Evaluating Earnings Quality, Part 1

Steps in evaluating the quality of financial reports:

Step 1: Understand the company, its industry, and the accounting principles it uses and why such principles are appropriate.

Step 2: Understand management including the terms of their compensation. Also evaluate any insider trades and related party transactions.

Step 3: Identify material areas of accounting that are vulnerable to subjectivity.

Step 4: Make cross sectional and time series comparisons of financial statements and important ratios.

Step 5: Check for warning signs as discussed previously.

Step 6: For firms in multiple lines of business or for multinational firms, check for shifting of profits or revenues to a specific part of the business that the firm wants to highlight. This is particularly a concern when a specific segment shows dramatic improvement while the consolidated financials show negative or zero growth.

Step 7: Use quantitative tools to evaluate the likelihood of misreporting.

Quantitative Tools

The Beneish Model

The Beneish model is a probit regression model that estimates the probability of earnings manipulation using eight independent variables. The M score determines the probability of earnings manipulation, higher values indicate higher probabilities.

Where:

M score > −1.78 (i.e., less negative) indicates a higher than acceptable probability of earnings manipulation.

Days sales receivable index (DSRI): Ratio of days’ sales receivables in year t relative to year t − 1. A large increase in DSRI could be indicative of revenue inflation.

Gross margin index (GMI): Ratio of gross margin in year t − 1 to that in year t. When this ratio is greater than 1, the gross margin has deteriorated. A firm with declining margins is more likely to manipulate earnings.

Asset quality index (AQI): Ratio of non current assets other than plant, property, and equipment to total assets in year t relative to year t − 1. Increases in AQI could indicate excessive capitalization of expenses.

Sales growth index (SGI): Ratio of sales in year t relative to year t − 1. While not a measure of manipulation by itself, growth companies tend to find themselves under pressure to manipulate earnings to meet ongoing expectations. Depreciation index (DEPI): Ratio of depreciation rate in year t − 1 to the corresponding rate in year t.

The depreciation rate is depreciation expense divided by depreciation plus PPE. A DEPI greater than 1 suggests that assets are being depreciated at a slower rate in order to manipulate earnings.

Sales, general and administrative expenses index (SGAI): Ratio of SGA expenses (as a % of sales) in year t relative to year t − 1. Increases in SGA expenses might predispose companies to manipulate. Accruals = (income before extraordinary items − cash flow from operations) / total assets.

Leverage index (LEVI): Ratio of total debt to total assets in year t relative to year t − 1.

This is likely why Beneish created an alternate model (not included in the curriculum) that excludes SGAI and LEVI.

Limitations of the Beneish Model

The Beneish model relies on accounting data, which may not reflect economic reality. Deeper analysis of underlying relationships may be warranted to get a clearer picture. Additionally, as managers become aware of the use of specific quantitative tools, they may begin to game the measures used.

This concern is supported by evidence indicating that the predictive power of the Beneish model is decreasing over time.

Altman Model

While not directly related to earnings quality, Altman’s Z score model was developed to assess the probability that a firm will file for bankruptcy.

Altman’s model relies on discriminant analysis to generate a Z score using five variables. The variables used include net working capital as a proportion of total assets, retained earnings as a proportion of total assets, operating profit as a proportion of total assets, market value of equity relative to book value of liabilities, and sales relative to total assets. Each variable is positively related to the Z score, and a higher Z score is better (less likelihood of bankruptcy).

Hence, higher values of any of the five variables reduce the probability of bankruptcy under Altman’s model.

One limitation of the Altman model is that it is a single period static model and, hence, does not capture the change in key variables over time. Additionally, similar to the Beneish model, Altman’s model mostly uses accounting data. Other market based data sources may provide more meaningful information for evaluation of default risk.

High quality earnings are characterized by two elements:

1. Sustainable: high quality earnings tend to persist in the future.

2. Adequate: high quality earnings cover the company’s cost of capital.

As stated previously, high quality earnings assume high quality reporting. In other words, low quality earnings come about due to (a) earnings that are below the firm’s cost of capital and/or (b) earnings that are not sustainable and/or (c) poor reporting quality (i.e., the reported information does not provide a useful indication of a firm’s performance).

Sustainable or persistent earnings are earnings that are expected to recur in the future. Earnings comprised of a high proportion of non recurring items are considered to be non sustainable (and hence low quality).

Classification of items as non recurring is highly subjective and, hence, is open for gaming. Classification shifting does not affect the total net income but rather is an attempt to mislead the user of the financial statements into believing that the “core” or “recurring” portion of earnings is higher than it actually is. One way to overstate persistent earnings is to mis classify normal operating expenses as expenses from discontinued operations.

Analysts should be wary of large special items or when the company is reporting unusually large operating income for a period. Companies may include non GAAP metrics such as pro forma income which excludes non recurring elements. Analysts should review the disclosures reconciling pro forma income to reported income to evaluate whether the items that are labeled as non recurring are truly non recurring.

One way to gauge earnings persistence is to use a regression model such as:

In this model, a higher value of β1 indicates higher persistence of earnings.

Accruals

Under the accrual basis of accounting, revenues are recognized when earned and expenses are recognized when incurred, regardless of the timing of cash flow. Unfortunately, accrual accounting requires considerable subjectivity because of the many estimates and judgments involved with assigning revenues and expenses to appropriate periods. Due to this subjectivity in revenue and expense recognition, disaggregating income into its two major components, cash and accruals, further enhances its quality as an input for forecasting future earnings.

The accrual component of income is less persistent than the cash component. In the following regression model, β1 > β2:

It is important to recognize that some accruals occur as part of normal business. Such accruals are called as non discretionary accruals. Discretionary accruals result from non normal transactions or non normal accounting choices, and are sometimes used to manipulate earnings.

One mechanism to separate discretionary and non discretionary accruals is to model (using regression) total accruals as a function of a set of factors that typically give rise to normal accruals (e.g., growth of credit sales, amount of depreciable assets). The residuals from such a model would be an indicator of discretionary accruals.

Finally, a major red flag about earnings quality is raised when a company reports positive net income while reporting negative operating cash flow.

Other Indicators

One metric used to identify potentially low quality earnings is to look for those companies that repeatedly meet or barely beat consensus estimates. While this is not a foolproof metric, analysts should be wary of a company that narrowly beats its benchmarks consistently.

External indicators of low quality earnings include enforcement actions by regulatory authorities (e.g., SEC) and restatements. External indicators are not very useful as they cannot be used to forecast deficiencies before such deficiencies are publicly known.