Analyzing Earnings Quality as a Financial Forensics Tool

fVLe issue 39 october/november 2012 Page 16

forensics/fraud - earnings Quality and the Beneish Model

why Earnings Qualityearnings quality is an important aspectof evaluating a company’s financialhealth. It has been studied since atleast 1934, as referenced in Grahamand dodd’s Security Analysis. Indeed,Warren buffett considers the book’sguidance as one of the four most treas-ured books that he owns.1 earningsQuality relates to the ability of a com-pany’s reported earnings to best sym-bolize its true earnings. earnings qual-ity also refers to the stability, or lackthereof, in a company’s reported earn-ings.

analysts, investors and manage-ment have deployed dozens of forensicindices that aid the forensic accountantin assessing the probability of earningsmanipulation by a suspect company.The wide variety of forensic indiceswas developed to evaluate earningsquality, and identify whether a compa-ny’s financial statements may poten-tially reflect manipulated earnings.The quality of a company’s earnings isone facet of an investigation that isoften overlooked in the financial foren-sic process.

This article discusses variousaspects of earnings quality, and dis-cusses a method for detecting earningsmanipulation. The beneish M-ScoreModel (beneish Model), deployed as afinancial forensic tool, can assist inevaluating the probability of earningsmanipulation in a company, as well asidentifying areas that may requiregreater scrutiny.

DEtErmining Earnings Quality earnings quality focuses on the extentto which the reported net income devi-ates from the actual, true earnings of acompany. a review of a company’searnings over time may be an indica-

tion of how consistent, and repeatable,the earnings are for a particular periodand in the future. However, as thefinancial statements are the responsi-bility of the company’s management,transactions can be structured to bestachieve a desired accounting result byreporting key financial transactions tothe company’s advantage. Therefore,the impact of a given transaction canvary from company to company, aswell as between different industries. Inaddition, the earnings determinationprocess becomes more problematic,and increasingly difficult to bench-mark, as a result of the myriad report-ing practices that individual compa-nies employ to report profits, such asoperating earnings, net income andincome before taxes. all of these differ-ing factors create variability, whichcould increase the chances that a com-pany’s reported earnings may havebeen manipulated, and not uncovered.

Some examples of accountingmanipulations to be aware of whenexamining financial statement data fora company’s earnings quality include:

• Recording revenue too soon or with questionable quality

• Recording fictitious revenue• boosting income with one-time

gains• Shifting current expense to a

different period• capitalizing otherwise currently

recognizable expenses• failing to record, or improperly

reducing, liabilities

The earnings reported by a companyare a result of its accounting methods,in conjunction with its customary busi-ness practices. In this connection,accruals function as temporary adjust-ments to address the timing of compa-

ny transactions by providing manage-ment’s best assumptions and estimates.as a result, accruals could representacceptable accounting estimates, basedon rational assessments of the future,or a prime management tool for com-mitting earnings manipulation.although large accruals are not auto-matically an indication of fraud, theforensic accountant should be awarethat a fundamental component ofmany financial statement frauds is cen-tered on earnings manipulation utiliz-ing accruals. It is often difficult to dif-ferentiate between the two, but themagnitude and purpose of accrualscan be used as a barometer to assist theforensic accountant in the detection ofearnings manipulation.

The balance sheet, income state-ment and statement of cash flows areinterrelated and should be analyzedtogether. When the numbers withinthese statements do not make sense, amore in-depth investigation is

Earnings Quality

expertTIPThe Beneish M-Score Model

(Beneish Model), deployed as a

financial forensic tool, can assist

in evaluating the probability of

earnings manipulation in a compa-

ny, as well as identifying areas that

may require greater scrutiny.

Continued on next page

Analyzing

as a Financial Forensic Tool

mark warshavsky,cpa/abv/cff, cva, cba, aSa,

cfe, dabfa, cffa

Gettry Marcus Stern & LehrerWoodbury, NY

[email protected]


forensics/fraud - earnings Quality and the Beneish Model, continued

required. Generally, the statement ofcash flows may be considered a moreobjective measure of performance, anda better predictor of a company’s trueearnings and future stability. a dispar-ity between a company’s reported highnet income and its low cash flowshould prompt an investigation.

DEtEcting Earnings manipulation One quantitative forensic indexingmethod developed to assist in detect-ing whether a company has manipulat-ed its earnings is the beneish Model,developed by professor Messodbeneish.2 The beneish Model is similarto the altman Z-Score,3 but is used touncover earnings manipulation ratherthan as a predictor of bankruptcy.companies with higher beneish scoresare more likely to be manipulators. Inhis study, beneish’s Model was able toidentify as many as three quarters ofthe companies who had manipulatedtheir earnings. The beneish Model canbe calculated by using a company’sannual financial statements, and differ-entiates manipulators from non-manipulators by applying variablesfrom the financial statements. furtherinvestigation must be completed toidentify potential legitimate causes forresulting abnormalities.

The beneish Model uses eightfinancial variables viewed as beingindicators of companies prone to finan-cial statement manipulation. There isan increased probability for manipula-tion when company financial state-ments reveal statistically significantchanges in accounts receivable, deteri-orating gross margins, decreasing assetquality, sales growth and increasingaccruals, among others.

The eight variables employed inthe beneish Model are presentedbelow as “indexes” to avoid confusionwith financial “ratios” (note: the cur-rent year financial statement numbersare indicated as “cy” and the prior yearfinancial statement numbers are indi-cated as “py”):1) Days’ Sales in Receivables Index(dSRI) – Measures the ratio of the days


that sales are inaccounts receivable,and benchmarks thisratio against the prioryear. a result of greater than 1.0 wouldindicate that accounts receivable, as apercentage of sales, has increased fromthe prior year. Therefore, although adisproportionate increase in accountsreceivable relative to sales could be theresult of a company’s normal opera-tions, such as legitimate end-of-monthsales or general business credit deci-

sions, it may also be indicative ofinflated revenues.2) Gross Margin Index (GMI) – Meas-ures the ratio of a company’s prioryear’s gross margin to the currentyear’s gross margin. The company’sgross margin has deteriorated whenthe results are greater than 1.0. Grossmargin deterioration is a negative indi-cator of a company’s prospects, mak-ing such companies more prone tomanipulate earnings.

3) Asset Quality Index (aQI) – Measuresthe quality of a company’s assets bycalculating the ratio of non-currentassets, other than plant, property andequipment (ppe), to total assets. Itindicates the amount of total assetsthat are less certain to be ultimatelyrealized, identified as asset quality. anaQI greater than 1.0 indicates that thecompany has potentially increased itscost deferral or increased its intangibleassets, and created earnings manipula-tion. Therefore, the greater the aQI,indicating a reduction in asset quality,the greater the probability of earningsmanipulation.

4) Sales Growth Index (SGI) – a result ofgreater than 1.0 represents salesgrowth compared to that of the prioryear. Sales growth itself is not indica-tive of earnings manipulation, howev-er, growth companies are more likelyto commit earnings manipulation.

5) Depreciation Index (depI) – a depIgreater than 1.0 may be an indicationof an upward revision of the estimatedlives of a company’s property, plantand equipment, which would increaseits income.

6) Sales, General and AdministrativeExpenses Index (SGAI) – Measures theratio of a company’s SGaI to sales. adisproportionate increase in sales, ascompared to SGaI, would serve as anegative indication concerning a com-pany’s future prospects.

7) Total Accruals to Total Assets Index(TATA) – accruals, calculated in thisformula as working capital4 other thancash, are estimates of the short termforecasted inflow and outflow activi-ties of a company. excluding anymajor changes within the company,these accruals should be fairly consis-tent within some acceptable measureof statistical variation. However,accruals have consistently provided aconventional opportunity to perpetrate

Accounts Receivable (cy)

Sales (cy)

Accounts Receivable (py)

Sales (py)

Depreciation Expense (py)

Depreciation Expense (py) + PPE (py)

Depreciation Expense (cy)

Depreciation Expense (cy) + PPE (cy)

Sales (cy)

Sales (py)

1-(Current Assets (cy) + PPE (cy)) / Total Assets (cy)

1-(Current Assets (py) + PPE (py)) / Total Assets (py)

Sales (py) – Cost of Sales (py)

Sales (py)

Sales (cy) – Cost of Sales (cy)

Sales (cy)

Sales, General and AdministrativeExpenses (cy)

Sales (cy)

Sales, General and AdministrativeExpenses (py)

Sales (py)



a fraud. as a result, higher positiveaccruals are associated with the poten-tial for earnings manipulation, asshown above.8) Leverage Index (LvGI) – Measures theratio of a company’s total debt to totalassets. When the LvGI is greater than1.0, it indicates an increased leverageand, therefore, a company more proneto financial statement manipulation.

although each variable can be usedindividually to assess a specific com-ponent of a company’s financial state-ment data, the beneish Model takes acompany’s results of these eight vari-ables and applies them in the follow-ing formula:

The figure of -4.84 is applied as a con-stant in the formula, and each of theeight variables is multiplied by itsrespective constant number. Whenapplying the beneish Model, a score ofgreater than -2.22 (i.e., less of a nega-tive) is an indication that the compa-ny’s financial statements may havebeen manipulated.

application of thE BEnEishm-scorE moDElThe following two companies havebeen well-publicized for earningsmanipulation of their respective finan-cial statements. The forensic indicesfrom the beneish Model had beenapplied to these two companies for a

specific year, prior to when the compa-ny fraud was reported to the public.The results are quite interesting, andrevealing, from the perspective of afinancial forensic application.

Enron corporationenron corporation (enron) represent-ed one of the largest fraud scandals inthe history of the United States. beforeits collapse in 2001, enron was number7 on the Fortune list of the 500 largestcompanies. at that time, enron wasknown as one of the most advanced,and well-regarded, businesses in theworld.

enron employed an assortmentof deceptive and fraudulent account-ing practices to obscure its actualfinancial position. However, therewere financial forensic warning signsrevealed in the company’s financialstatements prior to its downfall andeventual bankruptcy filing on decem- Continued on next page

ber 2, 2001. Two of these warning signswere how enron measured, andreported, its total revenues, and itsnegative operating cash flows. How-ever, the biggest red flag in enron’sfinancial statements were signs of poorearnings quality as indicated by sever-al key forensic measures.

The application of the beneishModel to enron’s financial statementsindicated that the company may havebeen manipulating its earnings as farback as 1997. The calculations for thefirst two columns of Table 1, below,represent the beneish findings5 andcategorize the companies into twogroups, non-manipulators and manip-ulators. The column to the right, titledenron, represents the results of apply-ing the beneish Model to enron’s 1997financial statements.

Table-1 demonstrates that enronscored close to, or higher, than manip-ulators in three of the eight variables:Gross Margin Index (GMI), assetQuality Index (aQI) and Sales GrowthIndex (SGI).

(Working Capital (cy) – (py)) – (Cash (cy) – (py)) + (Income Tax Payable (cy) – (py))+ (Current Portion of Long Term Debt (cy) – (py)) – Deprec. & Amort. Expense (cy)

Total Assets (cy)

Long Term Debt (cy) +

Current Liabilities (cy)

Total Assets (cy)

Long Term Debt (py) +

Current Liabilities (py)

Total Assets(py)

M = -4.84 + (0.92*DSRI)

+ (0.528*GMI) + (0.404*AQI)

+ (0.892*SGI) + (0.115*DEPI)

- (0.172*SGAI) + (4.679*TATA)

- (0.327*LVGI) VariablesNon-

ManipulatorsManipulators Enron7

Days in Sales in Receivables (DSRI) 1.031 1.465 0.625

Gross Margin (GMI) 1.014 1.193 1.448

Asset Quality (AQI) 1.039 1.254 1.308

Sales Growth (SGI) 1.134 1.607 1.526

Depreciation (DEPI) 1.001 1.077 1.017

Sales, General and Administrative (SGAI) 1.054 1.041 0.649

Total Accruals to Total Assets (TATA) 0.018 0.031 0.012

Leverage (LVGI) 1.037 1.111 1.041

Table 1

Enron Corporation

Benchmarking to the Beneish M-Score Model

Per Beneish-Mean6



was sentenced to 25 years in prison forsecurities fraud, money laundering,tax evasion, and other charges.

Table 3 below is a summary ofselected variables calculated from Z-best’s 1986 financial statements, andcompares the results of four mean vari-ables from the beneish Model.

Days’ Sales in Receivables Index formanipulators, according to the beneishModel, had a mean index of 1.465, andZ-best’s index was an incredible177,622. Z-best had no accountsreceivable in 1985, but reported almost$700,000 in 1986. It was determinedlater that the 1986 accounts receivablewere fictitious.

The mean Asset Quality Index is1.039 for non-manipulators and 1.254for manipulators. Z-best had an indexof 2.043, or 63 percent higher than themean for manipulators, and 97 percenthigher than the mean for non-manipu-lators. Z-best’s aQI of 2.043 at the endof 1986 was the result of increased non-current (soft) assets. Z-best, as withmany companies that commit fraud,was guilty of capitalizing certainexpenditures as deferred costs. atdecember 31, 1985, Z-best’s non-cur-rent assets were approximately 40 per-cent of its total assets. by december 31,

based upon its financial statements,the gross margin for enron decreasedfrom 21.15 percent in 1996 to 14.61 per-cent in 1997, and then to 6.22 percentin 2000.8 enron’s Gross Margin Index of1.448 exceeded the mean index of 1.193for the manipulators’ category by 21.4percent.

The Asset Quality Index for enronof 1.308 was higher than the manipula-tors mean index of 1.254, and 25 per-cent greater than the mean of 1.039 fornon-manipulators.

The Sales Growth Index for enronwas 1.526, which represented an indexclose to the mean of manipulators asdetermined by the beneish Model.enron used inflated revenues as itsstrategy in order to give the impressionto outside investors that the companywas an innovative, high-growth invest-ment. In the four-year period prior toits collapse, enron’s reported revenuesincreased from $13.3 billion in 1996 to$100.8 billion in 2000, representing anincredible 750 percent gain. Thegrowth of enron’s reported revenuesfor the period 19999 to 2000, from $40.1billion to $100.8 billion, was 151 per-cent and unprecedented in any indus-try.

The overall result for enronwhen applying the beneish Model cal-culations was -1.89, which is greaterthan the -2.22 standard used to meas-ure the likelihood of manipulation.This result places enron as a potentialcandidate of financial statement earn-ings manipulation in 1997, years beforethe fraud became public.

The detailed calculations forenron, using the beneish Model, arepresented in Table 2 at top right.

ZZZZ Best company, inc.another real life example of financialstatement earnings manipulation wasthe ZZZZ best company, Inc. (Z-best),and the activities of its founder barryMinkow. Z-best appeared to be anextremely successful carpet-cleaningand restoration company, but was afaçade for a massive ponzi scheme. InJanuary 1986, Z-best went public.Three years later, in 1989, Mr. Minkow

Table 2

Calculation of the Beneish M-Score Model

Applied to Enron’s 1997 and 1996 Financial Statements

M = -4.84 +0.92*DSRI + 0.528*GMI + 0.404*AQI + 0.892*SGI

+ 0.115*DEPI – 0.172*SGAI + 4.679*TATA – 0.327*LVGI

M = -4.84 + (0.92*0.625) + (0.528*1.448) + (0.404*1.308)

+ (0.892*1.526) + (0.115*1.017) – (0.172*0.649) + (4.679*0.012)

– (0.327*1.041)

M = -4.84 + .5750 + .7645 + .5284 + 1.3612 + .1170 - .1116 + .0561

- .3404

M = (1.8898) rounded to (1.89) = greater (less of a negative) than (2.22)

Table 3

ZZZZ Best Company, Inc.

Benchmarking to the Benish M-Score Model

Per Beneish-Mean10

Variables Non-Manipulators Manipulators Z-Best11

Days in Sales in Receivables (DSRI) 1.031 1.465 177,622

Asset Quality (AQI) 1.039 1.254 2.043

Sales Growth (SGI) 1.134 1.607 3.905

Total Accruals to Total Assets (TATA) 0.018 0.031 0.064



forensics/fraud - earnings Quality/Beneish Model, continued front Page - continued

1986, the non-current assets represent-ed over 65 percent of Z-best’s totalassets.

The mean Sales Growth Index fornon-manipulators and manipulatorswas 1.134 and 1.607, respectively. Z-best had an astonishing index of 3.905,or a 143 percent increase over themean manipulators. Sales growth byitself is not an indicator of fraud. How-ever, Z-best, being a company in agrowth mode, was more inclined tocommit fraud due to the pressuresplaced upon it to continue its patternof rapidly growing sales. Salesincreased from $1,240,534 in 1985 to$4,845,347 in 1986, almost 300 percent.Not surprisingly, fictitious sales werepart of Z-best’s fraudulent financialreporting.

Total Accruals to Total Assets fornon-manipulators had a mean index of0.018, and the manipulators’ categoryhad a mean index of 0.031. Z-best’sindex of 0.064 was over 100 percenthigher than the mean manipulators’category.

conclusionQuality of earnings continues to be animportant element in the investigationof a company’s financial statements.The accrual quality, a component ofearnings quality, can be determined bythe extent these accruals ultimately arerealized in the company’s cash flows.

1 Warshavsky, Mark S., contributing author to Financial

Forensics Body of Knowledge, Dorrell & Gadawski;

John Wiley & Sons, Inc., New York, 2012.2 Beneish, M.D., “The Detection of Earnings Manipula-

tion,” Financial Analyst Journal: 24-36, June, 1999. 3 Altman, Edward I., “Financial Ratios, Discriminant

Analysis and the Prediction of Corporate Bankruptcy,”

The Journal of Finance, September, 1968, Vol. 23, No.

4, pages 589-609.4 The term “working capital” is mentioned for brevity as

Professor Beneish’s TATA formula does not use this

term. Instead, his formula uses the actual formula for

working capital of current assets minus current liabili-

ties.5 Beneish, M.D., “The Detection of Earnings Manipula-

tion,” Financial Analyst Journal: 24-34, June, 1999,

Table 2.6 Beneish, M.D., “The Detection of Earnings Manipula-


Table 2.7 Enron Corporation financial statements for the years

1996 and 1997.8 Enron Corporation financial statements for the year

2000.9 Enron Corporation financial statements for the year

1999.10 Beneish, M.D., “The Detection of Earnings Manipula-


Table 2.11 ZZZZ Best Company, Inc. financial statements for the

years 1985 and 1986.

Tuesday, May 10, 2011transaction Databases update -pratt’s stats, BiZcomps and the iBa(Jim Hitchner, Marci bour)

1. Which of the transaction databas-es do you use for small companiessuch as less than $1 million? (can pick more than one)a. bIZcOMpS – 72%b. donedeals – 11%c. Iba – 42%d. Mergerstat – 10%e. pratt’s Stat – 74%

2. Which of the transaction databas-es do you use for larger compa-nies such as greater than $20 mil-lion? (can pick more than one)a. bIZcOMpS – 28%b. donedeals – 28%c. Iba – 13%d. Mergerstat – 42%e. pratt’s Stat – 82%

3. If you only have the usual mini-mal data concerning a transactiondo you usually rely on transactiondatabases as: a. a primary method and value

– 2%b. a secondary or corroborating

method and value – 76%c. dismiss the method and value

as unreliable – 22%

4. Have you ever had troubledefending your use of transactiondatabases where there was onlyminimal data?a. Yes – 51%b. No – 49%

5. do you use the Iba direct Marketdata Method?a. Yes – 33%b. No – 67%

6. Where there is minimal data onthe transactions what is the mini-mum range of transactions thatcan be relied upon?a. Zero – 3% b. One or two – 1%c. Three to five – 19%d. Six to ten – 50% e. More than ten – 26% c

The beneish Model starts with utiliz-ing the existing analytical ratios andapplies its indexes to develop a strongpredictor of earnings manipulation.The strength of the beneish Model isthat it applies eight unique indices,both individually and collectively.Therefore, the forensic accountant candeploy these findings to develop aroadmap for further investigation. c

Documents

Analyzing Earnings Quality as a Financial Forensics Tool