- 1 -
The influence of brand value on stock return, equity-holder
risk and trading activity:
a comparison between Interbrand, Millward Brown and BrandFinance.
R. van der Meer
1September 2012
Supervisors:
Dr. V. Angelini
Drs. M.M. Kramer
University of Groningen
Faculty of Economics and Business
MSc Business Administration
Specialization:
Finance
1
R. van der Meer, student number: 1536303, e-mail: rob_vd_meer@hotmail.com, University of Groningen, Faculty of Economics and Business, MSc Business Administration, Specialization:
- 2 -
Abstract
This master thesis investigates the relationship between the change in brand value and the change in stock return, the change in equity-holder risk and the change in trading volume. Relative brand values, based on brand value estimates over the sample period 2006-2011 from three different valuation companies are used: Interbrand, Millward Brown and BrandFinance. This research design enables to check whether these different estimated brand values are consistent with each other and have similar impact. Millward Brown and BrandFinance brand value estimate increases led to decreasing stock returns, whereas a positive relationship was expected. For Interbrand brand value estimates, no result was found. Differentiating for positive and negative brand value changes, results in increasing stock returns for decreasing brand values from Interbrand and Millward Brown. Positive brand value changes have no significant effect in those cases. Positive BrandFinance brand value changes result in decreasing stock returns, whereas decreasing brand values have no impact. An event study showed for none of the brand valuation companies an immediate effect on stock prices at the announcement date of the new brand values. Brand value changes from Millward Brown result in lower equity risk: lower stock volatility. Millward Brown brand value estimates outperform the other two brand valuations in this perspective, since Interbrand and BrandFinance brand value estimates show no significant result. No statistical inferences could be drawn on the effect of brand value on trading volume for Interbrand and Millward Brown brand value estimates. However, increasing brand value estimates of BrandFinance result in decreasing trading volume.
JEL classification: C23, G12, G14, M31
Key-words: brand value, stock return, equity-holder risk, trading volume, panel data, event
- 3 -
1. Introduction
The presence of very large global brands is overwhelming. Even during a hike in the jungle of Thailand, a can of Coca Cola will be available during your overnight stay with a native tribal. Back to the civilized world, still miles away from home, your driver takes you in his Toyota van to a local market. Walking around, you smoke a Marlboro cigarette before you buy a (fake) Louis Vuitton bag. Now it’s time for coffee, let’s go to Starbucks so you can access Google via the WIFI on your Apple IPhone.
The above mentioned brand names are just an example of exposure to brands we face every day. According to Phillips et al. (2003): “It is estimated that each American is exposed to well
over 2,500 advertising messages per day, and that children see over 50,000 TV-commercials a year.”
During the 30th Olympic Games, recently held in London, there was a large TV-commercial by Coca-Cola. The title of the song supporting this campaign was: “Anywhere in the world”. In my opinion, this one sentence could not be more striking for a brand like Coca Cola (as for the Olympics). Nowadays, there are some brands that are worldwide known by many people. This is something that the seven mentioned brands above have in common. However, these seven brands have one more thing in common: they outshine others, being in the top 100 most valuable Global brands. This list is yearly publicized by Interbrand, a branding consultancy company that puts a price tag on brand value. In cooperation with BusinessWeek the top 100 most valuable Global brands is launched yearly. However, Interbrand is not the only company that does so.
- 4 - The paper of Salinas and Ambler (2009) is part of the special issue on Brand Value and Valuation of the Journal of Brand Management in 2009. The special issue starts with a guest editorial by Raggio and Leone (2009) introducing and underlining the importance of brand value and brand valuation. Raggio and Leone (2009) state: “Brands constitute the largest
asset for many firms, and brand valuations are increasingly being seen as an important performance metric for both companies and managers.” Since brand valuations are remarked
as an important performance metric, increasing brand values should ultimately result in increased firm performance. In addition, Raggio and Leone (2009) emphasize the fact that financial market performance is positively influenced by components of brand valuation models.
In general, marketing performance metrics should be linked to company performance measures in this era of marketing accountability, where marketing will be held accountable for economic outcomes and financial results (Stewart, 2008). Methods to measure return on marketing investment (Seggie et al., 2007) and marketing metrics that maximize profitability are under discussion (Petersen et al., 2009). Brand value is one of these metrics that is linked to firm performance and has numerous examples of studies in current marketing literature (Kerin and Sethuraman, 1998; Madden et al., 2006; De Beijer et al., 2008). This thesis explores the brand value and firm performance relationship and, in doing so, the difference between three different valuation methods is taken into account. This thesis makes a comparison, both in direction and magnitude, of the brand value – firm value relationship for brand value estimates from Interbrand, Millward Brown and BrandFinance.
Firm performance in terms of shareholder value has two aspects that are in classical finance literature always mentioned together; return is always associated with risk. For example, the capital asset pricing model (CAPM) relates risk to expected returns (Sharpe and Lintner, 1964, 1965). As mentioned above, there is growing evidence linking brands and firm stock returns, however little is known about the effects of brands on firm risk (Rego et al., 2009). Rego et al. (2009) show in their paper the “risk relevance” of consumer-based brand equity. Therefore, not only the relationship between brand value and stock return is researched in this paper, also the relationship between brand value and equity risk is explored.
- 5 - Based on some behavioral finance principles, this thesis explores the difference in effect between positive and negative brand value changes. In addition, the influence of brand value on trading volume is tested. It is of interest to test whether there is a linkage between brand value and trading volume since people tend to invest in familiar stocks. Moreover, research findings by Frieder and Subrahmanyam (2005) suggest that investors prefer visible brand name stocks in their portfolio.
A better understanding of the relationship between brand value and stock returns, equity risk and trading volume can provide an insight for both finance and marketing managers. This statement is strengthened by the notion of Keller and Lehman (2006) who remark branding as a top management priority and emphasize on the fact that brands are one of the most valuable intangible assets that firms have. Moreover, the impact of brand value also enables investors and stakeholders to make better-informed decisions.
This thesis contributes to the current literature by simultaneously linking brand values to stock returns, equity risk and trading volume for the period 2006-2011. The majority of academic research in this field contains data before 2006. Moreover, multiple brand valuation methods will be used for brand value estimates. This thesis explores the differences in the brand value –shareholder value relationship for three different valuation companies; Interbrand, Millward Brown and BrandFinance. Whereas prior studies consisted of cross-sectional analyses exploring the brand value – stock market relationship, this study uses a panel approach for the longitudinal data.
- 6 -
2. Literature
Traditionally, finance researchers are interested in the impact of firm strategies and decisions on investor expectations. Research focusses on the creation of shareholder value and shareholders constitute the central stakeholder group. In contrast, marketing researchers focus on customer reactions to marketing strategies and decisions. Research focusses on attitudes and behaviors that drive revenues in the marketplace, with consumers being the key constituency. Aligning marketing and finance metrics, bridges the gap between marketing and finance (Madden et al., 2006).
A growing literature in finance addresses determinants of investor holdings. Frieder and Subrahmanyam (2005) explore the impact of brand perceptions of companies’ products spill over to investment decisions in the market for companies’ stocks. For example, Black & Deckers’ portfolio of well-recognized brands was remarked by Standard & Poor to be one of their key competitive strengths and even resulted in a purchase recommendation of Black & Deckers’ stock (Frieder and Subrahmanyam, 2005). Furthermore, intangible assets, including brands, are given as an explanation for discrepancies between share prices for companies and their tangible assets. These discrepancies faced by analysts, even resulted in proponents and opponents of including brands on balance sheets (Salinas and Ambler, 2009).
In recent years, there has been a great focus on accountability within the marketing discipline. According to Stewart (2008), the frequency of academics publishing on this topic has been increasing. Examples of publications are Rust et al. (2004) with an overview on how to measure marketing productivity, Seggie et al. (2007) discuss methods to measure return on marketing investment and Petersen et al. (2009) discuss what metrics to choose in order to maximize profitability. Stewart (2008) states that marketing will be held accountable and that accountability is ultimately about economic outcomes and financial results. The inability for marketing practitioners to account for marketing contribution to the firm has undermined its standing within the firm (O’Sullivan and Abela, 2007). Verhoef and Leeflang (2009) found accountability and innovation within the marketing department to be the two major drivers for its influence within the firm.
- 7 - marketing, firm- and shareholder value (Rust et al., 2004, Lehmann, 2004, Frieder and Subrahmanyam, 2005, Petersen et al., 2009). This thesis deals with the relationship between brand equity and stock returns, the relationship between brand equity and equity risk and the relationship between brand equity and trading volume.
Brand equity is a popular topic among researchers (Leone et al., 2006) and some even concluded that brands are one of the most valuable assets that companies have (Aaker, 1991). Frieder and Subrahmanyam (2005) state: “An important concept in marketing is brand value
or brand equity”. The importance of brand equity is also underpinned in research by Bick
(2009). In a simultaneous assessment of the importance of brand equity and customer equity, both constructs are mentioned as clear drivers for shareholder value. Moreover, Keller and Lehman (2006) discuss in their paper influential work in the branding area. In their paper, they highlight what has been learned from an academic perspective on several important topics, including brand positioning, brand integration, brand-equity measurement, brand growth and brand management. The opening quote of Keller and Lehman (2006) emphasizes on the relevance and importance of brands: “Branding has emerged as a top management
priority in the last decade due to the growing realization that brands are one of the most valuable intangible assets that firms have.”
Brand Equity
After pointing out the importance of brands and the linkage between brand equity and shareholder value, this section continues with some definitions. The term brand equity emerged in the 1980’s and is defined by The Marketing Science Institute as follows: “Brand
Equity is the set of associations and behavior on the part of the brand’s consumers, channel members, and parent corporation that permits the brand to earn greater volume or greater margins than it would without the brand name and that gives the brand a strong, sustainable and differentiated advantage over competitors (Leuthesser, 1988).” Both in accounting
- 8 - Feldwick (1996) divides the variety of approaches into a classification of three different meanings of brand equity:
- the total value of a brand as a separable asset – when it is sold or included on a balance sheet;
- a measure of the strength of consumers’ attachment to a brand;
- a description of the associations and beliefs the consumer has about the brand.
The first meaning can be subscribed to the financial approach of brand equity, whereas the latter two fit into the customer based brand equity spectrum. Both customer based brand equity and financial based brand equity will be discussed below.
Customer Based Brand Equity
In the marketing literature, intangible brand properties became to be known as brand equity (Kerin and Sethuram, 1998). Aaker (1991) describes brand equity as the value consumers associate with a brand. This brand equity is built on four dimensions: brand awareness, brand loyalty, perceived brand quality and favorable brand symbolism and associations. The brand equity ten gives, grouped into five categories, ten measures of brand equity: price premium, loyalty, perceived quality, leadership, perceived value, brand personality, organizational associations, brand awareness, market share and price & distribution indices (Aaker, 1996). Keller (1993) gives a different definition of customer based brand equity: “Customer-based
brand equity is defined as the differential effect of brand knowledge on consumer response to the marketing of the brand.” In addition, by developing a scale to measure customer based
brand equity, Lassar et al. (1995) give five dimensions of customer based brand equity: performance, value, social image, trustworthiness and commitment. Feldwick (1996), as mentioned above, gives two different expressions of customer brand equity. The concept of measuring consumer’s level of attachment to a brand can be called ‘Brand Loyalty’. Feldwick (1996) uses his preferred Brand Strength for this sense in his paper. The concept of a description of the associations and beliefs the consumer has about the brand could be called ‘Brand Image’, according to Feldwick (1996), but in his paper the term Brand Description is used.
Financial Based Brand Equity
- 9 - could bring in the financial market. Simon and Sullivan (1993) define brand equity, in their financial approach of brand equity, as follows: “the incremental cash flows which accrue to
branded products over unbranded products.” Using brand related profits based on financial
market estimates, Simon and Sullivan (1993) present a new technique for estimating a firm’s brand equity. Feldwick (1996) states that brand equity is used in three quite distinct senses. One of the three expressions of brand equity is referred to as brand value: the total value of a brand as a separable asset. Feldwick (1996) refers to this financial approach of brand equity as brand value.
Brand Equity and Firm Value
In recent years, there has been done quite some research to the effects of brand equity (or according to the financial approach of brand equity: brand value) on firm and shareholder value. Barth et al. (1998) uses Financial World’s brand value estimates to test the relationship of those values with share prices of the particular firms owning the brands. In their paper, they showed that brands significantly affect stock prices and stock returns. On the contrary, no significant positive relationship between brand value estimates and sales growth was found (Barth et al., 1998). Kerin and Sethuram (1998) use Financial World’s brand value estimates to test the relationship between brand value and shareholder value, taking market-to-book ratios as a measure for shareholder value. They found a positive relationship between brand value and shareholder value, albeit this relationship is concave with decreasing returns to scale.
Earlier, Aaker and Jacobson (1994) found high brand equity levels to lead to higher stock returns. They found that stock return is positively related to changes in return on investment (ROI) and that, albeit not quite as large as the response to ROI, stock returns and changes in brand equity held the same relationship.
In a simultaneous assessment of the effect of brand value and advertising on firm performance, Eng and Keh (2007) found improved future accounting returns at the firm level. Spending on advertising results in higher brand sales and brand profitability, whereas brand value is found to be a good predictor of brand performance. Conversely, a minimal impact of advertising and brand value on future stock returns was found (Eng and Keh, 2007).
- 10 - ratios and stock market performance measures is proved. Additionally, their findings provide sufficient evidence to suggest that there is a significant relationship between brand equity and the performance of the brand owner (the firm) in the stock market. However, while comparing the returns model and the price model, a significant impact of brand value on share prices is found, whereas there seems to be no impact on market returns (Yeung and Ramasamy, 2008).
Pahud de Mortanges and van Riel (2003) test the relationship between brand equity (brand value) and shareholder value for Dutch corporate brands. They measure brand equity using the Brand Asset Valuater® model (Young and Rubicam, 2000) for the years 1993 and 1997. Shareholder value is measured by three different indicators: total shareholder return, earnings per share and the market-to-book ratio. Pahud de Mortanges and van Riel (2003) conclude that the performance of a brand, measured by changes in brand equity, may have significant impact on the value of a firm.
According to de Beijer et al. (2008), new brand value announcements should represent new information to the market. If stock markets are semi-strong form efficient, the stock market immediately and fully incorporates all value-relevant new information (Fama, 1970). Moreover, de Beijer et al. (2008) found this effect for Interbrand brand values. They claim to be the first to provide direct evidence on the magnitude of the impact of externally provided brand value announcements on firm value. Using announcements of the valuation method of Interbrand, they found stock prices reflecting the new information.
Brand Value and Risk
Madden et al. (2006) use the Fama-French model and Interbrand’s measure of brand value to provide empirical evidence for the branding-shareholder value creation link. Besides showing that strong brands deliver greater returns to stockholders than a relevant benchmark does, they also show that those strong brands deliver those greater returns with less risk. A comparison is made between the World’s most valued brands (WMVB) portfolio and two benchmark portfolios. The market performance of the WMVB portfolio is firstly compared to a reduced market portfolio, containing all firms in the Center for Research Prices (CRSP) database except those in the WMVB portfolio and secondly to a full market portfolio that contains all firms in the Center for Research Prices database without exception. The WMVB portfolio both outperformed the reduced-market portfolio and the full-market portfolio. Moreover, the results hold, controlling for market size and firm value (Madden et al., 2006).
- 11 - average CBBE of a firm’s brands have a robust impact in reducing debt-holder risk. They state that this reduced debt-holder risk directly contributes to lowering the cost of capital. Furthermore, they find that firms with strong CBBE are also able to significantly reduce their equity risk. In their paper, they also make a difference between systematic and unsystematic equity risk. Since, different from Madden et al. (2006), they asses single firms and do not use a portfolio approach.
Measurement of Brand Value
Several researchers in the field of linkage between brand equity and shareholder value have proposed different methods of measuring brand equity (Srinivasan et al, 2005). Srinivasan et al. (2005) list eight different methods in their paper and eventually come up with another, new method to measure brand equity. Availability of good brand equity measures is a key requirement for managing brands (Aaker and Joachimsthaler, 2000). In a special issue on brand value and brand valuation in the Journal of Brand Management, Raggio and Leone (2009) discuss the latest research and ideas among others on valuation methodologies and use of brand valuation in practice. One of the five papers mentioned and discussed by Raggio and Leone (2009) is the paper of Salinas and Ambler (2009): “Salinas and Ambler provide a
valuable service by identifying the separate types of methodologies currently used in practice, consolidating those that differ only by label, and distinguishing them from those methodologies that are only theoretical, or appear only in academic journals.”
The development of brand valuation methods has been driven by at least four factors: measuring marketing performance, justifying share prices, trading brands and tax management. Indirectly this thesis focuses on measuring marketing performance, which should be linked to firm performance. Explicitly this thesis focuses on the justifying share prices aspect of brand valuation methods. The different valuation methods are divided into three broader categories of brand valuation approaches (Salinas and Ambler, 2009):
- cost; value based on the historical cost of the creation of a brand; - market; value based on the trade of a brand;
- income; discounted future cash flows attributable to the brand.
- 12 - Millward Brown and Brand Finance are chosen, and this is based on their differences in methodology and differences in values for the same brands in the same years.
Within the commercial brand valuation industry, Interbrand can claim to be the pioneer; Millward Brown’s BrandZ is like Interbrand also evaluated in the comparison of Salinas and Ambler (2009) and under discussion in the work of Harish (2009). Besides Interbrand and Millward Brown, Harish (2009) also takes into account BrandFinance’s method. This method is different from Millward Brown’s BrandZ and Interbrand’s method. BrandFinance’s method is the third valuation method used in this paper. Brand value estimates from all three valuation companies are publicly available.
Figure 1: 2011 Brand values for Coca-Cola and Apple, estimated by Interbrand, BrandFinance and Millward Brown
Figure 1 gives the brand values of Coca-Cola and Apple for all three brand valuation methods taken into account in this research. In line with the statement of Salinas and Ambler (2009), the estimated brand values for Coca-Cola and Apple differ across the different brand valuation companies. Moreover, at first sight there are no two measures giving approximately the same values across different brands. Whereas the value of Coca-Cola is comparable for Interbrand (71.861 $m) and Millward Brown (73.752 $m), the value estimated by BrandFinance is only a little more than one third of this amount (25.807 $m). In contradiction the value of Apple is comparable for Interbrand (33.492 $m) and BrandFinance (29.543 $m), but much higher for Millward Brown (153.25 $m).
In an overview on current academic literature on marketing and firm value, Srinivasan and Hanssens (2009) show metrics, methods, and findings. Moreover, Srinivasan and Hanssens list 10 different directions for future research. Among others they ask for comparing different measures of brand equity: “We know that investors react to movements in brand value, but
$0 $20 $40 $60 $80 $100 $120 $140 $160 $180
Coca Cola Apple
- 13 -
are these brand metrics reliable and consistent with each other?” (Srinivasan and Hanssens,
2009).
Different brand consultancy companies calculate brand values on a yearly basis (Harish, 2009). Interbrand, Millward Brown (BrandZ) and BrandFinance all use brand valuation methodologies in order to construct brand values. All companies use the income-based approach to brand valuation. However, the economic use method is used by Interbrand and Millward brown, while BrandFinance adopted the royalty relief method. The former method is based on the economic value of the brand in its current use to current owner, whereas the latter method is based on the premise that if a brand is to be licensed from another party, a royalty would have to be paid on the turnover, for using the brand.
An important difference between the different methods arises with respect to transparency. Interbrand’s method is largely dependent on available published data, BrandFinance uses consumer survey data to a significant extent, while Millward Brown’s methodology is for the most part driven by consumer research data. This implies that Interbrand’s method is most transparent and Millward Brown’s method is least transparent.
One could argue that the more transparent method, based largely on publicly available data, has the least effect around the publication date with regard to new information coming to the market. Therefore, it is expected that the Interbrand brand value estimate announcement has the least effect on immediate stock price fluctuations, whereas Millward Brown’s method has the largest impact on immediate stock price fluctuations. The goal of this paper is to discover differences in the brand value – stock return relationship between three different brand valuation methods: Interbrand, Millard Brown and BrandFinance.
Based upon the relationship between brand equity and firm performance, the following hypotheses arise:
H1 An increase in Brand value, as measured by Interbrand, Millward Brown and BrandFinance, results in increasing stock returns.
H2 An increase in Brand value, as measured by Interbrand, Millward Brown and BrandFinance, results in declining equity-holder risk.
- 14 -
Neoclassical Finance versus Behavioral Finance
The traditional view that securities are rationally priced, reflecting all publicly available information, is under challenge by academic researchers. For example, the drift of stock prices after major corporate news announcements finds an extensive body of empirical literature (Frazzini, 2006). The notion of efficient markets is contradicted by recent literature in behavioral finance (Paudel and Laux, 2010). The emphasis on psychological biases influencing the behavior of investors and pricing of assets, results in a strong debate among proponents of behavioral finance and neoclassical finance. Also Baker and Wrugler (2007) argue that the standard finance model of unemotional investors forcing capital markets to equal the rational present value of expected future cash flows is under debate. They give examples of certain events that cannot be explained by classical finance models. Explanation for those examples is that investors are subject to sentiment.
Heuristics, mental shortcuts, are proposed as an explanation for irrational investor behavior. Tversky and Kahnema (1974) discuss three different heuristics that people rely on to reduce the complex tasks of assessing probabilities and predicting values to simpler judgmental operations. Representativeness, availability and adjustments & anchoring are three judgmental heuristics people rely on. The affect heuristic, wherein people rapidly consult their affective feelings, when making judgments and decisions is another example (Slovit et al., 2002).
Based on existing behavioral finance literature, a few implications are developed about the impact of increasing brand equity on stock returns and trading volume.
Positive versus Negative Brand Value Changes
- 15 -
Brand Value and Trading Volume
Shefrin and Statman (1985) explain the disposition effect in their paper. They found that the realization of a gain on stocks is made quicker than the realization of a loss: sell winners too early and ride losers too long. In past literature and hypothesized in this research, high brand values have a positive effect on stock returns. Stocks of companies owning brands with high brand values will be regarded as winners. In line with the disposition effect, those stocks will be sold early, in order to realize a gain and induce pride. Those stocks of companies with high brand values will be traded more frequently since investors sell them too early.
Furthermore, people prefer things that are familiar to them. Fans support their local sports teams and employees are often the proud owners of stocks of the companies they work for. This is because the sports team and the company are familiar to them (Norfsinger, 2011). Heath and Tversky (1991) even found that people pick a more familiar gamble, even if the odds of winning are lower. At least, with two different gambles, yielding the same odds of winning, people pick the better-known gamble.
Moreover, investors seem to trade in securities that are familiar to them. People feel comfortable in having their money invested in a business that is visible to them (Huberman, 2001). The United States already offers tens of thousands of potential stock and bond investments solely and worldwide those investment opportunities are even larger (Norfsinger, 2011). However, investors seem to choose their investment based on familiarity.
Huberman (2011) demonstrates that an investor’s perception of risk and return are affected by the strong persuasive influence familiarity has on investment decisions. In ignorance of (modern) portfolio theory, people seem to invest in familiar stocks due to the familiarity heuristic. Moreover, people expect those stocks to deliver a higher return at lower stock-specific risk. According to Huberman (2011), this results in undiversified portfolios, concentrated in stocks that people know, for instance due to visibility in an investor’s live or stocks that are discussed favorably in the media.
Since, in general, companies from an investor’s home country are more familiar to them than companies from foreign companies, people have a home bias. Research conducted by The International Monetary Fund shows that investors overwhelmingly keep their money at home. Their assessment of equity portfolios finds this home bias, resulting in undiversified portfolios (Norfsinger, 2011). However the part that people invest in foreign firms is invested in companies that are familiar to them. This results in investments in large companies, with recognizable products. Kang et al. (2005) found that foreign investors (non-Japanese) seem to hold stakes in the large Japanese companies.
- 16 - propensity towards companies with easily recognizable products. The preference of recognizable products is in line with the notion that individuals prefer to invest in stocks for which they have more high quality information (due to greater familiarity with the firm’s products). In their paper, Frieder and Subramanyam (2005) suggest a spillover effect of marketing actions to investment decisions in the market for companies’ stocks.
Lakonishok et al. (1994) deal in their paper with the discrepancy of a good company and a good investment. They describe the demand for ‘glamour stocks’, but simultaneously find underperformance of those ‘glamour stocks’ compared to ‘value stocks’. Suppose that investors might remark brands appearing in one of the brand lists of Interbrand, Millward Brown, or BrandFinance as glamorous. This implies that high valued brands do not make a good investment by definition, but do result in above average demand and trading.
Norfsinger (2000) discovers in his paper the effect of news in the Wall Street Journal on stock trading. In this study, both firm specific news and general economic news are taken into account. He found investors conducting a high degree of trading around news releases, especially earnings and dividend news. Both good and bad news had impact on the trading activity. Regarding the publication of the brand lists to be firm specific news, the publication of those lists might induce investors to trade. This effect should especially hold for the Interbrand and Millward Brown brand lists, since BusinessWeek and the Financial Times release these respectively.
Based on the theory of loss aversion (Kahneman, Knetsch and Thaler 1991), it is hypothesized that a decline in brand values has more impact on stock prices than an increase in brand value.
Well-known worldwide brands, brands listed on the Interbrand, Millward Brown and BrandFinance most valuable brand lists, make the companies owning those brands familiar to a lot of people. Due to the familiarity heuristic (Huberman, 2011) extensive trading in those companies is expected. Moreover, despite the home bias, evidence from Japan shows that people tend to invest their investment money reserved for investing in foreign markets in familiar companies (Kang et al. 2005). Frieder and Subramanyam (2005) also found that people prefer visible brand name stocks in their portfolios.
- 17 - The following hypotheses arise from a behavioral finance perspective, stressed to the relationship between brand value and stock performance;
H4 After a decline in brand value, the impact on stock return change will be larger than after an increase of brand value.
H5 An increase in Brand value, as measured by Interbrand, Millward Brown and BrandFinance, result in increasing stock trading frequency.
3. Methodology
To test the relationship between brand value and stock return, equity-holder risk and trading volume, three sample pools are formed based on the brand values given by three different valuation companies: Interbrand, BrandFinance and Millward Brown.
The corresponding stock is incorporated in the sample pool(s) in the years the brand is listed in the ranking, whenever the brand satisfies the criteria of selection, as described in the next section. The inclusion process results in three sample pools containing an unbalanced panel (again, one for every valuation method).
Longitudinal data, or panel data, is the denomination of a dataset containing both time series and cross-sectional elements. It is often of interest to examine how variables, or the relationship between them, change dynamically (over time). A long run of data, simply to get a sufficient number of observations to be able to conduct a meaningful hypothesis, would be required to do this using pure time-series data. The combination of cross-sectional and time series data can increase the number of degrees of freedom. The power of the test is increased, by employing information on the dynamic behavior of a large number of entities at the same time (Brooks, 2008). Besides a pooled regression, the simplest way to deal with panel data, two classes of panel estimator approaches will be used for the above-mentioned reasons. Both fixed effect models and random effects models are employed.
The basic setup for the first, second, fourth and fifth hypotheses, testing the relationship between brand value and stock return, risk and trading volume can be described econometrically in the following equation:
- 18 - where γi,t is the dependent variable, αi is the intercept term, β is a k x1 vector of parameters to be estimated on the explanatory variables, and xi,t is a k x1 vector of observations on the explanatory variables, t = 1, …., T; i = 1,…., N.
Pooled regression ~ the dataset for y is stacked up into a single column containing all the
cross-sectional and time-series observations, ai can be defined as a and has a single intercept estimate, and the observations on each explanatory variable would be stacked up into single columns in the x matrix.
Fixed Effects Model ~ given an expected average change in value of yi,t over time, cross-sectionally or both, a time fixed effects model allows the intercept to vary over time, entity or both respectively. The disturbance term, εi,t, could be decomposed into two elements. The decomposition contains an entity specific effect, time specific effect or both and a remainder disturbance term, vi,t, that varies over time and companies. The model could be estimated using dummy variables: least squares dummy variable (LSDV). To avoid a dummy variable trap, the ai is removed from the equations and the, ai,t,specifies the firm, time or combined unobserved heterogeneity term respectively as mentioned above. The specification becomes as follows:
(2)
Random Effects Model ~ different intercept terms for each entity, constant over time, are
proposed under the random effects model. However, in contrast to the fixed effects model, the intercepts for each cross-sectional unit are assumed to arise from a common intercept α plus a random variable i. The random effects panel model can be described as:
, (3)
- 19 - The above-described estimation models will be used for testing the first, second, fourth and fifth hypotheses. The methodology behind the event study of hypothesis three will be discussed accordingly. As actual return, Ri,t , the continuously compounded return is used, where RIi,t is the total return index of stock i at year t and RIi, t-1 is the return index on year t-1;
(4)
Mathematically, the expression of the main relationship between brand value and stock returns would look as follows:
(5)
The dependent variable, Ri,t , is the continuously compounded stock return of company i in year t; is the relative change in brand value for company i in year t ; and is the error term for stock i in year t.
In classical finance theory, return is always associated to risk. Abnormal returns are analyzed by controlling for risk factors by using the extended capital asset pricing model (CAPM), a model relating risk to expected returns (Sharpe and Lintner, 1964, 1965). However, empirical studies show that the CAPM does not properly describe the relation between risk and expected return (Fama and French, 1992). No reliable relation between stock returns and beta are found by Fama and French (1992) and for the period 1963-1990 they found the importance of size and book-to-market equity in explaining cross-section average stock returns. They discovered that two classes of stocks tended to do better than the market: small capitalization stocks and low market-to-book ratios stocks. Therefore, the market value or market capitalization of a firm (firm size) and the market to book value of a firm are specified in the Fama and French equation as additional systematic risk factors.
- 20 -
(6)
Where the dependent variable, Ri,t,is the continuously compounded stock return of company i in the year t; is the relative change in brand value of company i in year t; ln is the natural logarithm of market value of company i, in the year t; is the change of market to book value of company i, in the year t; and is the error term for stock i in year t.
Besides considering brand value to be a potential risk factor implicitly in the brand value –
stock return hypothesis, brand value is remarked explicitly as a potential risk factor in the brand value – equity risk hypothesis. In order to test this relationship between brand value
and equity risk, the method used by Rego et al. (2009) is used as a guideline. The variance of stock returns is used as a measure of total equity risk. Several control variables are added to the analyses, in order to control for factors that are already known to affect firm risk. The used control variables are Firm Size (measured as total assets, in line with Rego et al., 2009) and Market to Book Value (MTBV), both discussed by Fama and French (1992) and Rego et al. (2009). Furthermore, in line with the control variables added in the study of Rego et al. (2009) the following control variables are also added: Leverage, Return on Assets (ROA) and ROA variability. In contrast with Rego et al. (2009), but in line with the first hypothesis of this thesis, the relative change in brand value is used in this regression. It is of interest of this thesis to test the effect of the change in brand value on different dependent variables. To be consistent along the different variables the first difference of Risk is taken as dependent variable and with respect to the control variables also the first differences of leverage, MTBV, ROA and ROA variability are used. Lastly the natural logarithm of Firm Size is taken. Mathematically, the main relationship between brand value and equity-holder risk can be expressed as follows:
(7)
- 21 - The effect of new information on the market value of a firm’s stock is measured by event studies. When new information is released, investors update their beliefs about the firm’s expected future cash flows and react by trading in the firm’s shares. This leads to a change in the firm’s continuously compounded daily stock return, Ri,t as given under (1).
The actual stock return, Ri,t, is compared to the expected return under normal conditions (without the event), as given under (8):
(8)
Where E(Ri,t) is the expected return, Rm,t represents the continuously-compounded stock return over trading day t on a benchmark portfolio m. The MSCI index is used for this purpose. The parameters and are OLS estimates obtained from regressing Ri,t on Rm,t over a certain estimation period preceding the event. In this thesis the period over 250 trading days prior to the event date is used, similar to MacKinlay (1997).
Following Brown and Warner (1985), the market model is used to obtain estimates of expected returns. To find the effect of the event on the stock return, the difference between the observed actual return and the estimated expected return is calculated (9).
(9)
Where ARi,t is the abnormal return for company i on day t; Ri,t is the actual continuously compounded return for company i on day t and E(Ri,t,) is the expected return under normal conditions. The average abnormal returns are obtained by averaging these abnormal returns across firms in common event time (10).
(10)
Where is the average abnormal return, N is the number of firms in the sample and 0 refers to period 0 in event time.
The parametric tests rely on the assumption that individual firm’s abnormal returns are normally distributed. The standard statistic is given below:
(11)
Where is defined as above and S( ) is an estimate of the standard deviation of the
- 22 - In order to test the fourth hypotheses a dummy variable for positive brand value changes and a dummy variable for negative brand value changes is incorporated in equation (6). Mathematically the new equation can be expressed as follows:
(12)
Where the dependent variable, Ri,t,is the continuously compounded stock return of company i in the year t; is the relative change in brand value of company i in year t; DPOS is a dummy variable for positive brand value changes and DNEG is a dummy variable for negative brand value changes; ln is the natural logarithm of market value of company i, in the year t; is the change of market to book value of company i, in the year t ; and is the error term for stock i in year t.
The last hypothesis, containing a brand value ~ trading volume relationship, will be tested with a similar regression formula as the first hypothesis. Again, the regression will be controlled for market value (firm size estimate) and market to book value. The following expression gives, mathematically, the relationship between brand value and trading volume:
(13)
Where the dependent variable, ΔVOLi,, t,is the relative change in trading volume of company i in the year t; is the relative change in brand value of company i in year t; is the natural logarithm of market value of company i, in the year t; is the change in market to book value of company i; and is the error term for stock i in year t.
4. Data Collection
- 23 - different valuation methods is discussed. Lastly, the inclusion criteria and process of inclusion of data in the dataset is presented.
Description of the variables
1. Independent Variable
Brand Characteristics
The brand name, the name of the company owning the brand, the country of origin, the sector the company is operating in and the ISIN number of all brands are collected. Brand names, country of origin and the sector the brand is operating in are largely gathered from the Best Global Brands publicized by Interbrand, the BrandZ top 100 most valuable brands publicized by Millward Brown and the BrandFinance Global 500 publicized by BrandFinance (2011 publications can be found in Appendix A). Missing information and ISIN numbers are found in Orbis.
Absolute brand value (BVi,t)
Brand values estimated in the Interbrand best global brands ranking of Interbrand during the time span 2006-2011 are used. Furthermore, brand values estimated in the Millward Brown’s BrandZ top 100 most valuable brands during the time span 2006-2011 are used. Lastly, brand values from the top 100 companies estimated in the BrandFinance top 500 most valuable brands during the time span 2007-2011 are used.
Brand value estimates for BrandFinance are only available from 2007-2011, whereas brand value estimates for Millward Brown are only available from 2006-2011. In order to include a maximum number of brand value estimates, all observations from BrandFinance and Millward Brown are included in this thesis. However, with regard to the Interbrand brand value estimates the sample period is matched with the two other valuation companies. Interbrand brand value estimates from 2006-2011 are used, in order to compare the test results from the three valuation methods over a similar sample period.
Interbrand’s valuation method, Millward Brown’s valuation method and BrandFinance’s valuation method are summarized in the next subsection and all values are in millions of dollars.
Relative changes in brand value (∆BVi,t)
- 24 - value is the main independent variable in this thesis. The change in brand value gives information to the market concerning the development of brand values of firms. Investors will (instantly) buy or sell stock based on new expectations, altered by the change in brand value, they form on discounted future cash flow.
It can be expected that the increase in a firm’s stock return may be relatively modest if a firm already has a high accumulated brand value. When a firm’s accumulated brand value is small, a given increase in a firm’s brand value probably relates to a larger increase in a firm’s stock return. A 3 million dollar increase in brand value means doubling their brand value for LG Electronics, a 3 million dollar increase for Coca Cola is only a relatively small improvement. In order to account for the large difference in “base” brand values and to compare brand value changes of different companies relatively, relative changes in brand value are used in the regressions.
Dummy variable for positive brand value changes (DPOS)
This is a binary variable coded 1 if the relative brand value change is equal or larger than zero.
Dummy variable for negative brand value change (DNEG)
This is a binary variable coded 1 if the relative brand value change is smaller than zero.
Publication dates of brand rankings
To test the third hypothesis on shocks in stock prices around the publication date, the publication dates are needed. The publication dates are presented in table 1.
Table 1: Publication dates of brand lists by Interbrand, Millward Brown and BrandFinance
Publication dates are found on the Interbrand, Millward Brown and BrandFinance websites. For Interbrand and Millward Brown the websites of BusinessWeek and Financial Times are also consulted. Moreover e-mail contact with representatives of all firms has confirmed the correct dates.
Year Publication Date
Interbrand Millward Brown BrandFinance
- 25 -
2. Dependent variables
The total return index (RIi,t)
Yearly total return indices are downloaded via Datastream (Thomson Reuters, 2011) for all companies. These concern the total stock return index of the companies that own the brands listed in the brand value lists of Interbrand, Millward Brown and BrandFinance. Returns rather than raw price series are collected, since working with returns is preferable. Moreover, returns have the benefit to be unit free (Brooks, 2008) The total return index can be defined as the stock price time series adjusted so that the dividends are added back. The total return index recorded on every last day of the year is recorded for the time span 2006-2011.
The natural logarithm of the annual change of the total return index (Ri,t), the continuously compounded return, will be used as the dependent variable in the regression equation, testing the first and fourth hypothesis. As an extra control variable, a lagged dependent variable will be included in some of the regressions (Ri,t-1).
Equity-holder Risk (Riski,t)
Risk is measured in this paper by volatility. Volatility for all different companies is calculated as the deviation of daily stock prices. Daily stock prices are obtained via Datastream. The
formula used to calculate volatility is as follows: . (14)
The calculated volatility is based on daily retrieved official closing prices of all stocks for all companies owning a brand listed in the brand value list of Interbrand, Millward Brown and BrandFinance for the time span 2006-2011. The year to year change in risk (∆Riski,t) is taken as dependent variable for the second hypothesis. In order to create consistency and to dive into the effect of the change in brand value, also the change in risk is used. For all (control) variables, it holds that either the natural logarithm or the first difference of the variables is used.
Trading Volume (VOLi,t)
This variable shows the number of shares traded for a stock on a particular day. Numbers on trading volume are gathered via Datastream on a yearly basis. The trading volume recorded on every last day of the year is recorded for the time span 2006-2011. The trading volumes are given in millions.
- 26 -
3. Control Variables
Market Value (MVi,t)
The total amount of market value is calculated as shares outstanding times the share price. Data concerning the market value of a firm, or market capitalization are retrieved via Datastream. The Market Value recorded on every last day of the year is recorded for the time span 2006-2011. In order to control for the size effects (smaller firms have relatively higher returns), market value data is collected. Two easily measured variables, size and book-to-market equity are found by Fama and French (1992) to explain average stock returns. Market values are displayed in millions of dollars.
The natural logarithm of market value (lnMVi,t) is used as a control variable in the regression equations regarding the first, second, fourth and fifth hypothesis. The natural logarithm of market value is used in order to create a normally distributed variable.
Market to book value (MTBVi,t)
The market to book value is retrieved via Datastream at the security level. The Market to book value is defined as the market value of ordinary (common) equity divided by the balance sheet value of the ordinary (common) equity in the company. Market to book value is gathered for all companies that own brands in the brand value lists of Interbrand, Millward Brown and BrandFinance for the time span 2006-2011.
Stocks with low market-to-book ratios tend to have higher returns than stocks with high market-to-book ratios (Fama and French, 1992). To eliminate the book to market effect bias, the change in market to book value (∆MTBVi,t) is used as a control variable. In order to be consistent in the equations, the first difference of market to book value is taken and included as a control variable in the regression equations for hypothesis one, two, four and five. Consistency along the variables is created by taking either the natural logarithm of the variables or the first difference. Since the market to book value refers to a ratio that also can take on non-positive numbers, the absolute change is used.
Firmsize (FirmSizei,t)
- 27 -
Leverage (Levi,t)
Leverage is computed as the net debt divided by the total assets. Leverage is included as a control variable, since leverage is related to equity risk in financial literature (Rego et al., 2009). Net debt and total assets are gathered via Datastream on a yearly basis for the time span 2006-2011 for all companies that own a brand listed in the brand value lists of Interbrand, Millward Brown or BrandFinance.
In order to be consistent in the equations the first difference of leverage (∆Levi,t) is taken and included as a control variable in the regression equations to test the second hypothesis. Consistency along the variables is created by taking either the natural logarithm of the variables or the first difference. Since leverage refers to a ratio that also can take on non-positive numbers, the absolute change is used.
Return on Assets (ROAi,t)
Return on Assets (ROA) is retrieved via Datastream on a yearly basis. The return on assets is reported for the time span 2006-2011 for all companies owning a brand listed in the brand value lists of Interbrand, Millward Brown and BrandFinance. Greater ROA should be associated with lower equity risk because it indicates the uncertainty of the firm’s likely future financial health (Rego et al., 2009).
As with the market to book value and leverage, also ROA refers to a ratio that can take on non-positive numbers. The first difference of ROA (∆ROAi,t) is used as control variable in the second hypothesis.
Return on Assets variability (ROA_vari,t)
The Return on Assets variability is computed as the standard deviation of the prior five years’ ROA, based on yearly retrieved ROAs of all stocks for all companies owning a brand listed in the brand value list of Interbrand, Millward Brown and BrandFinance for the time span 2006-2011.
Greater ROA variability should be associated with higher equity risk because it indicates uncertainty of the firm’s likely future financial health (Rego et al., 2009). The formula used to
calculate ROA variability is as follows: . (15)
- 28 -
Description of the valuation methods
Interbrand’s Method
Interbrand’s method of computing brand value is based on three steps. During the first step the financial performance is analyzed. Economic profit is calculated based on a terminal value after the explicit forecasting period and a defined five year forecasting horizon. The economic profit is calculated as net operating profit minus taxes (NOPAT: net operating profit after taxes) discounted by an industry weighted average cost of capital.
In the second step, the economic profit is multiplied by the role of brand resulting in branded earnings. The role of brand derives from, depending on the brand, primary research, a review of historical roles of brand for companies in that industry or expert panel assessment.
In the last step, the brand value is computed, based on the branded earnings multiplied by the brand strength discount rate. This brand strength is reported on a 0-100 scale and consists of ten brand strength components: clarity, commitment, protection, responsiveness, authenticity, relevance, differentiation, consistency, presence and understanding.
Millward Brown’s Method – BrandZ
Millward Brown’s methodology to compute brand values is based on an economic use approach. The process is divided into three steps. In the first step, the financial information of companies is assessed. Based on corporate earnings, branded earnings and finally branded intangible earnings are calculated. In the second step, the brand contribution is determined. This shows the portion of these branded earnings that are generated due to the brand’s close bond with its customers. In the third step, the brand multiple is constructed. The brand multiple has a future looking purpose. This multiple determines the growth potential of the brand-driven earnings.
In order to calculate the brand value of a certain brand, the items from the different steps are multiplied: branded earnings times brand contribution times brand multiple.
BrandFinance’s Method
- 29 - In the fourth step, the appropriate discount rate is determined and this rate is used in the fifth step. This last step shows brand values, equal to the net present value of post-tax royalties.
The royalty Relief approach is based on the assumption that if a company did not own any trademark it would need to license them from a third party trademark owner instead. Ownership therefore ‘relieves’ the company from paying a license fee (the royalty) for the use of the third party trademarks.
A graphical overview of the different valuation methods (Interbrand, Millward Brown and BrandFinance) is given in Appendix B.
Correlation of the three valuation methods
Interbrand, Millward Brown and BrandFinance all use an economic use approach to estimate brand values. However, differences in valuation methodology and differences in estimated brand values are observed and mentioned in the literature part. This subsection shows the correlation between the brand value estimates of the three brand valuation companies. Table2 shows the correlations between the estimated brand values, reported by Interbrand, Millward Brown and BrandFinance for all brands that appear on every brand value list. A total of 48 have estimated brand values from all three brand valuation companies.
Table 2: Correlation between brand value estimates from Interbrand, Millward Brown and BrandFinance
As can be seen in table 2, almost no correlation is found between the method of Millward Brown and one of the other two methods. BrandFinance’s brand value estimates and Interbrand brand value estimates show a correlation of 0.746. The absence of correlation between Millward Brown and the other two brand valuation firms raises the question whether the brand valuations of three different firms are consistent with each other. The correlation between Interbrand and BrandFinance brand value estimates suggest that the influence of changes in brand value on changes in stock return, changes in equity-holder risk and changes in trading volume might follow a similar pattern, whereas the influence of Millward Brown might result in different outcomes. Conversely, it should be noted that many companies only appear on one (or two) out of the three brand value lists used in this thesis. The correlations shown in table 2 are based on a subsample of all the companies included in this thesis.
Interbrand Millward Brown BrandFinance
Interbrand 1 0.030 0.746
Millward Brown 0.030 1 -0.002
- 30 -
Inclusion criteria dataset
In order to include the data on brand values, stock returns, trading volumes and risk, the dataset is adjusted according to some selection criteria. In this section, the process of inclusion or exclusion of the different brands and companies owning those brands is described. Selection criteria are the same for all three valuation companies and their brand value lists (Interbrand, Millward Brown and BrandFinance).
At a first stage, all brand names and brand values are taken from the reports published by Interbrand and Millward Brown for the time span 2006-2011 and BrandFinance for the time span 2006-2011. Names of the parent companies owning the brands appearing on at least one of the brand value lists are linked to every brand. Corresponding ISIN numbers of all parent companies are subtracted via Orbis. Using those ISIN numbers, data on different variables as described in the previous subsection are acquired through Datastream. This first stage resulted in 600, 600 and 500 brand value estimates in the three separate datasets containing data from 129, 152, and 154 companies for Interbrand, Millward Brown and BrandFinance respectively.
At a second stage, all companies that do not contain brand value estimates for two subsequent years are removed from the samples. This is simply done, because without brand value estimates for two subsequent years, the change in brand value cannot be calculated. The removal of companies without subsequent brand value estimates result in 117, 121 and 107 companies for Interbrand, Millward Brown and BrandFinance respectively.
At a third stage, all privately held companies are removed from the sample. Datastream provides no data on privately held companies. After losing all privately held companies, there are 112, 114 and 99 companies left for Interbrand, Millward Brown and BrandFinance respectively.
At a fourth stage, companies with incomplete data, at least for one of the variables, for all years of the sample period are removed from the samples. This results in a dataset containing 106, 108 and 95 companies for Interbrand, Millward Brown and BrandFinance respectively.
- 31 - Interbrand and Millward Brown respectively and 299 brand value changes over the time-span 2007-2011 for BrandFinance.
Summary statistics of the estimated brand values by Interbrand, Millward Brown and BrandFinance in the final sample are given in table 3. Extensive descriptive statistics on all different variables as described in the first part of this section are given in Appendix C. The data selection and data adjustment process is summarized in the table shown in Appendix D.
Table 3: Mean, Maximum, Minimum and Standard Deviation of the estimated brand values by Interbrand, Millward Brown and BrandFinance over the time span 2006-2011($m)
5. Results
For all three brand valuation methods included in this study a sample of brand value estimates is used to execute four different regressions on the effect of brand value changes on changes in stock returns, changes in equity-holder risk and changes in trading volume. In addition, an event study is performed to test for immediate stock reactions to the release of new brand value estimates.
With regard to the regression equations, three different estimation methods are used. Firstly, a pooled ordinary least squares regression is performed. Subsequently, a time fixed effects model, an entity fixed effects model and a both time and entity fixed effects model is estimated. Finally, a random effects model is also performed. A likelihood ratio test (redundant fixed effects test) is performed, both in χ2 and F-test versions, to determine
whether the fixed effects are necessary or not. A Hausman test (χ2) reveals the appropriateness of a random effects model.
Except for the regression equation on the relationship between brand value changes (in case of Interbrand and BrandFinance brand value estimates) and changes in equity-holder risk, a both time fixed and entity fixed effects model is statistically most appropriate in all regression estimations. This confirms the expectation, from an economical view, that both time fixed effects and entity fixed effects are due. With regard to the exceptions, the above-mentioned Interbrand case statistically requires a pooled ordinary least square, whereas in the BrandFinance case a random effects model suits best.
Interbrand Millward Brown BrandFinance
Mean 13.012 20.794 16.041
Maximum 71.861 153.285 45.441
Minimum 1.350 18.407 2.867
Standard Deviation 13.207 18.388 8.222
- 32 - For every single regression, the estimation results corresponding to the most appropriate estimation model are presented and discussed below. Estimation output of all different estimation models on all regressions can be found in Appendix E.
Regression Results
Brand value – Stock Return
In the first series of regressions, equation (6), the logarithm of the annual change in the return index is the dependent variable. The year-to-year change in relative brand value is the main explanatory variable. The regression equation includes two control variables: market value and market to book value. The natural logarithm of market value and the first difference of the market to book value are used.
Interbrand As can be seen from table 4, the coefficient of relative brand value change is not significant. The intercept is negative and significant. The natural logarithm of market value (control variable) gives a positive and significant result at a 1 % level, whereas the change in market to book value (control variable) gives no significant result.
Millward Brown In case of the Millward Brown brand value estimates the intercept is negative and significant (table 4). The coefficient of brand value change (β1), the slope, has a negative sign (β1 = -0,129) and is significant at a 5% level. Both control variables, the natural logarithm of market value and change in market to book value, give a positive and significant result at a 1 % level.
BrandFinance As can be seen from table 4, the intercept is significant and negative. The coefficient of brand value change (β1) is significant and has a negative sign (β1 = -0,156). With regard to the control variables, the natural logarithm of market value gives a positive and significant result, whereas the change in market to book value shows no significant result.
- 33 - explanation could come from a behavioral finance perspective in line with Lakonishok et al. (1994). Lakonishok et al. (1994) define “glamour” stocks as stocks that have done well in the past or are recognized to represent well-run companies. However, they note a discrepancy between a good company and a good investment. Those “glamour” stocks are not a good investment by definition and deliver worse returns than “value” stocks do. Suppose that increasing brand value leads to an increase in “glamour” and an increase in “glamour” results in a worse stock return. Eventually, increasing brand values result in decreasing returns.
Brand Value – Equity-holder risk
For equation (7) equity-holder risk, measured as the variance of stock returns, is the dependent variable. The year-to-year change in relative brand value is the main explanatory variable. The regression equation includes five control variables: firmsize, leverage, return on assets (ROA), ROA variability and market to book value. The natural logarithm of the firm size is taken and the first difference of the other control variables is used.
Interbrand As can be seen from table 5, the coefficient of brand value change is not significant, neither is the intercept. At a 10% level the change in leverage is significant and has a negative sign. None of the other control variables show any significant results.
Millward Brown As can be seen from table 5, the coefficient of brand value change (β1) has a negative sign (β1= -0.600) and is significant at a 1% level. For both time-fixed effects and entity-fixed effects model none of the control variables give a significant estimate.
