Following Asness et al. (2013), we consider quality as the consistent ability to generate strong future cash flows. We assess quality from several perspectives: profitability, operating efficiency, earnings quality (accruals) and leverage. Current profitability is related to future levels of profitability and the persistency of profitability is a key indicator of quality. Profitability improvements that are the result of increased operating efficiency or asset utilisation are likely to be more sustainable and therefore symptomatic of quality. The level of accruals (Sloan (1996)) may also identify the recurring component of earnings and therefore act as an additional measure of quality. Conversely, a high level of debt is typically perceived as an indicator of low quality.
The building blocks of each quality measure are accounting ratios sourced from consolidated annual company reports. We examine the performance of individual measures in each Developed market region from two perspectives; improvements in company fundamentals (operating performance) and stock performance.
- Return on Assets (ROA) and change in Asset Turnover (ATO): We find that historically companies with high current levels of ROA and larger changes in ATO have displayed superior subsequent operating performance. Historically, quality companies identified using these measures have outperformed those with lower levels of ROA and smaller changes in ATO. Highly profitable companies that display improvements in operating efficiency also exhibit lower levels of volatility and systematic risk.
- Accruals: Historically, companies with higher levels of accruals have been associated with lower levels of future profitability and display lower risk adjusted performance outcomes.
- Operating Cash flow to Total Debt (OPCFD): ROA, change in ATO and accruals assess earnings quality from a profitability perspective. Leverage provides another perspective on quality. We find that historically, OPCFD is positively associated with future profitability, i.e. increased levels of leverage are associated with lower levels of future profitability.
We examine the degree of independence between the various quality measures and assess the incremental improvement in performance from combining alternative measures of quality. We conclude that it is appropriate to form a composite measure of quality based on profitability and leverage measures.
All quality metrics are based on annual company fundamentals, implying an annual index rebalance. By the end of September, fundamental data for the majority of large countries is typically available in Worldscope. Consequently, we rebalance the quality indexes annually in September and employ a six-month lag on all fundamental data to mitigate foresight in our back-tests.
The Quality factor indexes are designed to exhibit a greater exposure to quality factors. However, liquidity, turnover and diversification considerations also influence the approach to index construction. We examine a set of broad and narrow quality indexes, where the latter are designed to exhibit higher levels of exposure to our preferred measure of quality whilst remaining well diversified.
High quality companies prove relatively resilient during periods of economic hardship. We illustrate the historical performance of quality indexes across the economic cycle. A broad quality index historically outperforms the capitalisation weighted equivalent index during recessionary periods. Furthermore, a narrow quality index, displaying increased levels of exposure to quality, outperforms a broad quality index during these same periods.
The structure of this document is as follows: Section 1 provides a review of the literature on the use of quality as a factor; Section 2 discusses alternative definitions of quality; Section 3 defines a set of quality factors; Section 4 examines the performance of the individual accounting measures associated with each quality measure; Section 5 considers practical issues regarding index construction.
1. Is quality a risk factor?
The literature focuses on whether investors pay a premium for firms exhibiting quality characteristics. Asness et al (2013) propose a general definition of quality arising from a re-formulation of the Gordon growth model, where P, D, r and g are the current stock price, dividend, discount rate and growth rate in dividends respectively. If high quality securities possess common characteristics, equation (1) suggests that these attributes may include profitability, growth in earnings, the required return (i.e. safer stocks) and the proportion of earnings returned to the shareholders as dividends (i.e. payout ratio).
Asness et al (2013) find empirical evidence that investors are willing to pay a premium for stocks that display quality characteristics. Furthermore, they observe that the premium attached to quality varies through time: It tends to be lower during less turbulent periods and higher during periods of crisis. They highlight that equation (1) gives no consideration to the premium attached to quality and propose that quality should be viewed in the context of value in order to identify quality at a reasonable price.
Bender and Nielsen (2013) examine a narrower definition of quality: Earnings quality or accruals1. They find strong empirical evidence of a quality effect that persists after controlling for common risk factors such as size, value, momentum and volatility. The question of whether quality is a risk factor remains. Bender and Nielsen (2013) examine accruals as a risk model factor and find that it is not statistically significant. They conclude that an accruals measure of earnings quality is not a good risk factor.
Earnings variability and leverage however, which fall into broader definitions of quality, are considered risk factors. Hunstad (2013) demonstrates that high quality stocks earn a risk premium. He suggests risk-averse investors hold high quality stocks in order to achieve greater certainty in investment outcomes, i.e. high quality stocks should exhibit lower price volatility and risk-seeking investors bid up the price of low quality stocks, resulting in a quality premium.
2. Definitions of quality
High quality firms are frequently described as those with sustainable earnings backed by robust cash flows. There are several candidates that may proxy for Profitability. Return on Equity (ROE), measured as the ratio of earnings to shareholders’ equity focuses on the returns to equity holders. From an accounting perspective, ROE includes interest income and cash holdings.2 Performance targets linked to ROE may incentivise companies to generate improvements in ROE, either by boosting assets through mergers and acquisitions (M&A) or gearing up their balance sheet, particularly during periods of cheap credit (e.g. banks in the late 90s, (Haldane 2011)). Return on Capital (ROIC)3 is a broader measure of profitability reflecting a firm’s corporate structure and incorporates the return to equity holders and the cost of debt in the form of interest expense.
Profit Margin (PM), measured as the ratio of net income (or operating net income) to sales, captures pricing power resulting from product innovation, product positioning, and brand name recognition – see Soliman (2008). PM exhibits significant variability across industrial sectors; for example, typically the PM of a food retailer is lower than that of a tobacco business. A high PM tends to attract new entrants into a particular industry, with the result that competition tends to erode high margins through time.
De Chow et al (2010) examine various proxies for earnings quality. Earnings variability is used as a proxy for earnings persistency as indicator of quality. The variability of earnings relative to the variability of cash flows is a measure of earnings quality. These measures are subject to possible opportunistic earnings management and vary in their sensitivity to the business cycle, making it difficult to disentangle the sources of earnings variability.
Profitability is typically negatively related to leverage. Nissim and Penman (2003) suggest that an increase in financial leverage has a negative effect on future earnings. Firms that are highly profitable generate positive free cash flow and tend to employ it in order to repay debt and acquire financial assets.
Debt measures such as Debt to Equity (or Debt to Total Assets) may partly be determined by a company’s ability to time equity issuance. Baker and Wurgler (2002) document that the market valuation of firms has a strong and persistent effect on the choice of capital structure; low leverage firms raise funds when their market valuations (price-to-book ratio) are high, whilst high leverage and presumably financially distressed firms raise funds when their market valuations are low.
Rajan and Zingales (1994) highlight that total assets is not an ideal base for determining leverage, since accounts payable (contractual trade obligations) and other liabilities (e.g. assets held against pension liabilities) should not affect assessments of leverage.
3. FTSE quality factors
A definition of quality as the ability to consistently generate strong future cash flows is consistent with the general definition provided by Asness et al (2013). In this Section, we specify potential measures of quality factors from the perspectives of profitability, growth and leverage.
3.1.1 Return on assets
There is some evidence that companies with high current levels of profitability tend to exhibit high future levels of profitability; see Novy-Marx (2012). We represent profitability by Return on Assets (ROA), defined as current fiscal year net income divided by average total assets of the current and preceding fiscal year. ROA incorporates the entire corporate financial structure and is not therefore distorted by differing levels of leverage. All else being equal a company that targets earnings through M&A and leverage would exhibit a lower level of ROA than ROE.
3.1.2 Change in asset turnover
The mechanisam used to achieve improvements in profitability is important. Profitability improvements that are the result of increased operating efficiency or asset utilisation are likely to be sustainable and therefore symptomatic of quality. Asset utilisation is measured by asset turnover (ATO). Soliman (2008) shows that changes in ATO are an indicator of future profitability and that performance is positively related to changes in ATO after controlling for the level of profitability (PM) and ATO.
An alternative perspective on quality is provided by an examination of non‑cash balance sheet items or accruals. Sloan (1996) suggests that earnings can be divided into stable and transitory components – and inventories. Net non-current operating assets are the difference between the change in non-current operating assets and non-current operating liabilities. The major components of non-current operating assets are Plant, Property and Equipment (PP&E) and intangibles. The change in net financial assets is measured as the change in short-term investments, plus the change in longterm investments less changes in debt and preferred stock.
Mohanram (2005) finds empirical evidence that high growth firms outperform low growth firms. He shows that growth firms are more likely to exceed earnings forecasts and earn significant abnormal returns around earnings announcements. Asness et al (2013) demonstrate that strong risk-adjusted performance is associated with growth. In a similar vein, we examine ROA growth over the past five years and explicitly consider the price of such growth by relating it to current valuation levels (Price to Book – P/B), to form a growth at a reasonable price measure (ROA-GARP).
We employ the ratio of net operating cash flow to total debt (OPCFD) to measure leverage. Specifically, to account for industry differences in financial structure, the ratio of operating cash flow to debt relative to the regional ICB industry median level is employed.
Typically, a company will first draw on internal sources of funds (cash and marketable securities) to meet obligations or fund investment programs. If additional funding is required, external financing is used. This suggests relating the level of debt to operating cash flow is important.
OPCFD relates operating cash flow to interest charges and debt-repayment. Low levels of cash flow to debt have been shown to be related to the likelihood of business failure, see Fiedler (1971). Consequently, required rates of return or discount rates should be higher for riskier, more leveraged companies with low levels of operating cash-flow to debt.
3.4 Quality factors for financials
We define financials as Banks, Insurance and Financial Services companies (ICB Industry 8000, Financials). Owing to the specific nature of their business; from both an operating and financing perspective, we distinguish between financial and non-financial companies. For example, working capital, CAPEX and debt are not clearly defined under IFRS or US GAAP for Financials. Consequently, a number of the quality measures discussed, such as operating cash flow and accruals cannot meaningfully be calculated or are not applicable to financial companies.
For financial companies, ROA is the sole measure of profitability. Additionally, the (ROA-GARP) measure in section 3.2 is used to capture growth. No measure of leverage is employed for Financials.
4.1 Factor assessment
We assess the factors discussed in Section 3 by examining both future operating and stock performance. Specifically, we define future earnings as ROA in the next fiscal year. Fundamental differences across regions prompt us to consider each region separately. Figure 1 indicates that the median company ROA varies systematically over time and across regions, with North America exhibiting the highest and Japan the lowest levels of ROA.
The remainder of this section examines the simulated performance of each quality measure. We examine the operating performance of constituents of the FTSE Developed Index over the period September 2000 to September 2013. All accounting data is lagged by six months in these back-tests.
4.2 Profitability (ROA, change in ATO and accruals)
Table 1 displays the equally-weighted average future profitability (ROA in the next fiscal year) for North America (for other regions see Appendix A3) between fiscal years 2000 and 2012. Specifically, each fiscal year we sort firms into three groups based on the current level of ROA. Within each ROA group, we sort firms on the current year change in ATO. Stocks with missing data, no future ROA or that are loss making in the current fiscal year are removed.
Stocks that exhibit both high current ROA and high changes in ATO form the group in the bottom right corner of each table (High-High group), whilst those with low current ROA and small changes in ATO form the group in the top left corner (Low‑Low group). The fourth column and first row report the average future ROA from independent sorts on changes in ATO and current ROA respectively.
A high current level of ROA signals high levels of future profitability. For example, in North America the high current ROA group on average exhibits future ROA of 19.8%, compared to 7.1% for the low current ROA group. Furthermore, larger changes in ATO incrementally signal higher future average levels of profitability. Stocks with both high current ROA and large changes in ATO had on average future profitability of 21.4% compared to 19.8% for all high current ROA stocks and 6.8% for stocks with both low current ROA and small changes in ATO. Within each ROA group, stocks with larger changes in ATO exhibit higher than average future profitability.
These results are broadly confirmed in all regions, suggesting that sustainable profitability may be identified by current ROA together with the change in ATO. Current ROA is related to future ROA; yet high ROA prompted by increases in operational efficiency is likely to be more durable.
Table 2 contains the simulated historical performance metrics of equally weighted quintiles formed on profitability and changes in ATO. Each September, we sort constituents of the FTSE Developed (ex Financials) for each region into five groups based on these measures. Securities with the highest average ROA and change in ATO ranks form the high quintile group, whilst those with the lowest composite rank form the low quintile group. Historically, the high scoring group has performed better than the low scoring group, in terms of performance and lower volatility and draw-downs. Furthermore, the high scoring groups consistently exhibit lower levels of systematic risk (beta) than the low scoring group.
— Download the complete paper —