6.8 Forensic accounting: Beneish, Piotroski, and an Indian checklist

Motivation

A DCF assumes the numbers are true. In India the largest permanent losses for minority shareholders have come from governance failures and accounting fraud, not from paying 10% too much. Screening for manipulation is part of valuation.

Schilit's taxonomy (Financial Shenanigans, 4th ed.). Earnings manipulation (recording revenue too soon or bogus revenue; boosting income with one-time items; shifting expenses to later periods; hiding expenses; shifting income or expenses between periods). Cash-flow manipulation (financing inflows shown as operating; operating outflows shown as investing; CFO inflated by acquisitions or by unsustainable working-capital moves such as factoring receivables). Key-metric manipulation. Acquisition-accounting manipulation.

6.8.1Beneish M-score#

Definition

Beneish ("The Detection of Earnings Manipulation", Financial Analysts Journal 55(5), 1999, 24–36) estimated a probit model on manipulators and controls. With year $t$ and year $t-1$ data:

Index	Definition	What a high value signals
DSRI, days' sales in receivables index	$\dfrac{\text{Rec}_t/\text{Sales}_t}{\text{Rec}_{t-1}/\text{Sales}_{t-1}}$	Revenue booked but not collected
GMI, gross margin index	$\dfrac{GM_{t-1}}{GM_t}$ , $GM = (\text{Sales} - \text{COGS})/\text{Sales}$	Deteriorating margins: pressure to manipulate
AQI, asset quality index	$\dfrac{AQ_t}{AQ_{t-1}}$ , $AQ = 1 - \dfrac{CA + PPE + \text{Securities}}{TA}$	More "soft" assets: capitalised costs, intangibles
SGI, sales growth index	$\text{Sales}_t/\text{Sales}_{t-1}$	Fast growers face pressure to sustain growth
DEPI, depreciation index	$\dfrac{DR_{t-1}}{DR_t}$ , $DR = \dfrac{\text{Dep}}{\text{Dep} + PPE}$	Slowing depreciation rate inflates profit
SGAI, SG&A index	$\dfrac{SGA_t/\text{Sales}_t}{SGA_{t-1}/\text{Sales}_{t-1}}$	Falling administrative efficiency
TATA, total accruals to total assets	$(NI_t - CFO_t)/TA_t$	Profit not backed by cash
LVGI, leverage index	$\dfrac{LEV_t}{LEV_{t-1}}$ , $LEV = (CL + LTD)/TA$	Rising leverage: debt-covenant incentives

\begin{aligned} M = -4.84 &+ 0.920\,\text{DSRI} + 0.528\,\text{GMI} + 0.404\,\text{AQI} + 0.892\,\text{SGI} \\ &+ 0.115\,\text{DEPI} - 0.172\,\text{SGAI} + 4.679\,\text{TATA} - 0.327\,\text{LVGI}. \end{aligned} \tag{6.19}

A firm is flagged as a likely manipulator if $M > -1.78$ . Some practitioners use −2.22 as a more conservative threshold; that figure is a later convention, not from the 1999 paper. The model was estimated on US non-financial firms; it is not valid for banks or NBFCs. Note that the TATA coefficient is the largest by far: accruals carry most of the signal.

AQI has two published definitions. Beneish (1999) uses $AQ = 1 - (CA + \text{net PP\&E})/TA$ ; Beneish, Lee and Nichols ("Earnings Manipulation and Expected Returns", Financial Analysts Journal 69(2), 2013) also count securities as hard assets, $AQ = 1 - (CA + \text{net PP\&E} + \text{Securities})/TA$ , which is the form in the table above and in the course code (Exercise m06_forensics).

Worked examplelab fixture

Year $t-1$ : sales 1,000, COGS 600, receivables 100, current assets 400, PPE 300, securities 50, total assets 1,000, depreciation 30, SG&A 100, PAT 80, CFO 90, current liabilities 200, long-term debt 200. Year $t$ : sales 1,300, COGS 820, receivables 220, current assets 600, PPE 320, securities 50, total assets 1,300, depreciation 28, SG&A 115, PAT 130, CFO 20, current liabilities 260, LTD 300.

DSRI $= (220/1300)/(100/1000) = 1.6923$ . Receivables doubled-plus on 30% sales growth.
GMI $= 0.40/0.3692 = 1.0833$ .
AQI $= [1 - 970/1300]/[1 - 750/1000] = 0.2538/0.25 = 1.0154$ .
SGI $= 1.30$ . DEPI $= (30/330)/(28/348) = 1.1299$ . SGAI $= (115/1300)/(100/1000) = 0.8846$ .
TATA $= (130 - 20)/1300 = 0.0846$ . PAT rose 63% while CFO fell to 20.
LVGI $= (560/1300)/(400/1000) = 1.0769$ .

$M = -1.1197 > -1.78$ : flagged. If the two years were identical, every index is 1 and TATA $= (80-90)/1000 = -0.01$ , giving $M = -2.5268$ .

6.8.2Piotroski F-score#

Piotroski ("Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers", Journal of Accounting Research 38 (Supplement), 2000) built nine binary signals to separate strong from weak firms among high book-to-market (cheap) stocks. Score 1 if true:

Group	Signal	Definition
Profitability	F_ROA	ROA $= NI_t/TA_{t-1} > 0$
	F_CFO	$CFO_t > 0$
	F_ΔROA	$ROA_t > ROA_{t-1}$
	F_ACCRUAL	$CFO_t/TA_{t-1} > ROA_t$
Leverage, liquidity, source of funds	F_ΔLEVER	$LTD_t/TA_t < LTD_{t-1}/TA_{t-1}$
	F_ΔLIQUID	current ratio rose
	EQ_OFFER	no new shares issued
Operating efficiency	F_ΔMARGIN	gross margin rose
	F_ΔTURN	asset turnover $\text{Sales}_t/TA_{t-1}$ rose

$F = \sum$ of signals, $0 \le F \le 9$ . Piotroski's high-F (8–9) cheap stocks outperformed low-F (0–1) cheap stocks in his US sample; read the paper for the magnitudes and the 1976–1996 sample design. The lab fixture (PAT 50→80, CFO 60→110, LTD 300→280, current ratio 1.60→1.85, no dilution, gross margin 33.3%→36.4%, turnover 0.947→1.100) scores 9; reversing the years scores 4.

6.8.3An Indian red-flag checklist#

Red flag	Check	Where
Cash conversion	5-year ΣCFO / ΣPAT below about 0.7	Cash-flow statement
Receivables outrunning sales	Debtor days $=$ receivables/sales × 365 rising year on year	Ratios
Cash that earns nothing	Implied yield $=$ interest income / average cash and investments, compared with deposit rates	P&L notes (Satyam)
High cash and high debt together	Why borrow at 10% while holding cash at 7%?	Balance sheet
Promoter pledging	% of promoter shares pledged, and the trend	Shareholding pattern
Related-party transactions	RPT / revenue; loans and inter-corporate deposits to promoter entities	RPT note, LODR Reg. 23
Auditor exit or qualification	Mid-term resignation, qualified opinion, Emphasis of Matter, CARO remarks	Exchange filings, AR
Capitalised costs	(CWIP + intangibles) / total assets rising	Balance sheet
Dilution and complexity	Share count growth; number of subsidiaries	AR
Contingent liabilities	Contingent liabilities / net worth	Notes
Lenders: NPA divergence	RBI-assessed gross NPA versus reported	Bank AR "divergence" note (Yes Bank)
Beneish $M > -1.78$	6.8.1	Computed

Practitioner notes

No single flag proves fraud; frauds rarely show only one. Schilit's rule is to read for patterns: profit growing faster than cash, assets growing faster than revenue, and a management that answers analysts' questions about either with hostility. Auditor resignation in India is the strongest single signal in this list, as the case studies show.