28 July 2008 49

Additional test- SOX effectDDAQ

Coefficient t p-value n R2POSAAF pre-SOX 387.563 1.695 0.092 214 0.211POSAAF post-SOX 819.714 3.377 0.001 214 0.388POSAAF-Overall 623.649 3.789 0.000 428 0.304

NEGAAF pre-SOX 9.053 0.233 0.816 229 0.347NEGAAF post-SOX 51.624 1.037 0.301 233 0.305NEGAAF-Overall 26.832 0.860 0.390 462 0.303

C-ScoreCoefficient t p-value n R2

POSAAF pre-SOX 0.114 0.504 0.615 173 0.122POSAAF post-SOX 0.340 2.647 0.009 201 0.270POSAAF-Overall 0.260 2.042 0.042 374 0.128

NEGAAF pre-SOX 0.164 1.925 0.056 188 0.089NEGAAF post-SOX 0.098 2.412 0.017 221 0.129NEGAAF-Overall 0.133 3.106 0.002 409 0.075

pre-SOX = before 2003post-SOX = 2003 onwards

Version 1.0 DBA VIVA Presentation 28 July 2008 50

Additional test - C-score comparison

Firm attributes n MeanC-Score t P-value

HSI 259 -0.7848Non-HSI 1665 0.3252 -13.980 0.000

PRC 459 0.0222non-PRC 1465 0.2239 -3.031 0.002

Big firms 663 -0.2230Small firms 1261 0.3855 -10.451 0.000

Page 26: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

26

Version 1.0 DBA VIVA Presentation 28 July 2008 51

Additional test – DDAQ and 90% abnormal audit fees metrics

DV=DDAQ Section APOSAAF

Section B90%

POSAAF

Section CNEGAAF

Section D90%

NEGAAF

Section EAAF

Section F90% AAF

Variable Expsign

Coefficient(t value)

Coefficient(t value)

Coefficient(t value)

Coefficient(t value)

Coefficient(t value)

Coefficient(t value)

(Constant) N/A -4532.69 -3443.81 -3077.81 -2595.15 -3191.20 -1968.90(-8.80***) (-7.67***) (-10.35***) (-9.33***) (-13.30***) (-11.26***)

SIZE (-) 558.21 433.51 410.80 355.37 426.14 277.32(11.02***) (9.38***) (13.08***) (11.62***) (17.19***) (14.59***)

OPCYCLE (+) -0.27 -0.17 -0.02 0.01 -0.05 0.02(-1.47) (-1.32) (-0.17) (0.16) (-0.60) (0.28)

SALESV (+) 78.82 60.22 290.39 200.60 214.19 158.26(0.33) (0.35) (1.70*) (1.38) (1.50) (1.72*)

LOSSDUMMY (+) 737.46 526.45 487.72 382.77 612.53 279.66(4.86***) (4.76***) (4.19***) (3.72***) (6.44***) (4.36***)

ABFees 623.65 467.86 26.83 19.96 56.93 36.24(3.79***) (3.75***) (0.86) (0.76) (-1.52) (1.54)

n 428 390 462 413 890 796

Adj R2 0.3040 0.2490 0.3030 0.2680 0.2930 0.2350key

*** 1 % significance** 5 % significance

* 10 % significancet statistics in parentheses

Version 1.0 DBA VIVA Presentation 28 July 2008 52

Additional test – C-Score and 90% abnormal audit fees metrics

DV=C-Score ExpSign

Section APOSAAF

Scetion B90%

POSAAF

Section CNEGAAF

Section D90%

NEGAAF

Section EAAF

Section F90% AAF

Coeffieicent(t value)

Coeffieicent(t value)

Coeffieicent(t value)

Coeffieicent(t value)

Coeffieicent(t value)

Coeffieicent(t value)

(Constant) 0.441 0.379 0.106 0.065 0.294 0.285(0.843) (0.762) (0.261) (0.148) (0.914) (0.896)

C-score t-1 (+) 0.165 0.218 0.032 0.046 0.098 0.111(4.303***) (5.443***) (0.952) (1.339) (3.894***) (4.269***)

DDAQ t (-) 0.000 0.000 0.000 0.000 0.000 0.000(-2.593***) (-1.147) (-3.6***) (-3.17***) (-4.471***) (-2.415**)

VOR t (+) -0.053 -0.016 -0.044 -0.045 -0.055 -0.025(-1.373) (-0.348) (-1.118) (-0.821) (-1.989**) (-0.683)

Age (-) -0.026 -0.026 -0.005 -0.005 -0.014 -0.017(-0.504) (-0.524) (-0.128) (-0.117) (-0.429) (-0.537)

Cycle t (-) -3.007 -2.152 -0.494 0.264 -1.916 -1.659(-1.415) (-0.942) (-0.205) (0.101) (-1.22) (-0.983)

ABFees t 0.260 0.229 0.133 0.139 0.141 0.137(2.042**) (1.869*) (3.106***) (3.153***) (3.227***) (3.189***)

n 374 340 409 367 783 699

Adj R2 0.128 0.111 0.075 0.063 0.095 0.056

key 1 % significance5 % significance10 % significance

Page 27: VIVA version 2 - polyudbaaa.compolyudbaaa.com/wp-content/uploads/2016/11/2008-Alan-Ho-VIVA-ve… · 3 Version 1.0 DBA VIVA Presentation 28 July 2008 5 The research gap Results of

27

Version 1.0 DBA VIVA Presentation 28 July 2008 53

Additional test - Nonaudit fees

Pearsoncorrelation ANF AAF POSANF POSAAF NEGANF NEGAAF

ANF 1-

AAF 0.0027 1(0.4891) -

POSANF 1.0000 0.4325 1(0.0000) (0.0013) -

POSAAF 0.3122 0.0714 0.7011 1(0.0121) (0.2542) (0.0003) -

NEGANF 0.9959 -0.0293 N/A 0.3467 1(0.0000) (0.4158) N/A (0.0259) -

NEGAAF 0.1422 0.2507 0.2560 N/A 0.2354 1(0.1623) (0.0069) (0.1034) N/A (0.1341) -