Equity Investing in the Age of Intangibles

Abstract

Expenditures on the creation of intangible capital have increased, but accounting standards have not kept pace. We investigated whether this has affected the value relevance of book value and earnings. We constructed a composite measure of intangible intensity by which to classify industries. The measure is based on intangible assets capitalized on the balance sheet; research and development expenditures; and sales, general, and administrative expenditures. We show that the value relevance of book value and earnings has declined for high-intangible-intensity companies in the United States and abroad, but for the low-intangible-intensity group, it has remained stable in the United States while increasing internationally.

Editor’s Note:

Submitted 19 October 2020

Accepted 30 December 2020 by Stephen J. Brown

Disclosure: The authors report no conflicts of interest.

This article was externally reviewed using our double-blind peer-review process. When the article was accepted for publication, the authors thanked the reviewers in their acknowledgments. John Adams and one anonymous reviewer were the reviewers for this article.

The authors gratefully acknowledge the suggestions of Executive Editor Stephen J. Brown and Co-Editor Steven Thorley, CFA. They also thank their colleagues at Bridgeway Capital Management, especially Andrew Berkin (head of research) for his guidance and encouragement and other members of the investment management team for helpful comments.

Notes

¹ The mismatch may be more acute for business entities that are in the early stages of their life cycle, when spending on activities that create intangible assets (e.g., research and development or customer acquisition) is high.

² In addition to net income and book value, they studied cash flow from operations, cash, total assets, intangible assets, sales, sales growth, R&D expenses, advertising expenses, cost of goods sold, capital expenditures, other comprehensive income, and special items.

³ Intangible capital items may include computerized information, innovation (including both scientific R&D and nonscientific discovery and development; Corrado, Hulten, and Sichel 2005), human resources (Pantzalis and Park 2009), organizational competencies (Lev and Radhakrishnan 2005), customer franchises (Bonacchi, Kolev, and Lev 2015), and brand values (Barth, Clement, Foster, and Kasznik 1998).

⁴ Li (2020) argued and showed that, at least in developed international markets, book value can be successfully adjusted without relying on the complex procedures suggested by Peters and Taylor (2017).

⁵ Of the 236,008 (411,330) company-year observations in our US (international) sample for which ubiquitous financial statement items, such as total assets, were available in the Xpressfeed database, 63,602 (46,160) company-year observations were for companies in the Banks, Insurance, and Diversified Financials industry groups.

⁶ In a pre-IFRS reporting regime, certain countries (including Australia, the United Kingdom, and France) permitted both expensing and capitalization of R&D expenditures. Oswald, Simpson, and Zarowin (2017) found differences in (1) the value relevance of the capitalized versus expensed development costs in a pre-IFRS regime and (2) changes in the value relevance of R&D expenditures before and after IFRS adoption for UK companies that switched from expensing to capitalization. Jaafar (2011) showed that the adoption of Australian-equivalent IFRS led to an increase in the value relevance of identifiable intangible assets.

⁷ For example, Lee and Lee (2020) stated that prior to 1999 in South Korea, R&D expenditures were classified as either ordinary or extraordinary depending on the characteristics of the activities; R&D expenditures that occurred in the ordinary course of business were expensed, whereas those not meeting this criterion were capitalized.

⁸ Most prior studies on this subject (e.g., Bublitz and Ettredge 1989) have shown that the life of brand value assets created by advertising expenditures is no more than one to two years.

⁹ In unreported results, we found that for our sample of US companies, advertising expenditures dropped from about 3.6% to 1.6% of total revenues while R&D expenditures rose from about 9.5% to more than 14.0% in the 1994–2018 period.

¹⁰ In descending order of GDP, the top 15 countries in the world are the United States, China, Japan, Germany, the United Kingdom, France, India, Italy, Brazil, Canada, Russia, South Korea, Australia, Spain, and Mexico. The complete list by GDP ranking is available from the World Bank (2019).

¹¹ Xpressfeed reports R&D expenses and SG&A expenses as separate components of the income statement item under “Other Operating Expenses.” Our three metrics had to have nonnegative values, and their intensity could not be computed if the scaling variable (total assets or total revenues) was missing or zero. Such cases (amounting to fewer than 0.4% of all available company-year observations) were treated as data errors and excluded from our sample.

¹² Relative rankings for intangible intensity based on alternative measures (total assets or total expenses) to scale R&D expenses and SG&A expenses were similar and are not reported here for brevity.

¹³ Table A1 in Appendix A provides some summary information about the availability of such data separately for developed and emerging countries.

¹⁴ In light of our focus on interindustry differences in intangible intensity, we provide, in Figure A3 and Figure A4 in Appendix A, information about the availability of the requisite data at the industry level for, respectively, the United States and other countries.

¹⁵ We recognize that this approach is imperfect. Within the four-digit GICS classifications that we used, intangible intensity can vary at the subindustry level; for example, within the Utilities sector, the wind and solar power sectors are likely to differ from those that rely on fossil fuels and nuclear energy. A more granular industry classification approach would yield additional insights but would come at the expense of reduced sample sizes at the industry level.

¹⁶Figures A3 and A4 confirm this conjecture.

¹⁷ We acknowledge that our assumption that all sources of intangible capital are equally important contributors to intangible intensity is subjective. The three metrics of intangible capital have different useful lives, and they differ in the amount and timing of the cash flows they generate. Accurate measurement of these attributes would enable assignment of more appropriate (unequal) weights to different sources of intangible capital.

¹⁸ We required at least three companies in an industry for estimation of the median intangible intensity, and we required that intangible intensity medians for at least two of the three metrics be available for computation of the composite intangible intensity in any year.

¹⁹ According to Core et al. (2003), this period was marked by several unusual economic developments, including high stock market returns, high valuations, and increased productivity driven by the declining price of computing power and investments in information technology and modern manufacturing facilities that benefit from information technology.

²⁰ The Theil–Sen estimator (Theil 1950; Sen 1968) is a nonparametric technique for estimating a linear trend by choosing the median of the slopes of all lines through pairs of points in the sample. This procedure produces a (statistically) efficient estimator that is insensitive to outliers. It can be significantly more accurate than a nonrobust simple linear (least-squares) regression for skewed and heteroskedastic data.

²¹ For the US universe, the 95% confidence intervals for, respectively, the high- and low-intangible-intensity groups are (–0.011, 0.000) and (–0.002, 0.007). For the international universe, the 95% confidence intervals are (–0.005, 0.004) and (0.005, 0.012).

²² See previous note.

²³ We repeated the analyses in Table 4 and Table 5 for a full “global” sample of companies. For this test, we combined US and international companies in the high-intangible-intensity groups, added an indicator variable to distinguish whether a particular company belonged to the US or international universe, and ran our primary annual cross-sectional regression of stock price on book value and earnings for this “global” sample of high-intangible-intensity companies. We obtained 25 R² values. We repeated the same procedure for the low-intangible-intensity companies. The Theil–Sen’s slopes (z-statistics) for the high- and low-intangible-intensity groups were, respectively, –0.00 (0.26) and 0.007 (3.10), and the coefficient (t-statistic) for $IN{T}_{-}{D}_h\times TIM{E}_t$ was –0.007 (–3.49).

²⁴ The coefficient on the TIME variable is positive and significant for both the US and international regressions, indicating that the combined value relevance of earnings and book value has increased for companies in the low-intangible-intensity group for the time period and sample of companies included in our study.