Ethics and Its Ties to Earnings Management

...

HYPOTHESIS AND RESEARCH METHODS

As a result of the Sarbanes-Oxley Act of 2002, senior management had to certify the faithful representation of the reported financial statements. In response to fraud activity coming to light, many CEOs and CFOs attempted to use the excuse that they had no idea earnings management and fraud had taken place because they were not accountants. With this act, top management is required to sign off on the financial statements indicating that they have reviewed the statements for any errors or intentional misstatements. Before the Enron scandal was uncovered, it looked as if Enron put on a dog and pony show in regards to social responsibility; however, the “tone at the top” was merely a depiction of deceit and simply its senior managers put the bottom line ahead of ethical behavior. Based on prior research, we hypothesize that:

H1: There is a direct link between ethics and earnings management – firms that have stronger ethical standards and stronger corporate social responsibility will tend to manipulate their financial statements less than firms that have weak ethical standards, and weak corporate social responsibility.

Because earnings management has had the largest impact on publicly traded companies, we will only include issuer companies in our sample. For the measures of companies that behaved ethically we will take the measures from previous research (Barton, Kirk, Reppenhagen, and Thayer 2010). The criteria outlined for our sample selection is as followed: (1) must be firms included in the 2008-2013 KLD STATS database (like Barton et al.’s 2010 research); however, we changed the time period to ensure the Sarbanes-Oxley Act of 2002 has sunk in with each company (2) firms must be either high, moderate, or low in corporate social responsibility activities involvement (3) firms must have required data on COMPUSTAT. From this selection, we will separate small companies from large, companies with executive bonus compensations from those without, and we will also consider whether those bonuses are determined using net income versus operating cash flow versus return on investment.

KLD STATS is “based on proprietary researched profiles of corporate environmental, social and governance (ESG) factors” (RiskMetricsGroup, p.3). We use KLD STATS in order to evaluate whether a firm is rated high, moderate, or low in regards to corporate social responsibility. If a firm was rated high then we considered this firm to be highly ethical; if a firm was rated moderate then we considered this firm to be moderately ethical; and lastly, if a firm was rated low then we considered this firm to have low ethical standards.

To measure earnings management, we will use the models offered by Cohen et al. (2008) that was referenced in Barton et al.’s paper. We will relate earnings management to abnormal levels of accruals, operating cash flows, production costs, and discretionary expense (Barton et al. 2010). The following models used estimate normal levels for accruals, OCF, production costs, and discretionary expense while residuals will represent abnormal levels.

***ACCRUALSit/ASSETSit–1 = α1kt(1/ASSETSit–1) + α2kt([∆SALESit – ∆ARit]/ASSETSit–1) + α3kt(PPEit/ASSETSit–1) + α4kt(OCFit/ASSETSit–1) + ϖit, (7) OCFit/ASSETSit–1 = δ1kt(1/ASSETSit–1) + δ2kt(SALESit/ASSETSit–1) + δ3kt(∆SALESit/ASSETSit–1) + ωit,

***OCFit/ASSETSit–1 = δ1kt(1/ASSETSit–1) + δ2kt(SALESit/ASSETSit–1) + δ3kt(∆SALESit/ASSETSit–1) + ωit,

*** PRODUCTION_COSTSit/ASSETSit–1 = θ1kt(1/ASSETSit–1) + θ2kt(SALESit/ASSETSit–1) + θ3kt(∆SALESit/ASSETSit–1) + θ4kt(∆SALESit–1/ASSETSit–1) + ξit,

*** DISC_EXPENSESit/ASSETSit–1 = ϕ1kt(1/ASSETSit–1) + ϕ2kt(SALESit–1/ASSETSit–1) + ζit,

To test whether a firm that had strong ethical standards had less earnings management we would generate a logistic regression model. Our dependent variable was a dichotomous measure relating to earnings management:

EM= 0 if the company did not performed perceived earnings management, and 1 if the company performed perceived earnings management.

Next we exhibit our control variables. These independent control variables were based

off prior research (Barton et al. 2010):

ETHICS = 0 if the company ranked low in corporate social responsibility, and 1 if the company ranked high in corporate social responsibility.

SIZE Natural logarithm of market capitalization of common equity, calculated as share

price multiplied by shares outstanding

ROA Income before extraordinary items, divided by total assets

LOSS 1 if income before extraordinary items is negative in at least three of the six years

in test period, 0 otherwise

PB Ratio of price to book value of equity

BONUS Bonuses paid to five highest compensated managers, divided by their total

compensation

STOCK Number of shares owned by five highest compensated managers, divided by total

common shares outstanding

OPTIONS Number of exercisable options owned by five highest compensated managers,

divided by total common shares outstanding

NEW_FINANCING Net proceeds from issuance of debt and equity, divided by total assets

MET_OR_EXCEED Number of times during test period that firm met or beat analysts’ consensus

forecast of quarterly earnings, divided by number of quarters with at least one

analyst forecast

LEVERAGE Total short- and long-term debt, divided by total assets

TAX 1 if marginal tax rate (Graham 1996, 2007) is larger than effective tax rate (i.e.,

income tax expense divided by pretax income), 0 otherwise

ENVIRONMENT Standardized sum of ENV_STR over test period, less standardized sum of