The Competitive Advantage Period Revived:
Investment Opportunity Approach vs. Economic Profit
Approach
Abstract
This paper looks into the measurement and usefulness of the competitive advantage period by drawing a comparison between the traditional investment opportunity approach and the economic profit approach. The sample consists of US firms from the Russell 3000 Index over the period 1996-2015 and includes 6,256 observations for the investment opportunity approach and 12,363 for the economic profit approach. The evidence of this study shows that the economic profit approach is a more reliable measure for the competitive advantage period than the investment opportunity approach, because it leads to both a higher number of observations and more stable estimates. Convincing evidence is also found that the competitive advantage period is sensitive to industry-specific effects and firm size and is not constant over time.
Keywords: Competitive advantage period, investment opportunity approach, economic
profit approach Iris Schoolderman s2199483 MSc Finance 13-06-2016 11,872 words
1
1. Introduction2
This paper takes the paper by Mauboussin and Johnson (1997) as a starting point and examines whether the competitive advantage period becomes more reliable in terms of volatility and extreme values when the economic profit approach is applied instead of the investment opportunity approach. Therefore, the main research question of this paper is:
“Is the economic profit approach a better approach to measure the competitive advantage period than the investment opportunity approach?”
To determine which approach is better both approaches are tested by looking at factors such as industry, volatility and size. T-tests are performed to see if there are large fluctuations in means or variances of CAPs using both approaches. When large fluctuations are found, this indicates that the CAP is not a very stable and reliable measure. Finally, regressions are performed to see the effect of some variables on the CAP, such as research and development costs. This paper uses the Russell 3000 Index, which is an index consisting of the 3000 largest stock-listed firms in the US. For these firms, the CAP has been calculated for the period from 1996 until 2015.
Even though the CAP is not used often throughout finance literature it can be a useful addition to the existing literature. The most important reason for this is that the CAP is a measure that can provide very quick insights into the corporate performance of a firm measured in competitive advantages. It is a raw measure, but for this reason it can also be quickly calculated, which is an advantage. The CAP is also easy to understand. The inputs for the formula are simple and can be obtained relatively easy. The outcome of the formula provides you with a clear number which represents the expected years of competitive advantage. The competitive advantage period according to the formula is based on the invested capital of the firm and the excess returns it earns on this invested capital. If there are no excess returns, the CAP will become negative. After all, firms with negative returns suffer from competitive disadvantages instead of advantages. This result is in accordance with the literature and thus makes the CAP a realistic measure. Finally, the CAP can show how successful firms really are and over what kind of period they will remain successful. This also means that you know how far to look ahead in the future and how to better allocate your resources to sustain your CAP. All this information makes the CAP a useful measure for managers, because it gives an insight into the sustainable value creation of a company. The CAP can also be of interest to investors, because they try to find firms to invest in that are mispriced relative to financial expectations (Mauboussin and Callahan, 2013). The CAP is a market-based measure that reflects the expectations of the market and usually firms with higher CAPs enjoy higher valuations.
Saint-3
Pierre, 2008). It can therefore complement finance theory. The concept is useful, because it has been applied in valuation before (Koller et al., 2010), but not many papers focus purely on the concept of CAP. This paper purely focuses on contributing to the knowledge on the CAP. The EPA is used for the first time to estimate the CAP. This approach gives less volatile results and fewer negative values for the CAP than the IOA, which implies that the EPA is a more reliable approach. One of the issues with the CAP that has been neglected in the literature is the fact that the CAP can return negative values in some cases. This aspect of the CAP is well-explained and identified in this paper for both approaches. This paper also uses a larger data set than has been used before, which improves the reliability of the results. Along with that, this paper uses a large timeframe, spanning an entire economic cycle. Finally, this paper complements the existing literature because it applies the enterprise method to evaluate the principle of CAP instead of the equity method. This is a more informative measure, because it focuses on the entire capital structure of the company and not just its equity.
The remainder of this paper is structured as follows: section two reviews the literature and develops hypotheses. Section three documents the data construction and explains the methodology. Section four presents the empirical results of this research and section five discusses the conclusions and recommendations of this paper.
2. Literature Review
This section reviews the concept of the competitive advantage period. Next, both the investment opportunity approach and the economic profit approach are explained. Finally, the hypotheses of this paper are developed. This literature review aims to provide structure and a guideline for the further analysis in this paper.
2.1 Competitive advantage period
4
based on an important prediction of economic theory, namely that profitability is mean reverting (Fama and French, 2000). Their results also show that the rate of reversion is around 38% per year and that mean reversion happens more quickly for firms that have profitability below the mean than for firms above the mean. This is because in the case of low profitability, firms are pressured by their competitors to make better use of their assets and allocate them more efficiently to survive. Therefore, they might be better able to change their course of action. In order for the competitive advantages of a firm to become sustainable, it must be difficult for current and potential competitors to duplicate the strategy (Barney, 1991; Grant, 1991). Therefore, the search for sustainable competitive advantages must focus on valuable, rare, imperfectly imitable and non-substitutable resources. When these conditions are met, it will be extremely difficult for competitors to copy a strategy and/or resources. This is in accordance with the resource-based view towards competitive advantage.
The resource-based view (RBV) is a way of looking at competitive advantages of firms in which one looks at firm-specific characteristics and the way in which these characteristics have an impact on the rate of mean reversion (Peteraf, 1993). The RBV is also described as the business strategy level of a firm and the relationships within that firm between resources, competition and profitability (Grant, 1991). The RBV is often used in the field of strategic management, because the RBV focuses on resources, products and their demand, which together form a strategic process. The resources of a firm are whatever exists within a firm that can form either a strength or a weakness for the firm (Wernerfelt, 1984; Barney, 1991). Whenever a firm is able to create durable and unique resources which are difficult for competitors to imitate, this could be a source of competitive advantage. So, creating durable and unique resources is a way in which firms can make excess returns, grow and create value. Therefore, it is crucial to focus on resource heterogeneity and immobility so that competitors cannot duplicate the strategy (Barney, 1991). Peteraf (1993) also focuses on firm heterogeneity, because this proves that some firms possess superior productive factors that other firms do not.
5
values cash flows more than earnings. The second reason is that firms usually apply a strategic-planning forecast period that is different from their CAP. This leads to more internal oriented expectations instead of market-based external expectations in their communication with investors.
According to Mauboussin and Callahan (2013) investors have a goal of understanding today’s stock prices and anticipating future changes in expectations. Therefore, they are primarily interested in sustainable value creation when making a decision about their investments. This value creation is based on the amount of economic profit that a company earns and the duration of excess returns. The period of excess returns can be defined as the CAP of a firm, which is the period during which a firm is expected to create value. The period of value creation is important because it determines the price an investor has to pay for a firm (Mauboussin and Callahan, 2013). A CAP can increase because of factors like economies of scale, management decisions and entry barriers. Therefore, the CAP can be a useful instrument to base your investment decisions on (Mouelhi and Saint-Pierre, 2008). Koller et al. (2010) determine firm value as the present value of future cash flows of a firm. Value creation can only happen through growth when the return on invested capital exceeds the cost of capital. If this is not the case, a firm is destroying value. The link between the CAP and valuation can be found in the sustainability of competitive advantages. The CAP can be used in corporate valuation by looking at the period during which a firm has excess returns and thus can create value. The CAP is a transformation of the market-to-book ratio, which is a measure that reveals the market’s expectations about the persistence of excess returns. Therefore, the CAP can also reveal market expectations and therefore it can be an important measure to determine firm value.
2.2 Investment opportunity approach
Previous literature on the CAP has focused on calculating the CAP using the investment opportunity approach (IOA). This method focuses on net investments and assumes a constant stream of profits from these investments. Firm value is measured based on two factors: The value of the assets that the firm has in place and the present value of the firm’s growth opportunities (Modigliani and Miller, 1961). Following Koller et al. (2010), the value of the growth opportunities can be measured as the net present value of the future net investments of the firm and there are only excess returns in case of positive net present value projects. The value of a firm when using the investment opportunity approach is equal to:
∑
,
(1)6
investments and i is return on net investments. The first part of the formula represents the value of assets-in-place and the second part the value of growth opportunities. A large part of the market value of firms is determined by the value of growth opportunities (Danbolt et al. 2000; Danbolt et al. 2011). Appendix A presents the extended derivation of the CAP. Future return is assumed to be equal to the return on net investments for the first T years. After this period excess returns disappear. This leads to an annuity that can be solved. From equation (A.4) the ones in the subscripts need to be eliminated. This is done by assuming that the investments do not lead to excess returns in the future, but including them in the term for growth opportunities. The final equation for the CAP using the IOA is:
-1 (2)
The most important flaw of this measure is that it assumes a constant stream of net profits from investments for the assets in place. This assumption is made even if there are substantial excess returns, while eventually there should be convergence (Mauboussin and Johnson 1997). This leads to too much value being attached to the assets-in-place and too little to the growth opportunities. Therefore, the CAPs become too short, or even negative.
A practical implementation of this argument can be observed when after-tax EBIT is high relative to the value. This causes a large value for the return measure and it does not properly take competition into account. In this way the value of growth opportunities is underestimated and the value of assets-in-place is overestimated. As a result, the CAP becomes smaller and sometimes even negative.
Another factor that might make the CAP negative is that net investments in some years can be negative when firms do not make new investments or have more write-offs than new investments. This causes a negative denominator, making the CAP negative. According to Koller et al. (2010), firms should not invest when the return on net investments is lower than the cost of capital, because then firms destroy value. However, in the data set of this paper this behavior is observed. These observations are deleted from the sample.
7
2.3 Economic profit approachThis paper applies the economic profit approach (EPA) to derive an alternative measure of the CAP. The EPA is based on the discounted cash flow approach and clean surplus accounting. The formula for the calculation of firm value applying the EPA is:
∑
,
(3)where V is enterprise value, A is invested capital, P is after-tax EBIT and k is cost of capital. Enterprise value here depends on the value of expected future economic profits. The first part of the equation refers to expected future economic profits on assets-in-place and the second part the expected economic profits on future net investments, so the value of growth opportunities. Appendix A presents the extended derivation of the CAP using the EPA. Future return is assumed to be equal to the return on invested capital for the first T years and afterwards excess returns disappear. This leads to an annuity formula which can be solved in the same way as for the IOA. The final equation for the
CAP using the EPA is:
,
(4)where a is return on invested capital. An important distinction between the two CAPs lies in the definition of the competitive advantage period. The IOA focuses on excess returns on new investments, while the EPA focuses on excess returns on all investments.
8
Hypothesis 1: “The competitive advantage period will be positive in more cases for the economic profit approach than the investment opportunity approach”. Hypothesis 2: “The competitive advantage period is on average larger for the economic profit approach than for the investment opportunity approach and the length of the competitive advantage period differs between industries”.
Hypothesis 3: "The variance of the competitive advantage period is larger for the investment opportunity approach than for the economic profit approach and the variance of the competitive advantage period differs between industries”.
2.4 Hypothesis development 2.4.1 Industry effects
Excess returns are created due to characteristics such as economies of scale, innovation and entry barriers. These factors are often industry-specific and these characteristics of industries are important determinants of performance (Porter, 1980). This is supported by Fama and French (2000) who emphasize that profitability is mean reverting to an industry mean. This implies that there is a certain threshold within an industry with regards to the return on investment that is possible in the long run and that this threshold is different between industries. It is therefore appropriate to also look into the industry-specific effects on the CAP, because one of the drivers of the CAP is the rate at which an industry changes (Mauboussin and Johnson, 1997). This means that firms in an industry with a lower rate of change might have less pressure from competitors to keep innovating or that the industry includes less competitors, which also lowers the rate of change. In industries with a lower rate of change it might therefore be easier to maintain a high CAP. Industry rate of change in this way might also lead to differences in the variance of the CAP. The variance of the CAP might be lower for industries with a low rate of change, because CAPs can be sustained easier and thus fluctuate less.
9
(2003), who observed a larger role for industry-specific than for company-specific factors in the determination of profitability. Company-specific factors only become more important for firms that are either value leaders or value losers.
2.4.2 Reversion of the competitive advantage period
Hypothesis 4: “The length of the competitive advantage period is not constant over time”.
CAPs rarely have a constant value over time (Mauboussin and Johnson, 1997). A constant CAP would imply that excess returns can persist, that the CAP is not mean reverting and that financial prices do not reflect fundamentals. If the value of the CAP does not change between two years, this means that the inputs for the CAP in the second year lead to the exact same value creation as in the previous year. This is a violation of economic theory (Fama and French, 2000) that can only be achieved for a few outstanding firms. In order for two consecutive years to have the same CAP, the inputs for the formula of the CAP need to stay the same. The passage of time without any additional changes does not lead to changes in the CAP. However, in practice the inputs for the CAP do not stay the same over time, because of changes in profitability, investments and returns between years. Overall, excess returns cannot persist because of competition and it is even more difficult for firms to sustain excess returns because of the current fast rate of innovations and technology (Mauboussin and Callahan, 2013).
2.4.3 Firm size effects
Hypothesis 5: “The size of a firm has a positive effect on the length of its competitive advantage period”.
10
2.4.4 Further hypotheses developmentHypothesis 6: “A higher ratio of fixed assets relative to total assets has a negative effect on the length of the competitive advantage period”.
If perfect accounting and perfectly competitive markets would exist, return on invested capital would equal the cost of capital. However, accounting literature has shown that accounting is almost never perfect and has a bias of around 1%-2%. This means that when return on invested capital is 10% and cost of capital is 8% it would be wrong to assume excess returns and growth. For firms with a higher ratio of fixed assets relative to total assets, the accounting bias is higher than for other firms. That implies that firms with a higher ratio of fixed assets to total assets can mistakenly believe that there are excess returns when this is actually an accounting bias. If firms base their decisions on this accounting bias, growth might actually lead to a lower enterprise value because there are no real excess returns. Therefore, a higher ratio of fixed assets to total assets might have a negative effect on the length of the CAP.
Hypothesis 7: “Higher return on invested capital has a direct negative effect on the length of the competitive advantage period”.
Firms with a higher return on invested capital are best positioned to benefit from possible competitive advantages (Mauboussin and Johson, 1997). This is either because the business of the firm is beneficial, or because it has been able to sufficiently reduce its cost level. A higher return on invested capital indicates that the firm has made beneficial investment decisions in the past, which could lead to competitive advantages. However, ceteris paribus a higher return on invested capital leads to a lower CAP. The CAP becomes smaller because the denominator in the formula becomes larger. Usually a higher return on invested capital means higher excess returns and then a firm can gain value through growth. In this way a higher return on invested capital can have positive effects and increase the CAP. However, an increase in return on invested capital and an increase in cost of capital can also move together, in which case there are no changes in the CAP. Overall, without knowledge of other changes in inputs for the CAP, a higher return on invested capital has a negative effect on the CAP.
Hypothesis 8: “Higher R&D expenditures have a negative effect on the length of the competitive advantage period”.
11
investments are left and the less the return on investment becomes. This operating profitability disadvantage is also mentioned by Dickinson and Sommers (2012), who do not see innovation leading to persistent economic rents for most firms. Because of fast innovations in technology the persistence of excess returns has become more difficult (Mauboussin and Callahan, 2013). This might also mean that more R&D expenditures are necessary just to be able to keep up with the competition and these R&D expenditures do not have a positive effect on a company’s CAP.
3. Data and Methodology 3.1 Data set construction
This paper uses a panel data set of US firms over the period 1996-2015. The US is chosen because it is a large market with many firms and the time period is as broad as possible to capture multiple economic cycles (Hawawini et al., 2003). The firms are selected from the Russell 3000 Index, because this index includes 3000 firms and a large database can thus be created. The main data source is Thomson Reuters Datastream. Data on the ten year US Treasury Bond rate from the US department of Treasury1 is used as a proxy for the risk free rate and the market risk premium is constructed in accordance with Damodaran2 (2016). All firms are assigned to one of ten different industries based on the sectors they were assigned to by the Russell Index3.
All firms in the sample must have at least ten years of consecutive data available. Not all 3000 firms in the index match this criterion, so these firms are deleted. Financials and utility firms are removed from the sample (Fairfield et al (2009); Fama and French (2000)). Firms in these industries use their assets and liabilities in a different way from the other industries and there are stricter regulations in these industries. Further exclusion of observations from the data set is based on possible inputs that lead to either extreme or unreliable (negative) CAPs. Observations with an abnormal return below 1% are deleted from the sample. These observations either have a negative return on invested capital or a return on invested capital that is below the cost of capital. It is not realistic that firms would continue to invest. The threshold of 1% is chosen to take possible accounting biases into account.
The next restriction is that both net investments and invested capital have to be positive and invested capital must be smaller than the enterprise value. This is not the case for some firms that have very large cash reserves, a negative book value of equity or a market value of equity that is lower than the book value of equity. A negative book
12
value of equity might seem impossible, but it is possible for successful firms. These firms can often borrow in excess of the book value of their assets, because creditors recognize future gains and thus assign a higher value to the assets. This leads to a debt level above the value of assets, which means that the book value of equity must be negative (Berk et al., 2014). Another data problem occurs for maturing firms. These firms have a low worth, but high current profits, which leads to a negative CAP. This is not reasonable and therefore observations where the enterprise value multiplied with cost of capital minus net operating profit is negative are also deleted. Finally, to take out the last outliers, the CAPs are trimmed at a 5% level. Table 1 summarizes the number of excluded observations for each of the conditions. In appendix B the percentage of excluded observations for each industry under each condition is calculated. A few things stand out. The number of observations that is excluded for the IOA is much larger than the number of excluded observations for the EPA, which was also expected. The number of CAPs that have positive values is much larger for the EPA than for the IOA, meaning that the first hypothesis is already confirmed. Next, a large number of observations are excluded because of excess returns below the 1% level. The information technology industry stands out from the other industries because it has a very low exclusion rate for the value-based conditions. In addition, a relatively high percentage of observations are excluded for this industry based on the trimming of data at the 95%. Therefore, information technology seems to be an industry with relatively high enterprise values. This could be expected, since information technology is a growing industry. The last striking observation is that the number of exclusions based on negative invested capital is very high for the healthcare industry. This means that these firms either have larger book values of cash than the book values of debt and equity combined or they have negative book values of equity.
3.2 Variable construction
Table 2 captures all raw data inputs that are used in the calculation of the CAP and table 3 shows the constructed data that lead to the CAP.
13
Table 1. Excluded observations and distribution of final sample Industry Initial sample
Trimming 5%-95% Final sample
Panel A: Investment opportunity approach
Total sample 25,550 10,695 4,264 2,778 861 696 6,256 Consumer discretionary 5,172 2,128 884 556 175 145 1,284 Consumer staples 1,750 439 421 314 72 56 448 Energy 1,872 917 162 363 66 29 335 Healthcare 3,989 1,613 750 431 155 108 932 Industrials 5,479 2,126 862 616 181 155 1,539 Information Technology 5,014 2,462 787 216 143 155 1,251 Materials 1,913 773 346 258 63 44 429 Telecommun ications 361 237 52 24 6 4 38
Panel B: Economic profit approach
Industry Initial sample
Trimming 5%-95% Final sample Total sample 25,552 10,695 707 413 1,374 12,363 Consumer discretionary 5,174 2,128 34 77 245 2,690 Consumer staples 1,750 439 3 45 75 1,188 Energy 1,872 917 5 69 135 746 Healthcare 3,988 1,613 514 37 206 1,618 Industrials 5,479 2,126 37 108 233 2,975 Information Technology 5,014 2,462 90 39 362 2,061 Materials 1,914 773 13 32 95 1,001 Telecommun ications 361 237 11 6 23 84
14
The return on invested capital is calculated as the average of the last two years to decrease the possibility of obscured results due to a non-recurring event in a certain year. The credit spread that is added to the cost of debt is industry-specific and based on the interest coverage ratio. The interest coverage ratio is linked to a credit spread through Damodaran’s interest coverage ratio tables4. The credit spread, equity beta, target D/E ratios and effective tax rate are all industry specific measures, based on the industry medians.
Table 2. Raw data for competitive advantage period
Symbol Variable Name Source
Market Capitalization Datastream
Code: WC07210
Book value of equity Datastream
Code: WC03501
Book value of cash Datastream
Code: WC02001
Earnings before interest and
taxes Datastream Code: WC18191
Reported Taxes Datastream
Code: WC08346
Net Income Datastream
Code: WC07250
Total debt Datastream
Code: WC03255
Interest expenses Datastream
Code: WC01251
Risk free rate US department of Treasury5
Market risk premium Damodaran (2016)6
Historical Beta
(Calculated over a five-year period using the local market index)
Datastream Code: 897E
Marginal tax rate OECD Tax Tables7
Net sales Datastream
Code: WC01001
Total assets Datastream
Code: WC02999
Research and development
costs Datastream Code: WC01201
Fixed assets as a percentage
of common equity Datastream Code: WC08266
Firm size Dummy variable for firm size
Author’s own estimation
Industry Dummy variables per industry
Russell 3000 Index
15
Table 3. Constructed data for competitive advantage period Symbol Variable Name Method of calculation
Scaled fixed assets
Scaled R&D expenditures
Book value of debt
Market value of debt
Enterprise value
Earnings before taxes
Effective tax rate
16
approach)Competitive advantage period (economic profit approach)
Note: This table contains the constructed data for the calculation of the competitive advantage period. Subscript i indicates company, j indicates industry and t indicates time.
These medians make the results less dependent on company-specific extremes and this approach is in accordance with Koller et al. (2010). The industry medians remain constant over the entire sample period, because it makes the data set less dependent on fluctuations in certain years. Robustness checks with a median that changes every year are done to check if the results are influenced by this decision. Fixed assets, return on invested capital and R&D are used as explanatory variables for the CAP in this paper. These factors are scaled to avoid the effects of spurious correlation (Brooks, 2014). Fixed assets are scaled with total assets and R&D is scaled with net sales.
3.4 Descriptive statistics
Table 4 presents the descriptive statistics of some of the main components of the CAP. The descriptive statistics for the return on invested capital, weighted average cost of capital and excess return are based on the excluded observations. All variables are trimmed at a 10%-90% level to eliminate the most extreme values.
Table 4. Descriptive statistics
Variable Obs Mean Median Std
Dev
Min Max Skewness Kurtosis
14,116 0.2 0.2 0.2 0.1 1.3 2.6 11.1 14,116 0.1 0.1 0.0 0.0 0.1 0.4 3.3
14.116 0.2 0.1 0.2 0.0 1.2 2.7 11.3 19.871 0.6 0.4 1.1 -1.5 3.8 0.7 3.3 19,707 18.2 17.5 8.9 -3.2 42.2 0.2 3.1 21,658 2.6 2.1 1.6 0.8 8.4 1.5 4.6
Note: This table presents the descriptive statistics of the main components of the CAP.
17
cases the mean is higher than the median, which makes sense here because reversion to the mean is slower for firms that are above the mean than for firms that are below the mean (Fama and French, 2000). The net investment rate and P/E ratio at the enterprise level have negative values in some cases. For both ratios this is theoretically possible when EBIT is negative and for the net investment ratio this can also happen when it saves more cash than it invests.
3.5 Estimation approach
The first hypothesis of this paper is that the CAP has a positive value in more cases for the EPA than for the IOA. In the process of excluding observations this hypothesis was already confirmed. The second and third hypotheses hypothesize that the value of the CAP is higher for the EPA and that the volatility is higher for the IOA. These hypotheses are tested with parametric tests. Parametric tests have more power than nonparametric tests and make no assumptions on equality of the dispersion of your data. Parametric tests do assume normality, but the data set is sufficiently large that non-normality is no problem here (Brooks, 2014). Therefore, independent sampled t-tests are performed to test the equality of the mean between the IOA and EPA and also between industries. In addition, a variance ratio test is performed to check if the variances are equal or if there are differences in variances for industries. The variance ratio test hypothesizes that the ratio of two variances is equal to one against the alternative that it is not. Since the median is also an important statistic, a nonparametric median test is performed to see if the median is significantly larger for the EPA and if this result is industry-specific.
The CAP will be tested with the following model:
(5)
Where
is a lagged value of the CAP for company i and size is a dummy variable with a value of one for firms with a market capitalization of over 5 billion and zero for firms with a lower market capitalization. Return on invested capital and the scaled R&D and fixed assets of company i in year t are explanatory variables for some of the hypotheses. I multiplied with the industry median CAP represents the industry effects.
18
size and the value of the CAP is expected to be positively significant. Return on invested capital, R&D expenditures and fixed assets are mentioned in hypotheses six until eight and therefore their relationship with the CAP is also tested. As previously mentioned, R&D and fixed assets are scaled to percentages and return on invested capital is already in percentages. These variables are also trimmed at the 1%-99% level to eliminate outliers. The 1%-99% level is chosen here instead of 5%-95% because these variables have less available data. Finally, to approximate the hypothesized relationship between industries and the CAP, dummies are created that indicate one when a company is in a certain industry and zero otherwise. Following Fairfield et al. (2005) this dummy is multiplied by the industry median to create an interaction variable that shows to what extent industry membership influences the CAP. The industry median CAP fluctuates for each year and multiplied by the industry dummy it becomes a regressor in this model.
The enterprise method is followed in this paper to calculate the CAP instead of the equity method. The equity method approaches the CAP by purely looking at measures such as the cost of equity, return on equity and the book value of equity. The enterprise method can be more informative for this sample, because it takes differences and changes in leverage between firms into account. However, the equity method is also tested in a robustness check.
Since many observations are excluded from the sample, there are now gaps in the data for the regressions. This makes it difficult to run a reliable regression, so two solutions are considered. First, firms with less than ten years of available data in a row could be deleted. However, this leaves only a very small number of observations. The second solution is therefore chosen, which is applying data carry forwards through imputation. This implies that when no CAP is available for the year 1999, the CAP for that year will equal the CAP of the previous year. An advantage of this approach is that the data set has fewer gaps and is more balanced. A disadvantage is that it causes a bias, because stable CAPs are assumed over multiple years.
The sample shows some skewness and the level of kurtosis shows that the peak in the sample is more extreme than would be expected in a normal distribution. Therefore, there is some evidence of non-normality. However, because of the large sample size this non-normality can be ignored (Brooks, 2014). To test whether the panel data set should be regressed using a fixed effects or random effects regression, the Hausman test is performed. The result of this test gives a p-value lower than 0.01, which means fixed effects are appropriate. Next, a modified Wald-test for groupwise heteroskedasticity in a fixed effects regression model indicates that there is heteroskedasticity in the sample9.
19
Finally, the Wooldridge test for autocorrelation in panel-data models is applied and this shows that autocorrelation is present in the sample10. Therefore, the standard errors in the regression could be biased and the estimation approach should correct for these factors. Three methods were considered; Newey-West standard errors, panel corrected standard errors(PCSE) through Prais-Winsten regression and Driscoll-Kraay standard errors. Newey-West standard errors could lead to biased results, because this approach is meant for time series instead of panel data and it cannot take fixed effects into account. The PCSE might be a poor estimator, because the number of observations in the sample is large compared to the number of time periods. The PCSE estimates a full cross-sectional covariance matrix and when the ratio of time periods to observations is small the approach becomes imprecise. Therefore, the Driscoll-Kraay standard errors are chosen. This method appropriately takes into account heteroskedasticity, autocorrelation, fixed effects and cross-sectional dependence (Hoechle, 2007).
4. Analysis and discussion of results 4.1 Descriptive statistics
Table 5 presents the descriptive statistics for the CAP economy-wide and industry-specific. Panel A uses the investment opportunity approach and panel B the economic profit approach. There is a higher degree of non-normality in the sample for the IOA than for the EPA. Both skewness and kurtosis are around double the size for the IOA compared to the EPA, which means that the peaks in the distribution are higher and the distribution is skewed more to the right. The skewness can also be observed when the minimum and maximum values are observed. These values are much more dispersed for the IOA than for the EPA. As a result of this dispersion the standard deviation is also higher for the IOA than for the EPA.
Figure 1 shows the length of the CAP over the entire time period for both the IOA and the EPA. The most obvious difference between the IOA and the EPA is that the CAP is larger for all years when the EPA is used. The pattern of the length of the CAP shows some spikes, but overall the average CAP is quite stable. During the dotcom bubble an increase in the CAP is observed for the IOA and there is also a heavy spike between 2013 and 2014. Finally, a remarkable observation is that the CAP increased during the financial crisis from 2007-2011. For the EPA, there is a sharp decrease in the CAP between 2007 and 2008, which is in line with expectations during a crisis period. However, afterwards there is a sharp increase and between 2012 and 2014 there is another peak. Overall, Figure 1 does not provide convincing evidence that movements of the CAP are synchronized with economic cycles.
20
The independent sampled t-tests are performed for the IOA, EPA and all industries. The first test hypothesizes that the mean of the IOA is on average the same as the mean for the EPA against the alternative that the mean for the EPA is higher. The test result is highly significant, so the hypothesis is rejected and the assumption made in hypothesis two about larger CAPs for the EPA is supported. Another test that is performed to confirm hypothesis two is the non-parametric median test. The hypothesis is that the median of the IOA is on average the same as the median for the EPA against the alternative that the median for the EPA is higher. This result is also highly significant, so the median CAP for the EPA is higher than for the IOA. Therefore, the first part of hypothesis two is accepted.
The second part of hypothesis two is that the length of the CAP differs between industries. To test this independent sampled t-tests and median tests are performed for all industries, with the null hypothesis that the mean/median CAP of that industry is equal to the mean/median CAP of all other industries. Many industries show highly significant results, only telecommunications is insignificant in all cases. This means that for all other industries in the sample there are some deviations from the mean/median CAP. The levels of significance are quite strong for some industries, specifically information technology and healthcare.
0 5 10 15 20 25 30 35 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 Co m p e titi ve ad van tage p e ri o d Year
21
Table 5. Descriptive statistics competitive advantage period
Industry Obs Mean Median Std
Dev
Min Max Skew ness
Kurt osis
Panel A: Using the investment opportunity approach
Total sample 6,256 16.2*** 7.8*** 21.5*** 0.5 121.4 2.4 8.9 Consumer discretionary 1,284 17.2** 8.2 22.1* 0.5 120.3 2.2 7.9 Consumer staples 448 13.7*** 6.6** 19.6*** 0.5 120.2 2.8 11.4 Energy 335 14.4 7.2 20.1 0.5 114.2 2.6 9.9 Healthcare 932 14.7** 7.1*** 19.9*** 0.5 121.3 2.6 10.5 Industrials 1,539 15.1** 7.3** 20.6*** 0.5 119.8 2.5 9.8 Information Technology 1,251 18.6*** 8.9*** 23.6*** 0.5 121.4 2.1 7.4 Materials 429 16.5 9.3* 20.9 0.5 113.6 2.3 8.6 Telecommunic ations 38 16.5 9.3 20.9 0.5 113.6 2.3 8.6
Panel B: Using the economic profit approach
Total sample 12,363 23.4*** 19.8*** 15.0*** 3.6 76.1 1.2 4.1 Consumer discretionary 2,690 22.9** 19.5 14.4*** 3.6 75.6 1.2 4.4 Consumer staples 1,188 23.2 20.8** 13.6*** 3.8 76.0 1.1 4.3 Energy 746 18.7*** 14.2*** 14.6 3.7 75.2 1.5 5.0 Healthcare 1,618 22.0*** 18.0*** 15.1 3.6 76.1 1.3 4.4 Industrials 2,975 23.6 20.3* 14.1*** 3.7 75.4 1.1 4.2 Information Technology 2,061 27.1*** 22.5*** 17.0*** 3.6 76.1 0.9 3.1 Materials 1,001 22.4** 19.0 14.2** 3.7 75.6 1.2 4.4 Telecommunic ations 84 25.4 21.4 16.0 4.6 73.1 0.8 3.0
22
Information technology shows significantly higher CAPs than the other industries, meaning that the firms in the information technology industry are better able to create and sustain competitive advantages. At the moment information technology is a booming sector, with large firms in Silicon Valley competing for their competitive advantages. On the other hand healthcare shows significantly lower CAPs than the other industries. This might be due to increasing competition in the healthcare sector, which makes it necessary to compete on prices. In addition, there are less well-known brands in the healthcare industry, because healthcare is an industry in which relationships and trust are very important (Shore, 2006). Overall, based on these results there is enough proof to reject the hypothesis of equal means/medians for all industries and thus the second part of hypothesis two is also confirmed.
The third hypothesis is that the variance for the CAP is larger when the IOA is used than when the EPA is used. This hypothesis is tested with a variance ratio test. The hypothesis of this test is that the ratio of the variance of the IOA to the variance of the EPA is equal to one against the alternative that it is larger for the IOA. The highly significant result indicates that the variances indeed are not equal, but larger for the IOA. This implies that hypothesis three is correct. Another part of the third hypothesis is that the variance of the CAPs is different between industries. This is also tested with variance ratio tests. The results are insignificant for the telecommunications and energy sectors, but all other industries have significant values for at least one of the two approaches. For the telecommunications these insignificant results seem counterintuitive. Over the last few years the telecommunications industry has evolved very fast because of the developments in the mobile communication sector. It would make sense if the variance of CAPs were significant, because stronger competition might lead to faster reversion to the mean. Overall, the test results indicate that the variance of the CAP is not the same for all industries, but that there are significant differences. Therefore, the third hypothesis is accepted.
4.2 Regression results
23
constant in all models is also highly significant and shows the long-term CAP that the model will revert to because of competitive forces.
In the second model hypotheses four until eight are tested simultaneously. The significance of the lagged CAP is still the same, indicating that hypothesis four can be accepted under all conditions. The adjusted R-squared of the second model is lower than for the first model for both the IOA and the EPA. A lower adjusted R-squared when more variables are added to the model implies that extra variables do not have any additional explanatory power for the variance in the dependent variable (Brooks, 2014). Especially for the IOA the second model might suffer from this, because some insignificant variables are observed here. Three industries do not show significant values and R&D expenditures also do not significantly contribute to the model. For the EPA only one industry is insignificant, namely telecommunications. Telecommunications is also insignificant for the IOA and this insignificance might be due to the low number of observations that are available for this industry. Few observations in an industry might cause a low explanatory value of that industry in the model. Another variable that is insignificant for the EPA is fixed assets. The sign for fixed assets is negative, as hypothesized, but it is highly insignificant.
Therefore, a third model is tested where the variables for fixed assets and R&D expenditures are excluded from the regression. The exclusion of these variables improves the fit of the model and also causes the significance of variables and industries that were previously insignificant. Therefore, the model with R&D expenditures and fixed assets included might be overfit, meaning that the adding of R&D and fixed assets as explanatory variables only causes random noise instead of a higher explanatory power. This can cause other, significant, variables to become insignificant. Another explanation might be that these variables do not contribute linearly to the model due to accounting biases. This is in line with Koller et al. (2010) who believes it can often be too difficult to capture reality in a model.
In hypothesis five, firm size is expected to have a positive effect on the length of the CAP. Figure 2 illustrates the size of the CAP for firm size. Here it seems that larger firms have larger CAPs for both approaches. The same result can be observed in the regression analysis. In all models the sign of the size dummy is positive and highly significant, meaning hypothesis five can be confirmed. These results are in accordance with findings by Baumol (1967) and Mauboussin and Callahan (2013) that larger firms can benefit from economies of scale and less rivalry.
24
highly significant. So, without knowing the response of an increase in the ROIC on other variables, it does not have a positive effect on its own. Therefore, hypothesis six can be confirmed.
Hypotheses seven and eight are about the effects of R&D expenditures and fixed assets on the length of the CAP. These variables have been excluded from the regression analysis in the third model since this improved the fit of the model. Therefore, even though R&D showed some significance for the EPA and fixed assets for the IOA, we cannot be sure whether this is real significance or a random noise and therefore hypotheses seven and eight are rejected. For R&D expenditures the unreliability of the results might be due to data availability. R&D expenditures are not available for all observations and are equal to zero in many cases. However, many firms do not report their R&D expenditures separately in their financial statements and therefore the value becomes zero. R&D expenditures did show some significance for the EPA, but this significance had the wrong sign, so hypothesis seven cannot be accepted.
Finally, strongly significant results can be observed for all industries in the case of the EPA and all industries except for telecommunications in the case of the IOA. The signs are all positive, which means that the industry-specific effects of a particular industry have a positive effect on the length of the CAP in that industry. The importance of industry-specific effects is therefore once more confirmed in addition to the performed t-tests and variance tests. Therefore, this paper joins authors such as Porter (1980) and Hawawini et al. (2003) in their belief that industry-specific factors are an important explanatory factor for profitability.
0 5 10 15 20 25 30
Small firm Large firm
Co m p e titi ve ad van tage p e ri o d Firm Size
Figure 2. The competitive advantage period for firm size
Investment opportunity approach
25
Table 6. Regression analysis for the competitive advantage period
Panel A: Investment opportunity
approach Panel B: Economic profit approach
Variables Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 CAP(t-1) 0.518*** (13.23) 0.498*** (13.27) 0.517*** (13.11) 0.469*** (15.73) 0.443*** (15.12) 0.462*** (16.71) Return on invested capital -2.600*** (-8.06) -2.323*** (-7.66) -4.790*** (-8.47) -5.125*** (-8.52) R&D 4.648 (1.06) 5.275** (2.21) Fixed Assets -3.556** (-1.98) -1.962 (-0.79) Constant 7.625*** (14.80) 6.061*** (10.10) 4.757*** (8.69) 12.317*** (22.31) 4.639*** (5.10) 3.690*** (5.68) Dummy Size dummy 2.002** (2.29) 1.513** (2.78) 2.860*** (6.13) 2.743*** (14.71) Industry Consumer discretionary 0.265 (1.33) 0.340*** (3.45) 0.560*** (7.04) 0.559*** (9.94) Consumer staples 0.298*** (3.22) 0.420*** (7.49) 0.629*** (7.22) 0.517*** (6.95) Energy 0.427** (2.02) 0.337*** (4.07) 0.728*** (4.36) 0.382*** (4.97) Healthcare 0.085 (0.82) 0.182** (2.27) 0.431*** (3.09) 0.416*** (3.65) Information technology 0.373*** (7.88) 0.344*** (7.09) 0.420*** (7.84) 0.439*** (10.09) Industrials 0.604*** (3.99) 0.538*** (4.92) 0.557*** (8.37) 0.510*** (8.21) Materials 0.269** (2.49) 0.376*** (5.33) 0.475*** (4.55) 0.464*** (5.25) Telecommun ications 0.041 (0.53) 0.051 (1.46) -0.004 (-0.11) 0.068** (2.33) Tests and R-squared Wooldridge test 0.000 0.000 0.000 0.000 0.000 0.000 Wald test 0.000 0.000 0.000 0.000 0.000 0.000 Adjusted R-squared 0.529 0.517 0.534 0.513 0.509 0.531 Within R-squared 0.265 0.257 0.271 0.222 0.244 0.261
Note: *** p<0.01, ** p<0.05, * p<0.1. T-statistics are given in parentheses.
26
4.3 Robustness checksTo test the robustness of the results, three robustness checks are performed. The results of these checks can be found in Appendix C until E. The first robustness check applies a constant risk free rate and constant market risk premium to see if this changes the results. The market risk premium is set at 5% based on calculations by Koller et al. (2010) and the risk free rate is the average of the ten year US treasury rate over the entire sample period. The second robustness check uses the equity method instead of the enterprise method. The last robustness check uses yearly fluctuating industry medians for the beta, credit spread, effective tax rate and debt-to-equity ratio instead of a constant industry median over the entire sample period.
The values of the coefficients change slightly when a constant risk free rate and market risk premium are applied, but overall the signs and the significances of the variables stay the same. The results of the t-tests, variance tests and median tests are also similar. Therefore, the results are robust for a constant risk free rate and market risk premium.
The results for the CAP are also approximately robust when the equity method is applied. However, these results do show a larger dependence on the return measure, because it has a much larger coefficient. This might be because of the difference in measurement between return on invested capital and return on equity. The coefficient for return on equity is very large, which might be at the expense of some other explanatory variables that now show a slightly lower level of significance. For the EPA the levels of significance are very similar to the original model, but for the IOA more insignificance for industries is observed.
Finally, the results for the CAP with yearly fluctuating industry medians are also robust. Especially for the EPA the results are identical in terms of signs and level of significance. This behavior could also be observed in the other two checks. For the EPA the results were more similar to each other than for the IOA. This again shows that the IOA is a more fluctuating and volatile measure than the EPA, because it responds more to changes in any of the inputs of the model.
5. Conclusions and recommendations 5.1 Conclusions
1996-27
2015. The enterprise method is used and a panel data model with Driscoll-Kraay standard errors and fixed effects is applied.
This paper shows that the EPA leads to a positive CAP in much more cases than for the IOA. Negative values for the CAP are deemed non-plausible, so the final sample for the EPA is larger than for the IOA. In addition, the EPA gives larger values for the CAP and less volatility. Especially the lower volatility makes the EPA more trustworthy than the IOA, because then CAPs show more stability. The regression results also give better results for the EPA than for the IOA in the sense that the significance levels are higher when the EPA is used. Finally, in the robustness checks the volatility of the IOA also becomes clear. Most results for the EPA are robust to changes in the measurement of the CAP, while the IOA shows some differences in levels of significance for each method. Based on these findings the EPA is a better measure of the CAP than the IOA.
This paper also has some more results on the CAP in general. The importance of industry-specific effects in the size of the CAP is shown both in the regression results and in the parametric and non-parametric tests that were performed. There are differences in the length of the CAP and the variance of the CAP among industries. In addition industry-specific effects are also highly significant and positive in the regressions of this paper, which means industry-specific effects contribute positively to the size of the CAP in that industry. Firm size is also an important determinant, because a larger firm size leads to a larger CAP. R&D expenditures and fixed assets do not show sufficient evidence of linear significance to the model. On the other hand, there is clear evidence that return on invested capital influences the length of the CAP. Finally, a model is employed to determine the stability of the CAP over time. The lagged CAP shows that the assumption of stable CAPs is not valid and that CAPs are mean reverting.
5.2 Limitations and recommendations
28
29
ReferencesBarney, J., 1991. Firm resources and sustained competitive advantage. Journal of Management, Vol 17 Issue 1, 99-120.
Baumol, W.J., 1967. Business behavior, value and growth. Harcourt Brace Jovanovich, New York.
Berk, J.B., De Marzo, P. and Harford, J. Fundamentals of corporate finance. Pearson, 3rd global edition.
Bou, J. and Satorra, A., 2007. The persistence of abnormal returns at industry and firm levels: Evidence from Spain. Strategic Management Journal, Vol 28 Issue 7, 707-722. Brooks, C., 2014. Introductory econometrics for finance. Cambridge University Press, 3rd edition.
Damodaran, A., 2007. Return on capital (ROC), return on invested capital (ROIC) and return on equity (ROE): measurement and implications. Stern School of Business, 1-69. Danbolt, J., Hirst, I. and Jones, E., 2002. Measuring growth opportunities. Applied Financial Economics, Vol 12 Issue 3, 203-212.
Danbolt, J., Hirst, I. and Jones, E., 2011. The growth companies puzzle: can growth opportunities measures predict firm growth? The European Journal of Finance. Vol 17 Issue 1, 1-25.
Danielson, M.G. and Dowdell, TD., 2001. The Return-stages valuation model and the expectations within a firm’s P/B and P/E Ratios. Journal of Financial Management, Vol 30 Issue 2, 93-124.
Dickinson, V. and Sommers, G.A., 2012. Which competitive efforts lead to future abnormal economic rents? Using accounting ratios to assess competitive advantage. Journal of Business Finance and Accounting. Vol 39, Issue 3, 360-398.
Fairfield, P.M., Sundaresh, R. and Yohn, T.L., 2005. Does industry-level analysis improve profitability and growth forecasts? Journal of Accounting Research, Vol 47 Issue 1, 147-178.
30
Fama, E.F. and French, K., 2000. Forecasting profitability and earnings. The Journal of Business. Volume 73, Issue 2, 161-175.
Grant, R.M., 1991. The resource-based theory of competitive advantage: implications for strategy formulation. California Management Review, Vol 33 Issue 3, 114-135.
Hawawini, G., Verdin, P. and Subramanian, V., 2003. Is performance driven by industry- or firm-specific factors? A new look at the evidence. Strategic Management Journal, Vol 24 Issue 1, 1-16.
Hoechle, D., 2007. Robust standard errors for panel regressions with cross-sectional dependence. The Stata Journal. Vol 7, Issue 3, 281-312.
Koller, T., Goedhart, M. and Wessels, D., 2010. Valuation, measuring and managing the value of companies, fifth edition. John Wiley and Sons Inc., New York.
McGahan, A. and Porter, M.E., 1997. How much does industry matter, really? Strategic Management Journal, Vol 18, 15-30.
Mauboussin, M. and Callahan, D., 2013. Measuring the moat. Assessing the magnitude and sustainability of value creation. Credit Suisse Global Financial Strategies.
Mauboussin, M. and Johnson, P., 1997. Competitive advantage period: The neglected value driver. The Journal of Financial Management Association, Vol. 26 Issue 2, 67 – 78. Miller, M.H. and Modigliani, F., 1961. Dividend policy, growth and the valuation of shares. The Journal of Business, Vol 34 Issue 4, 411-433.
Mouelhi, C. and Saint-Pierre, J., 2008. The competitive advantage period (CAP) as a basis for portfolio selection. Unpublished working paper. Université Laval, Québec.
Peteraf, M.A., 1993. The cornerstone of competitive advantage: a resource-based view. Strategic Management Journal, Vol 14 Issue 3, 179-191.
Porter, M.E., 1980. Competitive strategy. The Free Press, New York.
Shore, D.A., 2006. The trust crisis in healthcare: causes, consequences and cures. Oxford University Press, New York.
Stigler, G.J., 1961. The economics of information. The Journal of Political Economy. Vol 69, Issue 3, 213-225.
31
Appendix A: Derivation of the competitive advantage period Derivation of the competitive advantage period for the IOA: Value according to the investment opportunity approach equals:
∑
(A.1)
With
for
and
for
,
value can be written as:∑
∑
(
)
(A.2)
The power term can be written as
or as
.
Using the first approximation gives for value:
(A.3)
and for the competitive advantage period:
(A.4) To get rid of the t+1’s we can rewrite the investment opportunity approach as:
∑
(A.5)
With
,
value can be written as:∑
=
=
(A.6)
32
(A.7)
and for the competitive advantage period:
(A.8)
Equation (8) together with provides an estimate for the competitive advantage period using the investment opportunity approach.
Derivation of the competitive advantage period for the EPA: Value according to the economic profit approach equals:
∑
(A.9)
With
, value can be written as:
∑
=
∑
=
( )
=
(A.10)
Approximating the power term gives for value:
(A.11)
and for the competitive advantage period:
(A.12)
33
Appendix B: Excluded observations and distribution of final sample
Table B.1. Percentage of excluded observations per industry Industry Initial sample
Trimming 5%-95% Final sample Panel A: Investment opportunity approach
Total sample 100% 41.9% 16.7% 10.9% 3.4% 2.7% 24.5% Consumer discretionary 20.2% 19.9% 20.7% 20.0% 20.3% 20.8% 20.5% Consumer staples 6.8% 4.1% 9.9% 11.3% 8.4% 8.0% 7.2% Energy 7.3% 8.6% 3.8% 13.1% 7.7% 4.2% 5.4% Healthcare 15.6% 15.1% 17.6% 15.5% 18.0% 15.5% 14.9% Industrials 21.4% 19.9% 20.2% 22.2% 21.0% 22.3% 24.6% Information Technology 19.7% 23.0% 18.5% 7.8% 16.6% 22.3% 20.0% Materials 7.5% 7.2% 8.1% 9.3% 7.3% 6.3% 6.9% Telecommun ications 1.4% 2.2% 1.2% 0.9% 0.7% 0.6% 0.6%
Panel B: Economic profit approach
Industry Initial sample
Trimming 5%-95% Final sample Total sample 100% 41.9% 2.8% 1.6% 5.4% 48.4% Consumer discretionary 20.2% 19.9% 4.8% 18.6% 17.8% 21.8% Consumer staples 6.8% 4.1% 0.4% 10.9% 5.5% 9.6% Energy 7.3% 8.6% 0.7% 16.7% 9.8% 6.0% Healthcare 15.6% 15.1% 72.7% 9.0% 15.0% 13.1% Industrials 21.4% 19.9% 5.2% 26.2% 17.0% 24.1% Information Technology 19.7% 23.0% 12.7% 9.4% 26.3% 16.7% Materials 7.5% 7.2% 1.8% 7.7% 6.9% 8.1% Telecommun ications 1.4% 2.2% 1.6% 1.5% 1.7% 0.7%
34
Appendix C: Robustness check constant risk free rate and market risk premium Table C.1. Descriptive statistics competitive advantage period
Industry Obs Mean Median Std
Dev
Min Max Skew ness
Kurt osis
Panel A: Using the investment opportunity approach
Total sample 6,485 19.3*** 9.4*** 25.4*** 0.5 143.7 2.3 8.6 Consumer discretionary 1,343 20.6** 9.9 26.0 0.5 140.6 2.1 7.7 Consumer staples 504 15.7*** 7.9*** 21.9*** 0.5 140.8 2.8 12.4 Energy 340 16.5** 8.2* 22.7*** 0.6 131.8 2.7 10.7 Healthcare 984 17.3*** 8.1** 23.8*** 0.5 141.7 2.5 9.8 Industrials 1,595 18.0** 9.2 24.4*** 0.5 137.7 2.5 9.7 Information Technology 1,244 22.8*** 10.7*** 28.8*** 0.5 143.7 2.0 6.8 Materials 438 20.6 10.8* 25.3 0.6 140.5 2.1 7.2 Telecommunic ations 37 21.2 14.4 21.6 1.1 89.7 1.3 4.2
Panel B: Using the economic profit approach
Total sample 11,825 25.0*** 20.8*** 16.6*** 3.7 83.6 1.3 4.3 Consumer discretionary 2,563 24.7 20.5 16.3 3.8 83.1 1.3 4.6 Consumer staples 1,165 24.5 22.0** 14.5*** 3.8 82.1 1.2 4.6 Energy 706 19.7*** 14.5*** 15.4*** 3.7 81.0 1.5 4.9 Healthcare 1,582 23.5*** 19.0*** 16.5 3.8 83.5 1.4 4.6 Industrials 2,833 25.3 21.3** 15.9*** 3.7 83.4 1.3 4.5 Information Technology 1,959 28.8*** 23.6*** 18.7*** 3.8 83.6 1.0 3.3 Materials 937 23.9** 20.0 16.0 3.7 83.5 1.3 4.7 Telecommunic ations 80 29.0* 24.1 19.9** 4.8 78.5 1.0 3.2
35
Table C.2. Regression analysis for the competitive advantage period
Panel A: Investment opportunity
approach Panel B: Economic profit approach
Variables Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 CAP(t-1) 0.504*** (13.86) 0.492*** (14.33) 0.501*** (13.80) 0.483*** (14.59) 0.451*** (13.22) 0.473*** (15.02) Return on invested capital -3.846*** (-16.08) -3.299*** (-12.79) -5.043*** (-9.22) -5.479*** (-9.18) R&D 0.264 (0.05) 7.179*** (3.19) Fixed Assets 2.021 (0.77) -1.236 (-0.50) Constant 9.396*** (15.41) 6.754*** (17.52) 6.913*** (10.63) 12.819*** (23.23) 4.846*** (4.14) 3.925*** (6.21) Dummy Size dummy 5.312*** (10.05) 3.266*** (6.89) 3.120*** (4.96) 2.534*** (8.26) Industry Consumer discretionary 0.087 (0.59) 0.169 (1.19) 0.645*** (7.75) 0.621*** (6.81) Consumer staples 0.173 (1.49) 0.225 (1.37) 0.666*** (5.29) 0.527*** (7.91) Energy 0.877*** (2.62) 0.478*** (3.74) 0.613*** (3.10) 0.363*** (3.80) Healthcare 0.055 (0.42) 0.150 (1.05) 0.337** (2.47) 0.336** (2.73) Information technology 0.365*** (5.84) 0.342*** (5.73) 0.452*** (6.43) 0.497*** (6.77) Industrials 0.462*** (4.50) 0.468*** (3.82) 0.608*** (7.98) 0.568*** (13.66) Materials 0.331** (2.55) 0.380** (2.45) 0.363*** (3.69) 0.375*** (4.12) Telecommun ications 0.059 (0.95) 0.013 (0.42) 0.045 (0.98) 0.150*** (3.27) Tests and R-squared Wooldridge test 0.000 0.000 0.000 0.000 0.000 0.000 Wald test 0.000 0.000 0.000 0.000 0.000 0.000 Adjusted R-squared 0.515 0.517 0.519 0.519 0.515 0.534 Within R-squared 0.248 0.251 0.254 0.233 0.247 0.268
Note: *** p<0.01, ** p<0.05, * p<0.1. T-statistics are given in parentheses.
