Amsterdam Business School
Which ratio is more informative: financial ratio or adjusted
financial ratio?
Name: Chaoyang Huang Student number: 10457453 Supervisor: Dr. Sanjay Bissessur
2nd supervisor: Dr. Georgios Georgakopoulos Date: June 22, 2014
Master Thesis
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam Final version
Abstract
We extend Aliabadi et al. (2013)’s research regarding the informativeness of financial ratios by capitalizing R&D expenditure and their predictability. We find that adjusted financial ratios correlate much better than those without adjustment with current year stock return, which extends the research done by Fraser et al. (2009) and Wild (1992). However we do not find the increased predictability of adjusted financial ratios than those without the adjustments. Furthermore this thesis contributes to the accounting treatment of R&D expenditure for both US GAAP and IFRS in terms of informativeness. In particular, this research validates the informativeness of the two most popular adjusted financial ratio calculations: CFROI® and CROCI®.
Table of contents
1 Research Question and Introduction ... 5
2 Literature review ... 9
2.1 Value relevance, information content, accounting and earnings... 9
2.2 Financial ratios, earnings quality and stock returns ... 12
2.3 Adjusted financial ratio in practice - CFROI® and CROCI® ... 17
2.4 R&D expense, Disclosure and Capitalization ... 18
3 Hypothesis and Research Method ... 22
3.1 Hypothesis ... 22
3.2 Research Method ... 23
3.2.1 Sample selection and data collection ... 23
3.2.2 OLS Regression formula ... 25
3.2.3 Capitalization of R&D expenditure ... 27
3.2.4 An example of financial ratio calculation ... 27
4 Result ... 29
4.1 Regression of Fiscal Yeart stock return on Fiscal Yeart ratios ... 29
4.1.1 Descriptive statistics and Pearson Correlation Matrix for the dependent and independent variables. ... 29
4.1.2 Result of Regression of Fiscal Year t stock return on Fiscal Year t ratios ... 34
4.1.3 Analysis ... 36
4.2 Regression of Fiscal Yeart+1 stock return on Fiscal Yeart ratios ... 39
4.2.1 Descriptive statistics and Pearson Correlation Matrix for the dependent and independent variables. ... 39
4.2.2 Result of Regression of Fiscal Year+1 stock return on Fiscal Yeart ratios ... 42
4.2.3 Analysis ... 44
4.3 Low R2 ... 45
5 Conclusion and Recommendation ... 50
6 Limitations ... 52
Literature ... 53
1 Research Question and Introduction
This thesis compares the informativeness of two kinds of financial ratios for stock returns, the standard financial ratio represented by Return on Assets ratio (ROA) and the adjusted financial ratio represented by adjusted Return on Assets ratio (adjusted ROA)1.
Financial ratios are measured in monetary term and are calculated directly or indirectly based on the financial statement of a company. There are two kinds of financial ratios: ones (financial ratios) that are calculated directly from financial statements such as ROA and the other (adjusted financial ratios) are calculated indirectly from financial statements and are based on adjusted financial information such as adjusted ROA. In order to calculate adjusted financial ratios, there are often assumptions and economic calculations for the purpose of more informativeness. The two well-known examples are CFROI® and CROCI®2. The reason for using adjusted financial ratio is its more “informativeness” to investors, management and other users. The informativeness in this research is defined as the correlation between (adjusted) financial ratios and current or future stock return.
There are some researches on informativeness of financial ratios and they show evidence of predictive value, at least in respect to financial difficulties (Horrigan, 1968). Over the past 40 years, academic researches have been focusing on correlation and predictive value of financial ratios such as book-to-market ratio (B/M), dividend yield (DY) and earning yield (EY) or traditional profitability ratio like ROA. However there is no research addressing the informativeness of adjusted financial ratios3 although they are widely used in practice.
Even though financial information, upon which the financial ratios are based, is expected to be informative, this is not always the case according to various studies and practice.
Research and Development expenditures (R&D) are fully expensed based on USGAAP. According to IFRS, research expenditure is expensed while development expenditure may be capitalized subject to several specific criteria.
According to Dukes (1976), Zion (1978), Chan (1990), Lev et al. (1996), Zhao (2002), there are concerns over the value relevance of financial statements due to lack of capitalization of
1 Ratios can be defined as non-financial ratio and financial ratio. Non-financial ratio is not measured in monetary term such as market penetration rate. This thesis only examines the financial ratios.
2 CFROI® is registered trade mark of Holt, a subsidiary of Credit Suisse First Boston. CROCI® is registered trade mark of Deutsche Bank.
3 The famous EVA®, which is widely studied, is actually an adjusted firm value rather than a financial ratio composing of a numerator and a denominator.
economic assets such as R&D expenditure. In R&D expenditure case, if it is really an expense, the best way to boost the earning or stock return is to cut all the R&D expenditures. If the input to financial ratios lacks of value relevance, informativeness of the output from financial ratios is questionable.
CFROI® and CROCI® is an example to tackle the problem of traditional financial ratios by capitalizing R&D expense in their adjusted financial ratio calculation, which is a major step to improve the informativenss.
The aim of using adjusted financial ratios is to increase the informativeness of financial ratios and provide more useful information to the investors and other users.
To test the increased inforamtiveness of adjusted financial ratios with R&D capitalization, OLS
regression developed by Aliabadi et al. (2013) is used. Total number of company year in the
regressions is more than 87,000, a very large sample base.
The results of OLS regressions suggest that adjusted ROA ratio correlates much better with current year stock return but not with subsequent year stock return. This finding implies that information content due to capitalizing R&D expenditure in adjusted ROA ratio is fully incorporated in current year stock return and not reflected in subsequent year stock return. This casts some doubts about the improvement on predictability of adjusted financial ratios.
This thesis makes following contributions:
(1) This is the first research to examine the informativeness of research and development expenditure within the return on assets framework. According to Gentry (2010) and Aliabadi et al. (2013), there is a correlation between accounting ratios and market performance measures such as stock return where ROA is the most informativeness accounting ratio among 6 commonly used financial ratios. This thesis extends the scope of the researches done by Gentry (2010) and Aliabadi et al. (2013) in three areas: (a) it improves the informativeness of a key financial ratio by adjustments. (b) it investigates the informativeness not only from the perspective of current year stock return but also the subsequent stock return. (c) it updates the analysis up to 2012 with extended sample pool that covers more than 64,000 US company year and 23,000 international company year between 1967 and 2012 (compared with 1,208 company year between 2006 and 2009 in the research of Aliabadi et al. (2013)).
(2) This research further contributes to the accounting treatment/disclosure of R&D expenditure. We find that some of the research expenditures may be capitalized or minimally
historical success rate of research projects should be disclosed to facilitate proper valuation of the firm value. This treatment/disclosure therefore increase the informativeness of financial ratios.
(3) In addition to academic contributions, this research may be applied to investment community in selection of financial ratios and adjusted ratios.
(4) This research presents more “updated rule of the world”. It covers all USA and international companies with R&D expenditure and key accounting data available between 1975 and 2012 in Compustat/CRSP. Prior researches on informativeness of R&D expenditure were limited to USA and a few countries with limit number of companies and years due to research model capabilities. Furthermore, we also compare the informativeness of US companies before 1975 with that after 1975 to validate the role of change in accounting treatment of capitalizing R&D expenditure.
(5) This research presents an unique comparison of the informativeness of R&D expenditure between USGAAP and EU’s IFRS. It provides a convincing explanation to the hypothesis 3 of this research and shows which accounting treatment is superior.
(6) This independent research can serve as a proof if the adjusted financial ratios such as CFROI® and CROCI® promoted by some agencies really improve the information content of financial ratios or they are just marketing tools. Both CFROI® and CROCI® ratios are used widely by investors and even an investment fund was created based on them. For example DB Platinum IV CROCI US fund, a three Morningstar funds, uses CROCI® concept to pick the stocks. However the informativeness of these adjusted financial ratios has never been validated by an independent research. This research will fill in this gap.
(7) The result of this research may be applied to other financial ratio calculations such as ROCE (Return on Capital Employed) or ROE (Return on Equity) to make them more informative and thus more economic oriented. The research method can be used to test the informativeness of other accounting practice or accounting treatment proposal such as capitalizing operating lease contracts in lessee’s book, which is currently under consideration by IASB and FASB.
Our research finds that the ratios adjusted by capitalizing R&D expense correlate better with current year stock return than those without the adjustment. In this sense, adjusted financial ratios are more informative. However this is not the case if subsequent year stock return is used. Chapter 2 contains a detailed literature review in the areas of earnings and earnings quality, financial ratios, value relevance, informativeness and capitalization of R&D expense and a link
between these factors and stock returns. The application of adjusted financial ratios is also explained.
Chapter 3 provides 4 hypotheses and the quantitative research method.
Chapter 4 contains results, analysis and recommendations from this research. We also make robustness tests to confirm our result. Our research tries to limit the number of outliers. The reasons are explained in this chapter.
Chapter 5 gives conclusion remarks.
2 Literature review
This chapter contains a detailed literature review in the areas of earnings and the quality of earnings, financial ratios, value relevance, informativeness, capitalization of R&D expense and their relations with stock returns. By using prior literatures, we make a link among these factors. Earnings is the numerator and important part of ratio calculation. Earnings with good quality correlates better with firm value. Capitalization of R&D expense will make earnings and assets more value relevant, which in turn leads to more informative adjusted ratios that correlate with stock return better. We will explain how adjusted financial ratios work in practice. This chapter also reveals how the hypotheses are developed.
2.1 Value relevance, information content, accounting and earnings
There is a close relationship between value relevance and informativeness (Information content) of accounting information.
Value relevance of a performance measure can be defined as the relationship between a performance measure and current or future firm value. A performance measure is considered value relevant if it correlates with current or future firm value. Therefore value relevance is in terms of absolute value of a firm.
IFRS defines relevance as the predictive and confirmatory ability of financial information. “Financial information has predictive value if it can be used as an input to processes employed by users to predict future outcomes”. “Financial information has confirmatory value if it provides feedback about (confirms or changes) previous evaluations” according to The Conceptual Framework for Financial Reporting of IFRS (2010).
According to USGAAP, an accounting number is value relevant if the amounts reflect information relevant to the users to valuating the firm value.
Information content (informativeness) in finance and accounting is defined as the increase or decrease in the price of a security as a result of a release of new relevant information. For
example, a stock’s price drops because of a negative earnings announcement. More useful a piece of accounting information to the investors, more changes in price or trading volume. Therefore it is in terms of change in firm value or stock trading volume after factor A being changed to Factor B.
We can also view informativeness in another perspective. If factor B is proved to be more correlated with stock return or stock trading volume change than factor A, factor B should have more information content than factor A does.
It is logical to assume that if more value relevant information is used in investment decision process, an investment decision will be made with higher usefulness and informativeness. Who is the CEO of General Electric (GE) should be more value relevant to the firm value than the weather situation in New York. Therefore in terms of stock return of GE, the change of CEO of GE is more informative than the change of weather in New York. In our research, if we add value relevant information such as R&D expense to the components of financial ratio
calculation, we can make financial ratio more correlate with current or future year stock return. Thus more informative.
The stock return and trading volume change are good proxy for informativeness of accounting information.
According to Scott (2012), based on all available information at any given time, investors have prior beliefs about firm’s future performance. Upon release of current period’s earnings, some investors will revise their beliefs about future firm performance. The investors who revise their beliefs upward will tend to buy the stocks according to current stock price and vice versa. Trading volume of those stocks affected will be increased. The greater the difference between current belief and prior belief, the greater difference in the extent of the change in price and trading volume.
Various researches – Ball et al. (1968), Beaver (1968), Kim et al. (1997) confirm that the role of accounting information is to help investors to estimate the stock return and make buy or sell decisions. If the information released has no information content, the investors will not make buy or sell decision. The degree of information content can be measured by the extent of price and trading volume changes after the accounting information is released.
Ball et al. (1968) measure information content of earnings by good news, if current year earnings are greater than last year’s, and bad news, if earnings are lower than last year’s. Then they test abnormal stock return of the samples shortly before or after earning announcement dates. According to their research, the average abnormal stock market return of good news firms in the month of earnings release is strongly positive while the average abnormal stock market return of bad news firms is strongly negative. Furthermore, in a wide window period, 11 months prior to and 6 months after the earning announcement date, cumulative average abnormal stock
return of good news firms outperforms the market return and cumulative average abnormal stock return of bad news firms just underperforms.
The informativeness of accounting information can be observed not only from stock price change but also from trading volume change.
Beavers (1968) finds a dramatic increase in trading volume and price changes during the week of earning announcement. Kim et al. (1997) suggest that trading volume is noisier than price change as a measure of decision usefulness of accounting information. Both researches confirm the information content (informativeness) of accounting information.
Our research uses stock return as the proxy for infomrativeness of accounting information. Differences in accounting standard, disclosure practice and corporate governance can cause the difference in informativeness of accounting earnings, according to Alford et al. (1993).
To test the assumption, Alford et al. (1993) examine the information content and timeliness of accounting earnings in 17 countries against that in USA as they have different accountings standards, disclosure and corporate governance requirements. They use two research methods. One is the investment strategy similar to that in Ball et. al (1968). The other one is a regression model of 15 month stock returns on the contemporaneous level and change in earnings scaled by market value at the beginning of fiscal year.
They find that accounting earnings in Australia, France, the Netherlands, and the United Kingdom are more informative than accounting earnings in U.S. The results for Belgium, Canada, Hong Kong, Ireland, Japan, Norway, South Africa, and Switzerland are not conclusive. Accounting earnings from Denmark, Germany, Italy, Singapore, and Sweden are less informative than those in US.
Our research will segregate US companies with international companies and EU companies to examine more clearly the informativeness of ROA in several accounting regimes.
The management itself is also a cause of the difference in informativeness of accounting earnings.
The higher managerial ownership in a firm, the higher correlation between earnings and stock return (Warfield et al. 1995). Management sometimes uses their subjective judgment to signal private information (Bissessur, 2008). One area of management’s use of accounting judgment is research and development expense. However in USGAAP, this subjective tool no longer exists after 1975 because R&D expenditure needs to be written off to profit and loss accounts
immediately. This impairs the infomrativeness of ROA ratio since investors have to make judgments and adjustments themselves, which means a detailed disclosure of R&D is needed. In addition to net income or earnings, there are some researches on the information content related to other elements of financial statements.
According to Wild (1992), not only components of earnings (revenues, Operating expenses, depreciation, interests, taxes and other incomes) but also components of book value (working capital, capital expenditures, long term assets, long term debt and preferred stock values) are informative. Disaggregate accounting data are useful for shareholders’ valuation of firms.
Our research extends Wild’s research and will use ROA and adjusted ROA which contains both earnings component and book value component to test the informativeness of accounting information.
Other elements that received attention include intangibles where R&D is a part of. Fraser et al. (2009) investigate the information content of intangible- intensive companies. They conclude that the market impact of accounting disclosures such as Annual General Meeting(AGM), a part of the annual reporting cycle, is relatively greater for companies in industries characterized by relatively low market to book ratios (such as construction industry) than in those industries with higher market to book ratios (such as software industry with big research and development budget or oil and gas industry).
Fraser et al. (2009)’s research, therefore, supports the proposition that current corporate reporting practice may not work well with the characteristics of ‘new economy’ companies because value relevant information – research and development is missing in the financial statements. Their findings add urgency to calls for research to enhance the relevance of corporate reporting.
To echo the findings of Fraser et al. (2009)’s research, our research adds back R&D expense to net income and capitalize it to balance sheet to increase the information content of financial ratios.
2.2 Financial ratios, earnings quality and stock returns
Financial ratios are widely used by investors, management and other users because they are easy to understand, comparable across industries and informative. However there are also some disadvantages. For example financial ratios, similar to accounting profit, do not have all the information content.
According to Merchant et al. (2007), most companies reward their managers heavily on standard accounting based financial measures. There are two kinds of accounting base performance measures. One is accounting profit measures such as net income. The other one is ratio measures such as Return on Investment (ROI) and Return on Net Assets (RONA).
There are several advantages to use accounting base financial measures. They can be measured timely, objectively and precisely. They are largely controllable by the management,
understandable and inexpensive.
There are also disadvantages. Some researches have shown that the correlations between annual accounting profits and stock price changes are small since accounting profits do not fully reflect economic profits. Profit calculations ignore some economic values and value changes if
accountants feel that those cannot be measured accurately and objectively. For example research in progress is expensed immediately.
Return on investment (ROI) is one of the most important financial ratio used by companies. ROI is the ratio of accounting profit divided by investments. About 80% to 93% of the companies are using ROI to evaluate the performance according to the surveys.
There are some advantages to use ROI. It reflects the trade-off that managers must make between revenue, costs and investments. It can be used to compare the return of dissimilar businesses. It is expressed by percentage and can be compared with other financial returns such as the ones used for stocks and bonds. It can be understood easily by the managers and how the measures can be influenced. However the numerator of ROI is accounting profit and it has all the limitations of profit measures.
It is therefore important of selecting the most value relevant accounting performance measure in the ratio’s numerator.
Francis et al. (2003) analyse the variability in annual stock returns related to GAAP performance measures such as Earnings, earning before interests, tax and depreciation (EBITDA) and cash flow from operating activities (CFO) and non GAAP performance measures such as Revenue per passenger mile. They find that earning dominates CFO and EBITDA in explaining security returns in industries where earning is the preferred measure. They do not find CFO or EBITDA dominates earnings in explaining stock returns in industries where CFO or EBITDA is
preferred. They conclude that EBITDA and CFO are not superior performance measures to earnings.
Barton et al. (2010) examine the value relevance of a set of eight commonly used performance measures disclosed in the financial statements of almost 20,000 companies across 46 countries between 1996 and 2005 (the year the European Union directive which required listed companies
to prepare consolidated financial statements in accordance with International Financial Reporting Standards (IFRS)4).
The eight performance measures are: sales, earnings before interest, taxes, depreciation, amortization (EBITDA), operating income, income before taxes, income before extraordinary items and discontinued operations, Net Income (Loss) - Consolidated, total comprehensive income, and operating cash flows.
Regressing current year stock returns on each of the above eight performance measures calculated by current year financial statement and using adjusted R2 as a proxy for value relevance, Barton et al. (2010) find that the value relevance of the performance measures varies substantially across line items on the income statement as well as across countries. Each of the performance measures except sales is value relevant in at least one of the 46 countries. Income before taxes is found to be the most value relevant measure in 25 countries.
Although earnings is a better value relevant performance measure, earnings quality may not be the same for individual companies.
The earnings quality can be evaluated from accruals quality perspective. With earnings = cash flows + accruals, Sloan (1996) examines if stock prices reflect information about future earnings contained in the accrual and cash flow components of current earnings. He finds that earnings with higher magnitude of cash flow component exhibits higher earning persistency than the one with higher magnitude of accrual component. Investors fail to pick up this information until that information impacts future earnings.
Sloan (1996)’s study implies that higher magnitude of cash flow component in the current year earnings means higher earnings quality. However according to Dechow (1994), accrual improves earnings' ability to measure performance relative to cash flows. Therefore earnings is a better performance measure than cash flow. The study of Dechow et al. (2002) further suggests that precise estimation of accrual improves the beneficial role of accrual but imprecise estimation reduces it. Therefore only accrual with good quality improves the earrings quality.
Lev et al. (1993) evaluate the earning’s quality by fundamental analysis. They use 12 fundamentals such as change in receivables and R&D change against industry benchmark. They find that most of the 12 fundamentals add approximately 70%, on average, to the explanatory power of earnings with respect to abnormal stock returns.
4All listed EU companies are required to prepare their consolidated financial statements in accordance with IFRS
Among 12 fundamentals, R&D expense relative change is exceptional. Relative change in R&D expense does not always have positive or negative impact on cumulative abnormal stock return. This may be due to the misspecification in the research or investors do not view industry relative cut in R&D will affect the future firm’s performance negatively.
The conclusion suggests that besides earnings, other fundamentals of financial statements also have information content. In our research, we add R&D expense back to net income and capitalize R&D expense to the assets to test if adding R&D expense, a fundamental in the research of Lev et al. (1993), to the ratio, the informativeness of ratio will be increased.
Based on above researches, we use net income, an earnings performance measure in the ratio’s numerator and capitalize R&D expense in our research to improve its quality and value relevance. However we will also use Cash Flow from Operating activities (CFO) in our robustness test to see if adjusted financial ratio is more informative.
Among many financial ratios, ROA stands out as a good informative individual ratio.
Gentry et al. (2010) confirm that there are some correlations between accounting profitability ratio such as ROA and market performance such as market to book value ratio. However they also find that there is a need to pay attention to the difference between accounting profitability and market performance measures.
Aliabadi et al. (2013) extend studies by Francis et al. (2003) and Barton et al. (2010). They use four-stage model to analyze the link between financial ratios and stock return. Six accounting performance measures are examined to identify the most informative measure5 in certain industries globally between Year 2006 and 2009: return on equity, return on assets, operating income scaled by sales revenue, income before tax scaled by sales revenue, Net Income (Loss) - Consolidated scaled by sales revenue, and income before extraordinary items scaled by sales revenue.
Aliabadi et al. (2013) find that both American companies and companies using IFRS have significant correlation between current year market performance and current year accounting performance measures. The most informative accounting measure is return on assets (ROA) whose adjusted R2 is the highest of 0.0421 among those six ratios. One explanation is that ROA is the only measure among the six that includes both uses (revenues and expense) and sources of
5 In the article, most value relevant measures should be called most informative measures because the measures are financial ratios rather than absolute accounting measures such as profits and the research uses stock return as dependent variable.
assets. It should be noted that the outliers have been excluded without specifying the methodology.
Based on these reviews, we use ROA as a representative of all the financial ratios.
Aliabadi et al. (2013)’s research regresses current year stock returns on current year financial ratios and test the confirmatory ability of financial ratios. The following articles are more on the predictability of financial ratios.
Fama et al. (1988, 1992, 1995), Kothari et al. (1997) and Pontiff et al. (1998) find that book-to-market ratio (B/M) and dividend yield (DY) can strongly predict stock returns, especially in the long run. According to Fama et al. (1988), DY typically explain less than (R2) 5% of the variances of short term - monthly or quarterly returns but explained more than (R2) 25% of the variances of long term – two to four year returns.
Lewellen (2004) expands previous researches to improve the predictability of stock return by the financial ratios. His research includes both ratios B/M and DY, used by previous researchers, and earning yield (EY). The previous theoretical and empirical literatures indicate that these three financial ratios are most useful and effective on stock return predictability. Above several studies cover mainly US companies.
Using the result of Lewellen (2004), Kheradyar et al. (2011) test the predicting power of book-to-market ratio (Bt-1/Mt-1), dividend yield (DYt-1) and earning yield (EYt-1) individually and the
combination against subsequent year stock return (Rit) for the companies listed in Malaysia Stock
Exchange. Individual predicting power (adjusted R2) of the three ratios ranges from 0.0008 to 0.0247. After combination of the three ratios, the predicting power (adjusted R2) increases to between 0.0216 and 0.0264.
Ou et al. (1989) use one summary measure composing of 16 and 18 financial ratios for US companies between 1965 and 1977 and find that the result can predict one year ahead earnings change. The research uses multivariate logit earnings prediction model to regress future earnings on current year financial ratios (including ROA). The maximum likelihood estimate of the coefficient on the ROA is negative and statistically significant (i.e., there is a negative relationship between current year ROA and future year earnings and consequently stock return).
Based on this predictability of those financial ratios, Ou et al. (1989) set up long and short positions with zero net investments. The excess stock return is about 7% after adjusting the size effect.
Upadhyay et al. (2012) conduct a research on forecasting stock performance of Indian stock market. Using seven financial performance measures and ratios, Book Value (BV) ,
PBIDT/Sales(PBIDTS) and Earnings per Share(EPS), Percentage change in operating
profit(OP), Percentage change in net sales(NS), Price to Cash earnings per share(PECEPS) and Price to book value(PEBV) in the Multinomial Logistic Regression, they find that there is a predictability of seven financial measures and ratios in the regression. The model they build can predict excess stock return with 56.8% accuracy.
In summary, a single financial ratio or a set of financial ratios of current year have correlation relationship with current and/or future stock return. A single or even multiple financial ratios’ correlation with short term or long term stock return is generally small. Our research in chapter 4 generally has small R2 which conforms with these researches.
2.3 Adjusted financial ratio in practice - CFROI® and CROCI®
An adjusted financial ratio is used because it is believed to have more information content compared with a traditional financial ratio. CFROI® and CROCI® are two most well-known adjusted financial ratios in the investment community.
CFROI® (cash flow return on investment) framework was developed by Holt, a subsidiary of Credit Suisse First Boston. CFROI® is based on the ROA framework with adjustments in both numerator and denominator to arrive at gross cash flow (adjusted return) and gross investments (adjusted total assets). It falls into the category of adjusted financial ratio. The purpose is to provide the investors a consistent approach to benchmark corporate performance across time and countries. It is widely used in investment decision analysis (Holland et al., 2008).
More precisely, CFROI® = (Gross Cash Flow - Economic Depreciation)/ Gross Investment. Conceptually CFROI® tries to calculate Internal Rate of Return on an investment’ cash out and cash in over its life period (See the illustration graph from CFROI® below).
CFROI® is claimed to be a better financial ratio than a traditional accounting ratio such as ROE or ROCE to evaluate corporate performance (Holland et. al. (2008)). According to Holland et al. (2008), the CFROI® is based on economic principles and not accounting principles. Accounting measures compared with economic measures are more subject to human judgments and manipulations.
One important element to make CFROI ratio more economic principle based is to capitalize R&D costs in the ratio calculation.
In the formula to calculate CFROI®, Holt adds R&D expense back to the gross cash flow (numerator) and capitalizes R&D expenditure by adding back to gross investments (denominator).
Deutsche Bank uses the similar approach to capitalize R&D expense in their CROCI® calculation6. Deutsche Bank claims that CROCI® increase R2 from 0.3 to 0.5 between price/book value and return on equity to about 0.7 between enterprise value to net capital invested and to cash return on capital invested (CROCI®) (Deutsch Bank, 2013). A higher R2 means higher informativeness.
However there are no details released to substantiate the claim of CROCI® or CFROI ® that the ratios are with higher informativeness after capitalization of R&D expense. This study will fill in the gap in these applications.
Because CFROI® and CROCI® use a ratio similar to adjusted ROA7, we use ROA and adjusted ROA as a representative of financial or adjusted ratios in the research.
2.4 R&D expense, Disclosure and Capitalization
Current USGAAP prohibits capitalization of R&D expense. According to FAS 2, R&D expenditures have to be fully expensed after 1975. Capitalization of R&D expenditures was allowed before 1974. The exception is the development expenditure of computer software which may be capitalized subject to specific criteria (USGAAP: ASC 985-20, Software — Costs of Software to be Sold, Leased or Marketed and ASC 350, Intangibles — Goodwill and Other). The USGAAP is partially contrary to IAS 38 of IFRS which allows capitalization of development cost (not limited to development costs of computer software) if it meets specific requirements. The specific requirements include demonstrating technical feasibility, intent to complete the
6 Please refer to Deutsche Bank (2013) for the details how to calculate CROCI® ratio. 7 Similar: in terms of capitalization of R&D expense.
asset, and ability to sell the asset in the future etc. Research costs, similar to USGAAP, should be expensed when it occurs.
Key characteristic of USGAAP is that it is a conservative and practical solution which increases the reliability and comparability of financial statements across different entities. However it decreases the value relevance of financial statements.
“Immediate expensing of all R&D costs emphasizes that writing off as an expense of the present period expenditures made with the expectation of benefiting future periods, is an example of revenue/expense mismatching and cannot be justified on the grounds of sound accounting principles. Furthermore, precluding capitalization of all R&D costs removes from the balance sheet what may be a company's most valuable asset” (Gornik-Tomaszewski et al., 2005).
For example Bristol-Myers Squibb Co, a leading global pharmaceutical company, with revenue of USD 21.2 billion and total assets of USD 32.9 billion expensed USD 3.8 billion in R&D
expenditures in fiscal year 2011. Writing off such a huge potentially valuable assets affects the accounting profit in financial statement on one hand and may not affect the economic profit in investor’s mind, in reality and stock price on the other hand.
At the end of 2013, technology and healthcare sectors account for about 32% of S&P index. Impact of losing value relevance due to expensing R&D expenditure cannot be underestimated. Various researches support capitalization of R&D expenditure to increase the value relevance of financial statement.
Dukes (1976) concludes that investors adjust reported earnings for the full expensing of R&D. Zion (1978) shows that firms' market value minus book value is correlated with R&D
expenditures.
Similarly Chan et al. (1990) find that shareholders are positive about the R&D announcement and incorporate the positive view in the stock return.
According to Lev et al. (1996), capitalization of R&D expenditure of 1300 US companies between 1975 and 1991, the adjustment to reported earnings and book values for R&D capitalization strongly correlates with stock prices and returns, indicating that the R&D
capitalization process generates value-relevant information to investors. Furthermore, they find that the estimated R&D capital is associated with subsequent stock returns, indicating that the subsequent excess returns are compensating for an extra-market risk factor associated with R&D.
Nissim et al. (2000) conclude that capitalization and subsequent amortization of R&D expense improve the matching between the future benefits and costs for some industries but not so clear
for some other industries. Overall their study supports the capitalization of R&D expense though they would recommend not capitalizing for some industries.
Zhao (2002) further expands Lev’s scope from USA to a more international perspective. Four countries are selected, USA and Germany which prohibit capitalization of R&D expenditure, UK and France which allow conditional capitalization of R&D expenditure.
Zhao (2002) concludes that (a) The reporting of total R&D costs increases the association of equity price with accounting earnings and book value in countries with full R&D expensing model; (b) The allocation of R&D costs between capitalization and expense provides incremental information content over that of total R&D costs in countries with conditional capitalization of R&D expenditure model. The research supports the value relevance of R&D in investment decision process.
Using data of Swedish companies, Heshmati et al.(2005) find that R&D is not only a good predictor of future growth in foremost profit and employment, but also sales, value added and cash flows. Therefore R&D expense is an expense with future benefits. Therefore we can conclude that expensing R&D is not only conflicting with the matching revenue to expense principle, the basic accounting principle but also with the economic reality.
Cohen et al.(2013) find that stock market fails to take past track record of success rate of R&D investments into account when valuating US firms. A long-short portfolio strategy that takes advantage of the information in past track records earns an annual abnormal return of roughly 11%. Past track records can predict future success of R&D projects.
It should be noted that stock market does not always view R&D positively. As mentioned in the research of Lev et al. (1993), relative change in R&D expense is one of the exceptional
fundamentals that does not always have positive or negative relationship with cumulative abnormal stock return. According to Hall (1992), the investor valuation of intangible assets created by R&D in US manufacturing companies has dropped from roughly the same as ordinary tangible assets between 1979 and 1983 to about 20% to 30% between 1986 and 1991. Therefore R&D projects seem destroying shareholder value.
Although with those controversies, overwhelming researches and practices (refer to section 2.3: CROCI® or CFROI ®) support the value relevance of capitalization of R&D expenditure. In our research we follow this trend and add back R&D expenses to the income and capitalize them in the assets for a more informative adjusted financial ratio.
There are several facts worth pointing out. For US companies, accounting treatment of R&D expenditure changed after 1975 when no capitalization was allowed but some capitalizations were allowed prior to that. Therefore it is logical to assume that, for US companies, the
informativeness of ROA is higher before 1975 than that after 1975. This research will test this assumption in Hypothesis 4 outlined in later chapters.
Another point is about capitalization estimate. Even though users are capitalizing R&D expense by themselves according to various studies, the accounting disclosure is not always detailed enough.
In the case of 2011 annual report of Bristol-Myers Squibb Co. in Appendix 2, regarding R&D expense, 2 billion US dollar was spent on development out of 3.8 billion US dollar total R&D expense. Late stage development projects are explained in good details from page 5 to page 8 of the annual report. However no explanation was made on research projects.
Furthermore certain historical statistics could be given in quarterly and annual disclosures, such as the success rate of research and development projects and the amount spent on research and development projects in the last several years to facilitate the proper valuation of the R&D. Why? Because users may already have done it by themselves.
3 Hypothesis and Research Method
3.1 Hypothesis
Because overwhelming majority of researches in chapter 2 indicate that capitalization of R&D expense is value relevant, this research hypothesizes the following:
Hypothesis 1(H1): After capitalization of R&D expense, adjusted current year ROA has higher correlation with current year stock return than ROA.
If a piece of financial information is more informative, it will lead to bigger changes in stock return or stock trading volume (Scott, 2012; Ball et al., 1968; Beaver, 1968; Kim et al., 1997), therefore it will correlate with stock return or trading volume more closely. Gentry et al. (2010) and Aliabadi et al. (2013) both confirm the correlation of ROA with current year stock market performance. Compared with ROA, adjusted ROA adds R&D expense in numerator and R&D capitalization in denominator and will be more informative. Therefore we hypothesize that adjusted ROA should correlate better with current year stock return.
Hypothesis 2 (H2): After capitalization of R&D expense, adjusted current year ROA has higher correlation with next (subsequent) year stock return than ROA does.
Financial ratios have predictability power (Fama et al., 1988, 1992, 1995; Kothari et al., 1997, Pontiff et al., 1998 and Ou et al., 1989). Adjusted ROA with more value relevant components, R&D expense and capitalization of R&D expense in numerator and denominator respectively, is more informative and should correlate better with subsequent year stock return.
Hypothesis 3 (H3): After capitalization of R&D expense in ROA calculation, the improvement on correlation with current and subsequent year stock return of EU companies is smaller than that of US companies because development costs have been capitalized by EU companies. As reviewed in chapter 2.4, accounting treatments of R&D between US and EU are different where R&D expenditures are fully expensed in US and only research expenditures are expensed in EU. R&D expenditure in Profit and Loss accounts of US companies in terms of relative amounts (compared with full R&D expenditure) is much higher than that of EU companies. Thus the problem of mismatch between cost and benefit of EU companies should be lower than that of US companies. Adding back relative higher amount to numerator and denominator will generate more information content to the ROA of US companies than to that of EU companies. Therefore we hypothesize that adjusted ROA of US companies will correlate better with current and subsequent year stock return.
The hypotheses H1, H2 and H3 are based on the following time frame: US companies: 1975 - 2012; EU companies: 2005 to 2012.
The reason of choosing 1975 for US companies is because fully expensing R&D expenditure began in 1975 in US. Prior to 1975, capitalizing of R&D expenditure was allowed. Therefore we can assume that ROA ratio of US companies before 1975 already contains most of value relevant information regarding capitalization of R&D expenditure and thus already informative.
That leads to the following additional hypothesis:
Hypothesis 4 (H4): After capitalization of R&D expenditure for US companies, the improvement on correlation with current and subsequent year stock return for the period before 1975 (excl. 1975) is lower than the period after 1975 (incl. 1975).
In our research 1967 to 1974 is chosen to represent the period before 1975.
3.2 Research Method
3.2.1 Sample selection and data collection
R2 is used as the proxy of informativeness. Higher R2 means higher informativeness. In our research, R2 means adjusted R2.
Samples are divided into four groups, US companies between 1967 and 1974, and between 1975 and 2012, international companies (excl. US companies but incl. EU companies) and EU companies.
There are several reasons of this sample grouping.
The first reason has been explained. The accounting treatment of R&D expenditure for US companies is different before and after 1975. Therefore US companies are divided into two groups.
The second reason is that there are many more US companies in the samples. If we mix US samples with much fewer international samples, our research result will have statistical bias. The third one is because different accounting standards and disclosures have impact on the informativeness of financial ratios (Alford et al. 1993). To minimize this impact, we divide the samples into US companies, and international companies. According to IASplus website from Deloitte (2014), among 173 jurisdictions, IFRS is not permitted by only 25. More than 85% of jurisdictions allow full or partial usage of IFRS. Therefore it is a reasonable assumption that among international companies most of them are using IFRS or some part of it. Thus they are
under similar accounting regime. We further divide international companies into non EU companies and EU companies because compared with non EU companies, EU listed companies are obliged to use IFRS since 2005 and corporate governance is similar among EU companies 8. This quantitative research focuses on all listed companies whose key financials are available in Compustat/CRSP database such as total assets, R&D expense as well as stock return etc. 9
For US companies, annual accounting data are taken from the database of Compustat Monthly Updates - Fundamentals Annual. Stock market data such as dividend, opening, closing share price and number of shares are taken from the databases of CRSP/Compustat Merged Database - Fundamentals Quarterly and CRSP Monthly Stock .
For international companies (incl. EU companies), annual accounting data are taken from the database of Compustat Global - Fundamentals Annual. Stock market data are taken from the database of Compustat Global - Security Daily.
Data time coverage is between 1975 and 2012 for hypotheses 1 to 3 for US companies and is between 2005 and 2012 for international companies. The reason to use the data from 2005 is because all EU listed companies are required to report in IFRS after 2005 (including 2005). Please refer to appendix 1 for the list of EU countries defined in this research. For hypothesis 4, a period from 1967 to 1974 is used.
Ordinary least squares Linear (OLS) regression model is used.
Only companies with positive (debit to Profit and Loss accounts) R&D expenditure are used in the research. Companies missing key accounting data (net income and assets) or key stock return data (number of shares, opening or closing share price) are excluded.
No extreme data (outliers) are excluded from independent variables.
1% outliers of stock return are excluded from the dependent variables in each of the regressions. 1% outliers means that top 1% of highest value from the samples and bottom 1% of lowest value from the samples are excluded. Please refer to the following tables for the derivation of samples
8 Even among EU countries corporate governance can be quite different. However we assume that corporate governance inside EU is more similar than that outside of EU. For example corporate governance of Italy should be closer to that of France than that of India.
Table 3.1. Derivation of Samples ( Regression of Fiscal Year t stock return on Fiscal Year t ratios)
US 1975 - 2012 Firm year with available net income, assets, positive R&D expense and current year stock return 65,488 Firm year after excluding 1% samples with highest and lowest current year Stock Return 64,178 International 2005-2012 Firm year with available net income, assets, positive R&D expense and current year stock return 24,101 Firm year after excluding 1% samples with highest and lowest current year Stock Return 23,619 EU 2005-2012 Firm year with available net income, assets, positive R&D expense and current year stock return 4,777 Firm year after excluding 1% samples with highest and lowest current year Stock Return 4,681 US 1967 - 1974 Firm year with available net income, assets, positive R&D expense and current year stock return 3,603 Firm year after excluding 1% samples with highest and lowest current year Stock Return 3,531 Table 3.2. Derivation of Samples (Regression of Fiscal Year t+1 stock return on Fiscal Year t ratios)
US 1975 - 2012 Firm year with available net income, assets, positive R&D expense and subsequent year stock return 61,420 Firm year after excluding 1% samples with highest and lowest subsequent year Stock Return 60,192 International 2005-2012 Firm year with available net income, assets, positive R&D expense and subsequent year stock return 13,302 Firm year after excluding 1% samples with highest and lowest subsequent year Stock Return 13,036 EU 2005-2012 Firm year with available net income, assets, positive R&D expense and subsequent year stock return 2,367 Firm year after excluding 1% samples with highest and lowest subsequent year Stock Return 2,320 US 1967 - 1974 Firm year with available net income, assets, positive R&D expense and subsequent year stock return 3,362 Firm year after excluding 1% samples with highest and lowest subsequent year Stock Return 3,295
3.2.2 OLS Regression formula
Revision from the OLS regression model used by Aliabadi et al. (2013) is used.
Rit or Rit+1 = α0+ α1 ROAit or adjusted ROAit + α2 sizeit or adjusted sizeit+ α3 CAPEXit or adjusted CAPEXit+ α4 Leverageit or adjusted leverageit+ α5 liquidityit or adjusted liquidityit +εit
Rit (share price 3 months of firm i after the end of fiscal year t10 * number of shares11 + dividend per share12 during last 12 months of the same period*number of shares13) – (share price 3 months of firm i after the end of fiscal year t-1* number of shares). Compared with the approach that takes only closing share price into account, this approach can eliminate the mistakes due to stock split/merge and ignorance of dividends.
10 Share price: For US companies, item: PRCCQ -- Price Close – Quarter of database: CRSP/Compustat Merged Database - Fundamentals Quarterly is used. For international companies, item PRCCD -- Price - Close – Daily from database: Compustat Global - Security Daily is used and 01W of Issue ID - Daily Price is used.
11 Number of shares: For US companies, item: SHROUT -- Number of Shares Outstanding in in database: CRSP Monthly Stock is used. The reason why we do not use the number of shares in database: CRSP/Compustat Merged Database - Fundamentals Quarterly is because the number of share information of a large number of companies is missing. For international companies, item CSHOC -- Shares Outstanding from database: Compustat Global - Security Daily is used and 01W of Issue ID - Daily Price is used.
12 Dividend per share: For US companies, item: DVPSXQ -- Div per Share - Exdate – Quarter of database: CRSP/Compustat Merged Database - Fundamentals Quarterly is used. For international companies, item: DIV -- Dividends per Share - Ex Date - Daily (Issue) from database: Compustat Global - Security Daily is used and 01W of Issue ID - Daily Price is used.
13 Number of shares for dividend calculation: for US companies, number of shares of 3 months after the beginning of the fiscal year t is used. For international companies, number of shares at the end of every month with dividend pay-out between 3 months after the beginning of fiscal year t and 3 months after the ending of fiscal year t is used.
Rit+1: (share price 3 months after the end of fiscal year t+1* number of shares + dividend per share during last 12 months of the same period*number of shares14) – (share price 3 months after the end of fiscal year t * number of shares).
All financial ratios or adjusted financial ratios are based on financial information of firm i at the end of fiscal year t. For example ROAit means return on asset ratio of firm i at the end of fiscal year t. The data item abbreviation and data name (abbreviation letters -- item name) given below is the same as what is shown in Compustat.
ROA or adjusted ROA: ROA = NICON -- Net Income (Loss) - Consolidated / AT -- Assets - Total. Adjusted ROA = (NICON -- Net Income (Loss) - Consolidated + XRD -- Research and Development Expense)/(AT -- Assets - Total + 5 * XRD -- Research and Development Expense)
Size or adjusted size: size = log (AT -- Assets - Total). Adjusted size= log (AT -- Assets - Total + 5 * XRD -- Research and Development Expense)
CAPEX or adjusted CAPEX: CAPEX = CAPX -- Capital Expenditures / AT -- Assets - Total. Adjusted CAPEX =( CAPX -- Capital Expenditures + XRD -- Research and Development Expense)/( AT -- Assets - Total + 5 * XRD -- Research and Development Expense).
Leverage or adjusted Leverage: Leverage = (DLC -- Debt in Current Liabilities - Total + DLTT -- Long-Term Debt - Total)/ (AT -- Assets - Total). Adjusted leverage = (DLC -- Debt in Current Liabilities - Total + DLTT -- Long-Term Debt - Total)/( AT -- Assets - Total + 5 * XRD -- Research and Development Expense))
Liquidity or adjusted liquidity: Liquidity = CHE -- Cash and Short-Term Investments / AT -- Assets - Total. Adjusted liquidity = CHE -- Cash and Short-Term Investments / (AT -- Assets - Total + 5 * XRD -- Research and Development Expense).
Signs of ROA and adjusted ROA in the regressions are predicted to be positive. Higher profitability should cause higher stock return. Coefficient of ROA in the research of Aliabadi et al. (2013) is positive and statistically significant though dependent variable in that research is only current year stock return. Earnings measures have positive impact on current year stock return (Francis et al., 2003; Barton et al., 2010).
Ou et al. (1998) confirm that the relationship between current year ROA and future year stock return is negative. However most of literatures hold different views. Earnings yield and dividend yield are positively correlated with future year stock return (Fama et al. 1988, 1992 and 1995; Kothari et al., 1997; Pontiff et al., 1998; Lewellen, 2004; Kheradyar, 2011). Therefore in our
14 Number of shares for dividend calculation: similar to above footnote however the period here is 3 months after the end of fiscal year t and 3 month after the end of next fiscal year t+1.
research the correlation between ROA, with earnings in its numerator, and future stock return is predicted to be positive.
We will set predicted signs as positive in all the cases. Actual results are shown in chapter 4.1 and 4.2.
3.2.3 Capitalization of R&D expenditure
There is no research regarding how many years on average an R&D project spans. In Holland et al.’s book (2008), they use 5 years as an example to capitalize R&D expenditure. According to a presentation made by Mr. Soli Peleg (2008), most of R&D projects last between 2 to 9 years with an average of 5 years.
We pick 5 as a fixed capitalization factor in our research. We further assume that last 5 years’ (including current year) R&D expenditure is the same as current year and it is capitalized. Even if the 5 as a factor may not be an accurate estimation, the adjusted financial ratio has improved the informativeness dramatically in our research, which makes the conclusion more convincing. If an investor studies financial information such as annual accounts of an individual company deeper, he may be able to have better estimation than our model, which will further improve the informativeness of the financial ratio. This points to the need for companies to improve the disclosure regarding their R&D projects. Nevertheless, using a factor of 5 as a general rule is a limitation of our research.
3.2.4 An example of financial ratio calculation
In this chapter we take Bristol-Myers Squibb Co as an example to calculate all the ratios in this research. In fiscal year 2011, Bristol-Myers Squibb Co. spent USD 3,839 million on R&D. In our calculation, the capitalization of R&D expense is 5 times of USD 3,839 million, i.e. USD 19,195 million.
Table 3.3. An example of (adjusted) financial ratio calculation Fiscal Year 2011 Amount in million USD
Research and Development
Expense Net Income (Loss) Assets - Total Debt in Current Liabilities - Total Long-Term Debt - Total Cash and Short-Term Investments Capital Expenditures
3,839 3,709 32,970 115 5,376 8,733 367
ROA Size Leverage Capex Liquidity
11% 4.52 17% 1% 26%
Capitalization of Research and Development Expense
Adjusted Net Income
(Loss) adjusted Assets - Total Debt in Current Liabilities - Total Long-Term Debt - Total Cash and Short-Term Investments Adjusted Capital Expenditures
19,195 7,548 52,165 115 5,376 8,733 4,206
adjusted ROA adjusted Size adjusted Leverage adjusted Capex adjusted Liquidity
14% 4.72 11% 8% 17%
There is a material difference between ROA and adjusted ROA. After the capitalization, the ROA ratio is increased by about 27%. Leverage ratio is decreased by more than one third and Capex ratio jumps 7 times. The key ratios are changed.
4 Result
4.1 Regression of Fiscal Yeart stock return on Fiscal Yeart ratios
4.1.1 Descriptive statistics and Pearson Correlation Matrix for the dependent and independent variables. Table 4.1. Descriptive statistics: Regression of Fiscal Year t Stock return on Fiscal Year t ratios
Items
Current year stock
return ROA Size Leverage Capex Liquidity adjusted ROA adjusted Size Leverage adjusted adjusted Capex Liquidity adjusted capitalization of RD Assets - Total US companies (1975 - 2012) USD m. USD m. Mean 0.214 -0.092 2.137 0.187 0.054 0.232 0.025 2.284 0.148 0.091 0.148 407 2,522 Standard Error 0.003 0.002 0.004 0.001 0.000 0.001 0.001 0.004 0.001 0.000 0.001 9 68 Median 0.058 0.032 2.035 0.141 0.039 0.138 0.062 2.196 0.103 0.085 0.102 30 108 Standard Deviation 0.773 0.548 0.981 0.287 0.054 0.241 0.204 0.936 0.166 0.049 0.144 2,260 17,124 Range 6.078 67.646 7.258 39.593 1.039 1.008 25.014 6.402 3.786 0.882 0.994 60,915 797,769 Minimum -0.864 -61.197 -1.357 0.000 -0.037 -0.010 -19.897 -0.492 0.000 0.000 -0.006 0 0 Maximum 5.213 6.450 5.902 39.593 1.003 0.998 5.116 5.910 3.786 0.882 0.988 60,915 797,769 Sum 26,118,562 161,876,470 Count 64,178 64,178 64,178 64,178 64,178 64,178 64,178 64,178 64,178 64,178 64,178 64,178 64,178 International companies (2005 - 2012) Mean 0.143 -0.029 3.162 0.210 0.046 0.187 0.033 3.231 0.186 0.042 0.150 Standard Error 0.004 0.011 0.009 0.002 0.001 0.001 0.010 0.008 0.001 0.000 0.001 Median 0.009 0.032 3.215 0.164 0.029 0.125 0.045 3.257 0.145 0.025 0.109 Standard Deviation 0.636 1.755 1.344 0.371 0.087 0.190 1.471 1.295 0.205 0.052 0.140 Range 4.927 342.481 9.536 24.067 10.267 1.043 236.539 8.931 7.154 1.050 1.013 Minimum -0.805 108.333 -1.523 - 0.000 0.000 -0.043 -13.595 -0.917 0.000 0.000 -0.043 Maximum 4.122 234.147 8.013 24.067 10.267 1.000 222.944 8.013 7.154 1.050 0.970 Count 23,619 23,619 23,619 23,619 23,619 23,619 23,619 23,619 23,619 23,619 23,619 EU companies (2005 - 2012) Mean 0.130 -0.051 2.384 0.194 0.039 0.187 0.027 2.496 0.165 0.033 0.129 Standard Error 0.009 0.005 0.016 0.003 0.001 0.003 0.002 0.015 0.002 0.001 0.002 Median 0.009 0.031 2.263 0.162 0.027 0.106 0.052 2.369 0.132 0.021 0.087 Standard Deviation 0.644 0.337 1.103 0.224 0.045 0.206 0.155 1.054 0.160 0.040 0.126 Range 4.655 10.129 6.162 4.951 0.731 0.996 3.377 5.985 1.364 0.727 0.952 Minimum -0.842 -4.457 -0.623 0.000 0.000 0.000 -2.653 -0.440 0.000 0.000 0.000 Maximum 3.813 5.672 5.538 4.951 0.731 0.996 0.724 5.545 1.364 0.727 0.952 Count 4,681 4,681 4,681 4,681 4,681 4,681 4,681 4,681 4,681 4,681 4,681
Items
Current year stock
return ROA Size Leverage Capex Liquidity adjusted ROA adjusted Size Leverage adjusted adjusted Capex Liquidity adjusted capitalization of RD Assets - Total US companies (1967 - 1974) USD m. USD m. Mean -0.056 0.050 2.068 0.254 0.066 0.072 0.064 2.116 0.231 0.079 0.064 62 540 Standard Error 0.006 0.001 0.012 0.003 0.001 0.001 0.001 0.012 0.002 0.001 0.001 5 32 Median -0.095 0.053 2.008 0.252 0.054 0.049 0.067 2.049 0.225 0.071 0.044 8 102 Standard Deviation 0.332 0.066 0.731 0.151 0.048 0.072 0.060 0.729 0.142 0.045 0.064 288 1,875 Range 1.925 1.229 4.466 1.223 0.472 0.656 1.171 4.339 1.139 0.470 0.597 6,848 54,546 Minimum -0.705 -0.778 0.271 0.000 0.000 -0.005 -0.767 0.403 0.000 0.000 -0.005 0 2 Maximum 1.220 0.451 4.737 1.223 0.472 0.650 0.404 4.742 1.139 0.471 0.592 6,848 54,548 Sum 217,376 1,907,072 Count 3,531 3,531 3,531 3,531 3,531 3,531 3,531 3,531 3,531 3,531 3,531 3,531 3,531 Notes:
Current year stock return (Rit): (share price 3 months of firm i after the end of fiscal year t * nr of shares + dividend per share during last 12
months of the same period*number of shares) – (share price 3 months of firm i after the end of fiscal year t -1* number of shares). Compared with the approach that takes only closing share price into account, this approach can eliminate the mistakes because of stock split/merge and ignorance of dividends.
All financial ratios or adjusted financial ratios are based on financial information of firm i at the end of fiscal year t (i.e. current year). For example ROA means return on asset ratio of firm i at the end of fiscal year t. The item name is the same as what is shown in Compustat.
ROA or adjusted ROA: ROA = NICON -- Net Income (Loss) - Consolidated / AT -- Assets - Total. Adjusted ROA = (NICON -- Net Income (Loss) - Consolidated + XRD -- Research and Development Expense)/(AT -- Assets - Total + 5 * XRD -- Research and Development Expense) Size or adjusted size: size = log (AT -- Assets - Total). Adjusted size= log (AT -- Assets - Total + 5 * XRD -- Research and Development Expense)
Leverage or adjusted Leverage: Leverage = (DLC -- Debt in Current Liabilities - Total + DLTT -- Long-Term Debt - Total)/ (AT -- Assets - Total). Adjusted leverage = (DLC -- Debt in Current Liabilities - Total + DLTT -- Long-Term Debt - Total)/( AT -- Assets - Total + 5 * XRD -- Research and Development Expense))
CAPEX or adjusted CAPEX: CAPEX = CAPX -- Capital Expenditures / AT -- Assets - Total. Adjusted CAPEX =( CAPX -- Capital Expenditures + XRD -- Research and Development Expense)/( AT -- Assets - Total + 5 * XRD -- Research and Development Expense). Liquidity or adjusted liquidity: Liquidity = CHE -- Cash and Short-Term Investments / AT -- Assets - Total. Adjusted liquidity = CHE -- Cash and Short-Term Investments / (AT -- Assets - Total + 5 * XRD -- Research and Development Expense).
Standard Error: It is estimated by the standard deviation of one measurement of the mean divided by the square root of n:
n 1
n x x n s n 1 2 i
Standard Deviation: The standard deviation is a measure of how widely values are dispersed from the average value (the mean).
1 n x x s 2 i
Table 4.1 provides the descriptive statistics of the dependent and independent variables. International companies are represented by 78 countries (such as China) and regions (such as Hong Kong). EU companies are represented by 27 countries.
It reveals that stock returns of US companies between 1975 and 2012 are much higher than the rest three groups in both mean and median value. Compared with international and EU companies, stock returns of US companies between 1975 and 2012 are more volatile because standard deviation in the table is bigger than that of international and EU companies. The range of stock returns of US companies between 1975 and 2012 is also larger than that of international and EU companies.
However in terms of ROA, the picture is mixed. ROAs of US companies between 1975 and 2012 are less volatile than that of international companies but more volatile than EU companies because US companies have bigger standard deviations and larger range of ROA.
In general, compared with international and EU companies, US companies between 1975 and 2012 have higher current year stock returns but have lower mean current year ROA and the same median ROA. This is contradictory to the basic economic theory because it means companies with lower ROA will generate higher stock return.
Furthermore, US companies between 1975 and 2012 also have lower leverage, higher CAPEX and higher liquidity. Normally lower leverage and higher liquidity means lower risk profile. This indicates that risk profiles of US companies are probably not the cause of their higher stock returns.
To further investigate this contradiction, we conduct a similar analysis by using adjusted ROA ratio. The mean of adjusted ROA of US companies between 1975 and 2012 is 0.025 which is lower than that of international and similar to EU companies. However the median adjusted ROA of the same US companies is 0.062 which is higher than 0.045 of international companies and 0.052 of EU companies. US companies are more profitable if using median adjusted ROA. The adjusted ROA improves the informativeness of ROA.
With same adjustment, adjusted leverage ratio of US companies is much lower than that of international and EU companies and adjusted liquidity ratio is much closer to that of international and EU companies. We conclude that US companies between 1975 and 2012 have lower risk profile than that of international and EU companies but have higher stock returns. Thus US companies as a whole are better investment choices.
It should be noted that after the adjustment, difference in adjusted CAPEX ratio between the same US companies and international/EU companies becomes much bigger because mean and median of US companies are much higher than those of international and EU companies. EU
