septian bayu k. (0806479080) detecting earnings management dechow, sloan, sweeney (1995)

Post on 21-Jan-2016

212 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

SEPTIAN BAYU K. (0806479080)

Detecting Earnings Management

Dechow, Sloan, Sweeney (1995)

Outline

IntroductionStatistical BackgroundMeasuring DAExperimental DesignData AnalysisEmpirical ResultsConclusionsImplications

Introduction (1)

Analysis of earnings management focuses on discretionally accruals (DA) Separate total accruals to DA & NDA

The aim of research Finding the sophisticated model(s) to measure to

detect earnings management with DA/NDA

Research gap Modified Jones Model

Introduction (2)

Prior research DA: Healy (1995), DeAngelo (1996), Jones (1991) Accounting procedure changes: Healy (1985), Healy &

Palepu (1990), Sweeney (1994) Specific components of DA: McNichols & Wilson

(1988), DeAngelo et al (1994) Components of Discretionary Cash Flow (Dechow &

Sloan (1991)

Statistical Background

McNichols & Wilson (1988)

Problems: Incorrectly attributing earnings management to

PART Unintentionally extracting earnings management

caused by PART Low power test

Measuring DA (1)

The Healy model (Healy, 1985)

The DeAngelo model (DeAngelo, 1986) NDAτ = TAτ-1

The Jones model (Jones, 1991)

Measuring DA (2)

The Modified Jones model

The Industry model (Dechow & Sloan, 1991) NDA τ = γ1 + γ2 median1 (TA τ)

Experimental Design

Randomly 1000 firm-years (1950-1991)Firm-years experiencing extreme financial

performanceFirm-years with accrual manipulation

Expense manipulation Revenue manipulation Margin manipulation

32 firms that are subject to SEC enforcement actions

Data Analysis

Total accruals (TA)

CFO = Earnings – TA Using Z-statistic

Empirical Results

Random sample of firm-years Table 1, table 2

Samples of firm-years experiencing extreme financial performance Figure 1, table 3, figure 2, table 4

Samples of firm-years with artificially induced earnings management Figure 3, figure 4

Sample of firm-years in which of the SEC alleges earnings are overstated Figure 5, table 5, table 6, table 7

Conclusions

All of models appear well specified when applied to random sample of the firm-years

The models all generate test of low power of earnings management

All models reject the null hypothesis of no earnings management

Modified Jones model generate the revenue –based earnings management

Implications

Regardless of the model used to detect earnings management Further research: develop new model with more

powerful test to detect earnings management

Correlation between PART ad firm performance, considered the models

Consider about earnings management context

top related