The true earnings of a business are represented by the concept of return on investment – the amount of cash out we receive in return for a given amount of cash in. As a result of history¹ we measure these numbers using double entry accrual accounting, however, a number of factors mean that any measurement system will be imperfect.
In its broadest sense, the term earnings quality describes how closely measured (reported) earnings match true earnings:
Reported Earnings (RE) = Accounting Function (f) of True Earnings (X); RE = f(X)
There are two components to this idea – how well do we understand True Earnings (X) and how good a job does our measuring system ‘f’ do of representing it. This leads to two common usages of the term “Earnings Quality” by practitioners:²
- The Composition of earnings – How well do we understand True Earnings? Are they annuity style or one off? High risk or low risk? Cyclical or defensive?
- The Representation of earnings – How good a job does f do of representing X?
Within the Equity Toolkit we will reserve the term Earnings Quality to refer to this second usage. How accurately do reported earnings represent the underlying economics of the business?
Obviously there is a chicken and egg situation here. We cannot know the quality of this Representation unless we first have an understanding of true earnings, including their composition. This understanding is gained through analysis of the operational and returns characteristics of the business. So whilst the Composition of Earnings is an important consideration of overall earnings quality, for the purposes of the Equity Toolkit we have considered these issues separately under the following topics: Operational Frame (composition of cash flows), Market frame (valuation) and Financial Frame (Measurement of Returns).
Two further semantic points are worth clarifying. Firstly, there is a tendency for the expression “bad earnings quality” to be reserved for situations of accounting fraud or breaches of accounting standards, with the corollary that compliance with such standards is acceptable. However, this fails to recognise that earnings quality falls on a continuum. Fraud might represent an extreme of this continuum, but many non fraudulent and compliant activities might still fall on the “bad” end of this spectrum. The overriding objective of analysis should be to consider where on the spectrum of representation reported earnings lie, not to simply understand compliance to standards.
The second point is the distinction between good and bad earnings quality. In the academic literature, these terms are used to define how good a job f is doing in representing X. However, amongst practitioners, these terms are generally used to refer to the direction (or bias) of the representation. So a reference to good earnings quality would reference a situation where reported earnings understate underlying business economics, whilst a reference to bad earnings quality would imply the reverse.³
The crux of earnings quality analysis is as follows:
- Reported earnings represent an estimate of underlying performance.
- This estimate is subject to error and bias.
- These errors and biases are not properly understood and so represent opportunities for investors.
The source of these opportunities is that for a combination of behavioural4 and institutional reasons, investors tend to fixate on earnings rather than the statistical significance of these earnings. Practical experience tells us that this should offer investment opportunities and whilst the academic literature on earnings quality is vast and often conflicting, it demonstrates three broad ideas:
1. Companies deliberately manipulate earnings;
Anyone who has worked in finance for any length of time knows anecdotally that companies manipulate earnings. The incentives for this are varied, but the evidence from academic studies is reasonably robust. Distributions of reported earnings (whether absolute or relative to estimates) have a kink around zero – suggesting concerted efforts at the margin to boost reported profitability.