THE VALUE RELEVANCE OF NONFINANCIAL PERFORMANCE INDICATORS: NEW CUES FROM THE EUROPEAN FASHION INDUSTRY Francesco Dainelli1 – Francesco Giunta *
Abstract Intangible assets and related performance measures assume increasing importance in valuation processes. Value relevance studies testify to their importance through an analysis of market stock prices. We aim to examine the value relevance of non-financial indicators in European fashion companies. The indicator selected is the “change in mono-brand stores”. Applying the models proposed by current literature, we have refuted the value relevance hypothesis. However, refining both the “operationalization” of the concepts and the related result analysis procedure, the value relevance is confirmed. In this way, we contribute to increasing the generalizability of this research trend and to fuel the debate concerning the standardization process of this information. In particular, following in the footsteps of the Gartner/EBRC project, supported by AICPA (Gartner-EBRC, 2010), our results can help the national and international standard setters to pinpoint the indicators that really matter for the fashion industry and standardize their communication. Key words Intangibles; value relevance; financial measures; non-financial measures; fashion industry; mono-brand stores.
1
[email protected]
* The authors thank Marco Pericci for his precious contribution to data collection and elaboration.
Electronic copy available at: http://ssrn.com/abstract=1938979
Dainelli F. - Giunta F.
1. Introduction The competitive strength of companies is more and more based on intangible assets. As a consequence, classical tangible assets which are represented in financial statements have gradually been losing their economic importance (Amir and Lev, 1996; Collins et al., 1997; Francis and Schipper, 1999; Healy and Wahlen, 1999; Lev and Zarowin, 1999; Zambon, 2003). The intangible asset contribution to company competitiveness is related to the specific characteristic of these assets, as a lock-in effect (Shapiro-Varian 1999), “scalability (non-rivalry), increasing returns, and network effects versus partial excludability (the general lack of full control over the benefits of intangibles), inherent risk, and nontradability (absence of organized markets in intangibles)” (Lev 2001, p. 6). This fact is testified by the increasing gap between the book value and the market value of the companies’ stocks. During the period 1977-2001, in the USA stock market, the market-to-book ratio increased six times and some of the research argues that this growth depends on the intangible assets which are not recognized in financial statements (Amir and Lev, 1996; Chan et al., 2003; Collins et al., 1997; Hand and Lev, 2003; ICAEW, 2009; Skinner, 2008; Taub, 2003). No surprise then that, from the beginning of the 1970s, the attention of researchers has been focused on those performance indicators that can capture the value of intangibles (Lev and Zambon, 2003). On the one hand, the systemic approach to managerial control has put in evidence the limitations of the classic financial metrics (Edvinsson and Malone, 1997; Hofmann, 2008; Kaplan and Norton, 1992; Stewart, 1997; Sveiby, 1997). On the other hand, research which concerns the market efficiency asserts that companies have to communicate specific non-financial indicators to their stakeholders, above all to the stockholders (AICPA, 1994; Wallman, 1995). Looking at the financial markets, many studies throw into relief the leading role that financial analysts play and the growing use, on their part, of non-financial performance indicators (Breton and Taffler, 2001; Previts et al., 1994; Rogers and Grant, 1997). A quite recent European Commission survey has also come to the conclusion that about 35% of investing decisions are conditioned by information concerning intangible assets (Eustace, 2000).2 2
This survey is part of “The intangible economy impact and policy issued – Report of the European high level expert group on the intangible economy” which was edited in 2000 by the Commission of the European Communities Enterprise Directorate General.
2
Electronic copy available at: http://ssrn.com/abstract=1938979
The value relevance of non-financial performance indicators
Against this background, law makers and standard setters have gradually realized the importance of regulating non-financial communication in order to reduce information asymmetries, which could heavily influence the investor’s decisions (Zambon, 2003, chap. 3). At the beginning of the 1980s, the SEC is the first regulator that required companies to disclose non-financial indicators in the narrative section (MD&A) of their annual reports. In 1993, the Australian Securities Exchange (Review of Operations and Activities: Listing Rule 4.10.17 – 1993) and Great Britain (Operating and Financial Review, ASB, 1993) took a step in the same direction. From 2000 on, Canada (Canadian Institute of Chartered Accountants, Management’s Discussion and Analysis, 2002) and the European Union (EU) made the same choice. The EU, in particular, in the Directive 51/2003/CE requires limited companies to show in their management commentary “financial and, where appropriate, non-financial key performance indicators relevant to their particular business” (art.1, point 14, letter b). The meaning of the above mentioned regulation is contained in this question: are non-financial indicators really useful for the investor’s decisions? Value relevance studies try to answer this question. In short, these studies measure the usefulness of a specific indicator considering the relationship between the level of this indicator and the market value trend of the stock of the company to which the indicator refers. An indicator is considered useful to the investor if it is able to predict stock market performance; in this case, the market trend in turn is conditional on the indicator disclosure to the market. In the light of this, our paper examines non-financial indicators in the European fashion industry. The fashion industry has been selected on the basis of two considerations: the high technological and innovative content of its production and the wide gap between book and market value of the companies’ stocks. Our research, on the one hand, sets out to gauge the value relevance of a well-known non-financial indicator; on the other hand, it aims at verifying the capacity of the several models proposed by literature, to effectively work using datasets which are different from those on which they were built and put to the test. The non-financial indicator that we have chosen is the “change in monobrand stores”. This measure should be value relevant because it synthesizes the intangibles which are present in a fashion company (brand, sales and distribution channels, commercial strength, organizational capacity etc.) 3
Dainelli F. - Giunta F.
and the growth capacity of a company better than other indicators. This hypothesis is supported by the fact that the indicator at hand is the most widespread indicator in company’s financial statements and financial analyst’s reports. Generally, value relevance studies are based on linear regression models where the dependent variable is a variable which expresses the market value. Applying the regression models which are proposed in the main literature, we initially had to refute the value relevance hypothesis for the above mentioned indicator. This result has spurred us to critically revise the models we have referred to. In fact, the applied models differ among themselves with respect to either the “operationalization” of the concepts which they refer to or the results analysis. Refining the “operationalization” of the concepts which these models are based on, we managed to elaborate a new model that confirms the value relevance of the “change in mono-brand stores” indicator in the fashion industry. Our results contribute to fuel the debate concerning the importance of non-financial indicator disclosure for the efficiency of the financial market, In particular, the great information capacity of some indicators which are considered key measures for a specific industry, is confirmed. For this reason, from an operative point of view, our study buttresses the activity of the standard setters that are involved in the non-financial disclosure regulation. Following in the footsteps of the Gartner/EBRC project, supported by AICPA (Gartner-EBRC, 2010), our results can help the national and international standard setters to pinpoint the indicators that really matter for the fashion industry and standardize their communication. In addition, as far as we know, our study is the first research concerning the non-financial indicator value relevance that encompasses listed companies belonging to several European countries. This fact should contribute to increasing the “generalizability” of our research results. Even the non-financial indicator that we have chosen is public, as it has been drawn from the financial analysts’ reports and the companies’ financial statements. On the contrary, the existing literature has mainly used information that is not widely disclosed to the market. For this reason, it is plausible to hypothesize that our indicator is well known by the users, and it can be conditional on the stock market trends more so than a measure which is available only from a source difficult to access. In the following section, we will show a literature review, throwing into relief the analysis models used by the authors. The dataset, which our research is based on, will be presented in the third section. The fourth and 4
The value relevance of non-financial performance indicators
fifth section apply to our dataset the most perspicuous analysis models discussed in the second section. The sixth section proposes a few methodological refinements of the analysis models formerly applied to our dataset, and discusses the results that we have achieved due to these refinements. The sensitivity analysis is carried out in the seventh section. The last section, eventually, draws the conclusions.
2. Literature review Many pieces of research have shown the value relevance of nonfinancial performance indicators. Generally, this research uses linear regression models to estimate the correlation between these indicators and stock prices or returns. Birchard (1994) is one of the first researchers who states that companies should widen their disclosure to the market, communicating non-financial indicators. This statement is grounded on a case study. Birchard analyses the Whirlpool Company, using interviews and descriptive statistics. Amir and Lev (1996) develop their research looking at the U.S. cellular communications industry – when this industry was in full growth –, applying two sets of linear regression models. In the first set, the authors relate stock prices and returns to a few financial measures like Earnings Per Share (EPS), change in earnings per share, and Book Value Per Share (BVPS). They conclude that, in the examined industry, the financial indicators are of little significance for the market. In the second set of regression models, the authors include non-financial indicators. The chosen indicators are two: total population in the area in which each company is licensed to operate, multiplied by the company share of ownership (POPS) and the penetration rate, i.e. the ratio of subscribers to POPS (PEN). Amir and Lev demonstrate that these indicators increase the informative content of traditional financial measures and put in evidence that non-financial performance indicators are industry-specific. Behn and Riley (1999) move along the same route, showing that, in the U.S. airline industry, the non-financial indicators, which are related to customer satisfaction, are value relevant. In their research, the two authors use three different regression models. Through the first model, the relevance of a synthetic customer satisfaction indicator is proved. In the other two models a few financial measures (operating income, operating revenues, operating expenses) are used as dependent variables to throw into relief their “contemporaneous” and predictive relationship with the 5
Dainelli F. - Giunta F.
synthetic customer satisfaction indicator and other non-financial indicators (load factor, market share, available ton miles). Behn and Riley’s results show a negative and statistically significant relationship between financial and non-financial measures. In particular, non-financial performance information appears to be useful in predicting quarterly revenues, expenses, and operating income. Further research (Liedtka, 2002; Riley et al., 2003) has confirmed the long-term validity of these results. Riley et al. (2003) resort to a model in which stock returns (the dependent variable) are related to two financial indicators (EPS and change in abnormal EPS), and four non-financial indicators (customer complaints, load factor, market share, available ton miles). Comparing the “full” model results (where both financial and nonfinancial regressors are used) with two other partial model results (where the two regressor categories –financial and non-financial – are separately used), the authors succeed in demonstrating that financial and non-financial indicators are complementary. In 1998, Ittner and Larcker (1998) corroborated the negative and statistically significant relationship between customer satisfaction indicators and company financial performances. Their research concerns the telecommunications industry, and is conducted processing customer, business-unit, and firm-level data. Already in 1996, Lev and Sougiannis (1996) had shown the value relevance of R&D expenses. Disagreeing with SFAS n.2 statement, these expenses are considered an important signal of a company’s investment effort in intangible assets that are conducive to increasing the future company’s revenues. The authors “adjust” the reported earnings and book values of sample firms for the R&D capitalization and find that such adjustments are value-relevant to investors. In the paper, different models are applied, which relate net income, stock prices and returns (dependent variables) to measures of investments in intangibles (R&D expenses, annual advertising expenses, adjusted book value etc.). Connoly and Hirschey (1988) and Hirschey et al. (2001) have also come to the same conclusion, using non-financial measures that can be applied to hi-tech industries. In particular, Hirschey et al. (2001) base their conclusion on a regression model where the dependent variable is represented by the normalized market value of common equity (which is calculated dividing the discounted present value of all future earnings at an appropriate riskadjusted normal rate of return by the book value of common equity), and the independent variables are financial and non-financial (earnings before R&D expenses, R&D expenses, number of patents granted, citations-based 6
The value relevance of non-financial performance indicators
indicators of scientific merit, technology cycle time). Recently, Coram (2010) has shown that the non-financial performance indicator value relevance is related to the user: non professional users have a proclivity to underrate this kind of information. At the origin of the above mentioned studies, one can find a common general research hypothesis: the non-financial indicators are value relevant. From a methodological point of view, to test this hypothesis most of the research resort to linear regression models. The applied models have a few contact points, but also some differences. As regards the contact points, the dependent variable choice is usually represented by market-based variables like stock returns, as these studies are association studies or marginal information content studies3. The stock return calculation method is also well established in literature (Ali et al., 2000; Amir and Lev, 1996; Healy and Wahlen, 1998; Riley et al., 2003): the change in the adjusted4 stock returns is scaled by the stock market price at the beginning of the analysis period; from this return is subtracted the equally weighted index return that encompasses the stocks of all the companies which are part of the scrutinized financial market.5 As regards the main differences, these deal with the use of per share indicators. If we consider the financial independent variables, researchers 3
A well-known systematics of value relevance studies (Holthausen and Watts, 2001) pinpoints three different approaches: • relative association studies, that “compare the association between stock market values (or changes in values) and alternative bottom-line measures” (Holthausen and Watts, 2001, p. 5) by means of the R2 index comparison; • incremental association studies, that “investigate whether the accounting number of interest is helpful in explaining value or returns (over long windows) given other specified variables. That accounting number is typically deemed to be value relevant if its estimated regression coefficient is significantly different from zero” (Holthausen and Watts 2001, p. 6). • marginal information content studies, that “investigate whether a particular accounting number adds to the information set available to investors. They typically use event studies (short window return studies) to determine if the release of an accounting number (conditional on other information released) is associated with value changes. Price reactions are considered evidence of value relevance” (Holthausen and Watts 2001, p. 6). 4 The adjusted stock prices do not take into account the stock split effect (which causes an advance in stock prices) and the payment of dividends effect (which causes a decline in stock prices). 5 For this purpose, most research that concerns US companies uses the CRSP index (Center for Research in Security Prices – for more details, see: http://www.crsp.com/products/stocks.htm).
7
Dainelli F. - Giunta F.
agree upon the EPS use when the dependent variable is represented by stock returns, even if there are some variations. For example, Amir and Lev (1996) use cash flow measures. On the contrary, if the dependent variable is represented by stock prices, the book value per share measure is frequently used (Amir and Lev 1996). Other models incorporate the abnormal earnings per share measure – even if the calculation method of this indicator is not unambiguous –, which represents the portion of a stock’s realized return that is different from its expected return (given existing market conditions) (Riley et al., 2003). We have to notice that, very often, the financial independent variables and sometimes the non-financial variables (Hirschey et al., 2001; Riley et al., 2003) are scaled by a common factor which usually is the stock price. However, there are some discordances among the authors in the interpretation of these results. On the one hand, the most used method to show the value relevance of an indicator is based on the indicator’s statistical significance, which is measured by means of p-value analysis, and Student’s t distribution, carrying out either a one-tailed test or a twotailed test. On the other hand, a few researchers prefer to compare the R2 index that results from the application of different regression models (Riley et al., 2003). To deduce the complementarity between financial and nonfinancial indicators, Riley et al. (2003) consider the R2 differences between the models which incorporate only one kind of independent variable (financial or non-financial) and the “full” model in which both indicator categories are present.
3. Non-financial indicators in the fashion industry and the research hypothesis The value relevance studies on non-financial indicators are focused on industries with high potential intangible assets. To this regard, beginning in the 1990s, the fashion industry has been radically transformed. That revolution was determined by the import of low-cost textiles from other countries. Firms of more developed countries reacted by implementing a high differentiation strategy, based on creativity, design, and recognizability of their brand. Management attention, therefore, moved from process efficiency to the customer value. As a consequence, the main value drivers are no longer represented by tangible assets, but by intangible assets, linked to product research, design, and commercialization (Altan, 2004; Hand and Lev, 2003). The intangible assets in the fashion industry are identified by 8
The value relevance of non-financial performance indicators
Modina (2004) in the (potential and current) brand value, which is only partially recognized by accounting numbers and the drivers of which are based on customer loyalty, store localization, and product style. The relevance of this shifting from tangible to intangible assets is usually expressed by the divergence between accounting and market values. To this regard, European fashion companies registered an average market-to-book ratio around 3 between 2004 and 2008 (Table 1). Table 1 – Market-to-book ratio for European fashion companies Shareholders’ Equity (mean)
Market capitalization (mean)
Market-to-book (mean)
2004
1.685.060
4.705.572
2.79
2005
1.595.237
5.031.381
3.15
2006
1.662.504
6.172.457
3.71
2007
1.829.458
6.923.777
3.78
2008
1.959.014
4.783.491
2.44
Barth et al. (1998) provide more evidence concerning the capability of market prices to include intangible resources in the industry we analyze, finding a positive relation between the brand value and the market value. The same conclusion is drawn by Oliveira et al. (2010) who examine other kinds of intangibles, and by Seethamraju (2000) who highlights a relevant investor reaction when listed companies announce a brand acquisition. Looking at the annual reports of fashion companies, the “change in mono-brand stores” is a widespread indicator. In the European fashion industry, 17 out of 50 firms published that indicator in 2002, while this proportion reaches 38 out of 50 in 2008. The non-financial indicator we have searched for is more and more present also in analysts’ reports. We have scrutinized all the reports of the analysts that cover Italian fashion listed companies and we have found that 46% of them indicate the “change in mono-brand stores” as a key driver. If we average it out, the indicator at issue is used at least by one of the financial analysts that cover the company. The attention towards this non-financial measure is due to its capability to synthesize several factors. First, the differentiation strategy has induced firms to directly commercialize their products using mono-brand stores. This category includes DOS (Directly Operating Stores)6, franchise shops, 6
These stores are directly managed by the firm and they consist of: flagship store (large
9
Dainelli F. - Giunta F.
and mono-brand outlets. Therefore, direct distribution has today become a strategic factor to develop brand identity and corporate vision (Altan, 2004; Modina, 2004). Second, mono-brand stores also express other internal intangible factors, such as human resources and the organization’s structure that manage those retails. Since market prices incorporate future earnings expectations (Rappaport and Mauboussin 2001), the opening of new mono-brand stores seems a good driver of the growth capability of a firm, which, in turn, can be seen as a multiplier of future earnings. It is from these considerations that we can draw the research hypothesis that is at the basis of our study: the change in mono-brand stores is a value relevant indicator in the fashion industry. In concrete, this hypothesis is a specification of a more general research assumption on which a large part of the literature, concerning the value relevance of non-financial measures, is based. Specifically for this reason, our hypothesis is tested using the methods proposed by that literature. This choice permits us, on the one hand, to verify the generalizability of the studies on the value relevance of non-financial measures and, on the other hand, to identify a predictive indicator for the fashion industry to add to the indicators revealed in other studies, in order to develop a set of standard measures. The choice of the fashion industry is motivated also by its importance in the world economy. In fact, fashion product transactions represent 5,7% of global export (342 billion of dollars in 2001) and they have increased 60 times since the 1970s (Labory and Zanni, 2002).
4. Dataset Our hypothesis has been tested on a sample of firms that operate in the fashion & apparel industry. It consists of different clusters, with different weight and importance, strictly interlinked. Thus, it is impossible to define its borders. For this study, we have selected companies that belong to the clothing industry (and accessories), with a strong brand loyalty, capable of incorporating higher intangible values (Kotler, 2004). The identification of the sample firms meets the following three store in VIP locations), self-standing store (medium store) e shop in shop (small shop located in distribution chain).
10
The value relevance of non-financial performance indicators
conditions: • Companies listed on developed financial markets, which ensure higher investor capacity to evaluate intangible assets. Thus, our attention is focused on: the United Kingdom, Germany, France, Italy, Spain, Denmark, Sweden, Belgium, Netherlands, Switzerland7. • Companies belonging to the following industry sectors: a) retail (sub-sectors apparel and broadline retail), b) consumer goods (sub-sectors clothing/accessories and footwear); • Companies included in the “Luxury & lifestyle” Merryl Lynch index that groups all the globalized firms operating in the Luxury & lifestyle segment, with a medium-high market capitalization, listed in developed markets (53% Europe, 38% USA, 9% others). All these factors supposedly guarantee the international recognizability of a brand and its valorization in the stock price. The application of these selection criteria has generated a list of 97 firms. Nevertheless, a further one by one screening procedure has been necessary to exclude firms commercializing products different from clothing8. Starting from 2000 until 2008, we have manually scrutinized the annual reports of the sample firms, searching for “mono-brand stores”. We stop in 2008, assuming that the international crisis has since significantly influenced stock prices. Moreover, we have excluded from our sample those companies that do not publish the indicator for a consecutive fouryear series, in order to have a minimum time span to run a time series analysis. In the end, our dataset counts 30 companies and 156 observations (Table 2). Table 2 – Sample firms by country ITALY
UK
GERMANY
FRANCE
Mariella Burani
French Connection
Wolford
Christian Dior
CSP International
Burberry Group
Gerry Weber
Hermes
7 For multi-listed companies, we have chosen the market where the headoffices are located. 8 For instance, the Shieldtech PLC, company listed on London Stock Exchange has been excluded because produces and sells bulletproof vests.
11
Dainelli F. - Giunta F.
Geox
Next
Adidas
LVHM
Tod's
JD sports fashion
EDOB Abwicklungs
Camaieu
Luxottica
JJB sports
VET Affaires
Ted Baker
Etam developpement
Debenhams Marks&Spencer SPAIN Inditex
DENMARK IC Companys
SWEDEN
SWITZERLAND
H&M
Richemont
Fenix outdoor
Charles Vogele
Bjorn Borg
5. Testing the hypothesis on the literature models Our value relevance hypothesis has been tested on the dataset described in the previous paragraph, using the main models proposed in literature. In the following sections, the discussion of the models (5.1), and the related results (5.2) are presented.
5.1 The literature models chosen There are two main different models in recent literature: a) Amir and Lev (1996); b) Riley et al., (2003). The first model (Amir and Lev, 1996) is structured as follows: P = β0 + β1EPS + β2BVPS + β3∆MONO + β4COUNTRY + ε where: P (stock prices) is the average price registered in the four-month window surrounding the publication date of the annual report. EPS is calculated as net profit before extraordinary items, divided by the average number of outstanding common shares. BVPS (book value per share) is the shareholder’s equity (book value) divided by the average number of outstanding common shares. 12
The value relevance of non-financial performance indicators
ΔMONO represents the “change in mono-brand stores”. COUNTRY is the control variable that synthetizes the environmental differences. This is not present in the original model because it was developed on a one-country dataset. This variable is expressed as the average return of the stock market of each company’s origin country. The model elaborated by Riley et al. (2003) is structured as follows: R = β0 + β1EPS/p + β2ABEPS/p + β3∆MONO/p + β4COUNTRY + ε where: R (return) is the return that expresses the stock performance better than the price, because it takes into account the dividend effect. R is computed as the period (quarterly) change in stock price plus dividends. The return of the “MSCI Europe Equal weighted” index for the same period is subtracted from the calculated return in order to isolate the intrinsic variability of the stock.9 EPS/p is the first independent financial variable. It is calculated as net profit before extraordinary items divided by the average number of outstanding common shares. Earnings per share is scaled by beginning of the period stock price. ABEPS/p (abnormal earnings per share) is the second independent financial variable. It is determined by subtracting from EPS the beginning of quarter book value per share multiplied by a discount rate of 13%. The change in ABEPS is scaled by beginning of the period stock price. ΔMONO/p is the non-financial variable. In our case, it is represented by the “change in mono-brand stores”. This variable is also scaled by beginning of the period stock price. COUNTRY is the control variable that synthetizes environmental differences. This is not present in the original model because it was developed on a one-country dataset. This variable is expressed as the average return of the stock market of each company’s origin country. Table 3 shows the sources of data of all variables. Table 3 – The sources of data VARIABLE
COMPONENTS
SOURCE
9
NOTES
Riley et al. (2003), as most USA literature, use the “equal weighted CRSP” index to “normalize” stock returns. In Europe, we believe that the “MSCI Equal Weighted” is the most similar index.
13
Dainelli F. - Giunta F.
Δ MONO
Number of monobrand stores
Consolidated annual reports
RETURN
Stock price*
Thomson Financial
Dividend*
Consolidated annual reports
Discount rate returns
MSCI-Barra
Net profit*
Consolidated annual reports
Average number of outstanding common shares
Consolidated annual reports
EPS
“Yahoo finance”. It is the non-adjusted daily closing
We use/calculate the annual average number of shares
Book value per share* Consolidated annual reports
AEPS
Risk-free rate COUNTRY
Eurostat
Country stock market Thomson Financial returns
Decennial BUND returns Yahoo finance
* Values not expressed in Euro have been converted using the exchange rate of the related day.
The stock prices used in the model represent the market reaction that follows the non-financial information disclosure. It is plausible to assume that the annual report information is acquired by more sophisticated operators in the period that goes from the closing date of the annual report to its publication. Other operators are supposed to get acquainted with that information within two weeks after the publication of the annual report. Our stock prices, therefore, are represented by the average of daily closings within the window that goes from the end of the year until fifteen days after the publication of the annual report. 10
5.2 Results obtained with the literature models
10
Choosing this time window, our study can be defined as an incremental association study. These studies consider quite long time spans and do not emphasize the importance of timeliness of information communication, so that the annual report is deemed a significant source of information for the market (Healy and Palepu, 2001; Holthausen and Watts, 2001).
14
The value relevance of non-financial performance indicators
The results of the Amir-Lev’s model are shown in Table 4, which highlights the differences between the two partial models – one of which is only based on financial variables (FIN), while the other one only on nonfinancial variables (NONFIN) – and the full model (FULL) which encompasses all the variables. The Riley et al. model results are presented in Table 5. Table 4 - Results of the Amir-Lev’s model Variable Constant
Coefficient (std. error) FIN
NONFIN
FULL
8.653***
27.625***
6.663***
(2.053)
(3.412)
(2.255)
18.330
-0.064
(11.105)
(6.117)
Δ MONO EPS BVPS COUNTRY 2
R adjusted
10.631***
10.633***
(0.707)
(0.731)
0.228*
0.227*
(0.119)
(0.121)
14.385**
15.444
14.377**
(6.327)
(11.894)
(6.382)
0.720
0.013
0.718
133.71*** 2.030 F statistics ***, **, * indicate significance at 0.01, 0.05 and 0.10 levels.
99.63***
Table 5 - Results of the Riley et al. model Variable Constant
Coefficient (std. error) FIN
NONFIN
FULL
-0.124**
0.054
-0.122**
(0.053)
(0.045)
(0.055)
0.002
-0.0006
(0.002)
(0.0025)
Δ MONO/p EPS/p ABEPS/p
3.039***
3.06***
(0.555)
(0.563)
-0.056
-0.049
15
Dainelli F. - Giunta F.
(0.088) COUNTRY 2
R adjusted
(0.0025)
0.414**
0.578***
0.415**
(0.166)
(0.175)
(0.167)
0.224
0.067
0.218
15.684*** 6.607*** F statistics ***, **, * indicate significance at 0.01, 0.05 and 0.10 levels.
11.707***
The results of the two models demonstrate the statistic irrelevance of the ΔMONO variable, both in the FULL version and in the NONFIN version. This fact does not support the value relevance hypothesis of the nonfinancial indicator we have chosen. Moreover, it seems that the non-financial data does not increase the predictive capacity of the financial measures, as stated by Amir-Lev. In fact, the coefficients of EPS/p and ABEPS/p do not significantly vary between the FULL model and the FIN model. Furthermore, it has to be noted that, in the Amir-Lev model, the “EPS” variable is extremely significant both in the FIN and FULL model FULL (p-value respectively of 1.80e-07 and 2.19e-07), with positive Beta coefficients, as expected. The dummy variable (COUNTRY) is also statistical significant (p-value 0.0141 for the FIN and FULL model and 0.0012 for the NONFIN model) and positively associated with the stock returns. The same conclusion can be drawn from the Riley et al model. It demonstrates the value relevance of both financial variables (EPS e BVPS), with a good fitting. These results stress the primary role of the traditional accounting measures (earnings and equity), also in industry segments with high market-to-book value and intangible assets. A first possible explanation of a lack of value relevance of our selected non-financial variable is that corporates and analysts are wrongly oriented towards an indicator which is scarcely predictive of stock market performance. Nevertheless, a second explanation can be linked to a few limitations of the models elaborated by the literature, which reduce their level of generalizability.
6. Results of a refined model Considering the possible limitations of the models proposed in the 16
The value relevance of non-financial performance indicators
literature, we have tried to develop a new regression model. This is a crosssection model, which works with a dataset made of time series of different length (from a minimum of four years to a maximum of eight years). The model we use is based on stock returns instead of stock prices (Olson, 1995) and the market and financial variables employed are the same as Riley’s model. In particular, as we have run a stock return model, we have employed the ABEPS, a “flow” value, instead of the BVPS, which is a “stock” value. Riley’s model, however, also suffers from two conceptual limitations. The first one is that the increase/decrease in the stock price (return) depends on the absolute value of the financial variables (EPS and ABEPS). Considering that the returns represent changes in price, it is more logical to assume that they depend on a change in the financial variables11. In fact, Riley also reasons in terms of “change” in the non-financial variables included in the same model. Moreover, the choice to scale all the variables including the nonfinancial ones, for a single constant (the beginning of period stock price), is only justified in order to reduce the heteroscedasticity (Lev and Sougiannis, 1996; Kothari and Zimmerman, 1995; Barth et al., 1992). To this regard, however, Brown et al. (1999) have highlighted the existence of a “scale effect” for the indicators scaled by stock market bases (per share indicators). As a matter of fact, everything depends on the number of shares by which the equity is divided. The stock price of those companies where that number is high is obviously lower than the stock price of those companies where that price is lower. In this respect, Brown et al. (1999) report that the Berkshire Hathaway stock price was 45,000 dollars in 1997, while the price of IBM was around 100 dollars in the same year. As a consequence, the model we propose is the following: R = β0 + β1∆EPS + β2∆ABEPS + β3∆MONO + β4COUNTRY + ε where: R (return) is the stock return. Riley’s model is based on a compound interest rule. In accordance with the recent trends of finance research, returns are calculated on a logarithmic scale, adopting a continuously compounded return:
11
This critique is elaborated by Lambert (1998) in discussing Ittner and Larker (1998).
17
Dainelli F. - Giunta F.
R = ln[(Pt + dt) / Pt-1] - MSCI ΔABEPS. A limitation of Riley’s model is the generalized application of a normal discount rate of 13%, without defining the data source and motivating this choice. We have calculated a normal discount rate for each firm: Rn = Rfreet + βt i * (Rmt - Rfreet) The risk free rate (Rfree) is the decennial return of Bund, the Beta is a levered measure to take the global risk of a company into account, and the market return (Rm) is based on the average return of the European fashion industry (Source: Yahoo Finance - Thomson Financial). The following tables (Tables 6 and 7) show the descriptive statistics and the correlation matrix of the variables of our model. Table 6 - Descriptive statistics (n=156) VARIABLE
MEAN
STD. DEV.
MEDIAN
MIN
MAX
Return (R)
-0.05487
0.45765
-0.015
-2.46
1.50
EPS
-0.15513
1.775
0.11
-15.4
3.51
ABEPS
0.18383
7.972
-0.635
-30.51
54.33
MONO
0.15397
0.26641
0.10
-0.38
2.25
COUNTRY
0.01410
0.24881
0.10
-0.47
0.57
Table 7 - Correlation matrix (n=156) RETURN
Δ EPS
Δ ABEPS Δ MONO COUNTRY
1.0000
0.5263
0.0895
0.2864
0.3270
RETURN
1.0000
0.0975
0.1434
0.1654
Δ EPS
1.0000
0.0825
-0.0860
Δ ABEPS
1.0000
-0.0890
Δ MONO
1.0000
COUNTRY
The results of the regression analysis are presented in Table 8. Table 8 - Results of the regression test (n=156) Variable
Coefficient (std. error)
18
The value relevance of non-financial performance indicators
Constant
-0.1251 *** (0.0357)
Δ EPS
0.1109 *** (0.0169)
Δ ABEPS
0.0025 (0.0036)
COUNTRY
0.5204 *** (0.1194)
Δ MONO
0.5319 *** (0.1437)
2
R adjusted
0.38
24.81*** F statistics ***, **, * indicate significance at the 0.01, 0.05 and 0.10 levels.
The result of the Riley et al. analysis is affected by another limitation. The authors compare the R2 indices deriving from different regression models (FIN; NONFIN, FULL) instead of evaluating the p-value (Wooldridge, 2010). Looking at the p-value, our results show, as might be expected, that the ΔEPS financial variable is positively associated (p-value < 0.00001) with the stock returns, but it is not associated with the change in abnormal earnings. This evidence demonstrates the considerable importance of the earnings for the market operators in the fashion industry in Europe. In our model, the non-financial variable (ΔMONO) becomes extremely significant (p-value = 0.0003) and positively associated with the returns. The goodness of fit is motivated by an adjusted 0.38 R2 and by a highly significant F-statistics (p-value < 0.00001). Fixed effects of standard errors are White’s heteroscedastic consistent estimates. Moreover, the variance inflation factor is about 1 for each regressor we have used, and this fact brings us to reject the presence of multicollinearity among regressors (Table 9). Table 9 - Variance inflation factor (VIF) VARIABLE Δ EPS Δ AEPS Δ MONO COUNTRY
VIF 1,067 1.024 1.038 1.052
19
Dainelli F. - Giunta F.
Even though the normality of the distribution is not demonstrated, the model can be considered valid due to the large number of observations in the sample (n=156).
7. Sensitivity analysis In our sample, the high number of observations is one of the reasons that spurred us to choose a cross-section model instead of a panel data model. Nevertheless, a panel data model has been applied to a sample of 30 companies for the 2004-2008 period (n=90), excluding the years before 2004. The results are similar to those presented in the sixth section. The high values of the coefficient R2 (0.41) and the F-statistics (p-value < 0.0001) testify to the validity of our model. We reach analogous results on 20 firms for the 2002-2008 period (n=100): ΔEPS, COUNTRY and ΔMONO are positively and significantly associated with the returns. The model registers a 0.41 R2 and an F-statistics equal to 19.89 (p-value < 0.0001). In both of the panel data regressions, the homoscedasticity assumption of residuals (with the White test) is confirmed, as well as the absence of multicollinearity and the normality of statistic errors. We have also run our regression test delaying our non-financial variable one year, following the practice of certain literature (Amir and Lev, 1996). The results of these lagged regressions show the irrelevance of the “change in mono-brand stores” indicator. Nevertheless, in our case, the use of lagged variables appears not to be suitable and it weakens the analysis. Our dataset, in fact, is based on annual observations instead of four-month (or shorter period) observations as the above mentioned literature does. Thus, it is not plausible that the market includes the non-financial information in the stock prices one year later. The sensitivity analysis shows that the BVPS (book value per share) indicator is not capable of adding predictive power to the independent variables. The BVPS is significant (at 0.10 level) only when the dependent variable is represented by prices instead of returns. When the BVPS indicator is used in the model, however, the non-financial variable is no longer statistically significant. Finally, the sensitivity analysis has tested the use of the independent variables calculated as absolute change instead of relative change. Also in this case, the results are the same: ΔEPS, COUNTRY and ΔMONO (at 20
The value relevance of non-financial performance indicators
0.01 level) are very significant and positively associated with the market returns; ΔABEPS is not significant; the model presents a good fitting (R2 = 0.38, F-statistics = 24.89 with a p-value < 0.0001).
8. Conclusions In times in which the competitive strength of companies is more and more grounded on intangible assets, the usual financial accounting metric has gradually been tapering off its importance in favor of non-financial indicators. This fact is confirmed by many value relevance studies that predicate the ability of non-financial indicators to capture and also determine the market trend of a stock. In the light of this, our study follows this research avenue, and examines the value relevance of a well-known non-financial indicator which is used in the European fashion industry. In the fashion industry, after the entrance of low cost competitors in the market, the value creation is based on several intangible assets. The “change in mono-brand stores” (ΔMONO) is a non-financial indicator that synthesizes the effects of a few of these assets as it depends on the human, relational and organizational resources of a company. It is not by chance that this indicator is largely disclosed in the companies’ financial statements and it is the most used by financial analysts. It is logical to hypothesize, therefore, that this measure can be value relevant, i.e. it can determine the company's stock prices when the market gets acquainted with it. The above mentioned indicator has been collected from the European listed fashion companies’ financial statements over the period from 2000 to 2008, with a minimum of four consecutive financial years. After being skimmed, our sample consists of 30 companies and 156 observations. Applying the most well-known models worked out by literature to this sample, the hypothesis of the value relevance of “change in mono-brand stores” indicator cannot be supported. This result is peculiar if we take into account that: the examined companies belong to an industry with an evident high intangibles content (testified by the large gap between the stock book and market value); the non-financial variable is an indicator which the main financial market actors, i.e. analysts and companies, seem to consider the most significant; the dataset, to which we have applied the analysis models of the literature, is quite ample. For this reason, we have critically revised the models we have referred 21
Dainelli F. - Giunta F.
to. Refining the “operationalization” of the concepts which these models are based on, we managed to elaborate a new model that produces different results, and confirms the value relevance of the selected non-financial indicator for the fashion industry. To adapt the above mentioned models, we have: a) calculated the stock returns using the continuously compounding capitalization method; b) determined the regressors as relative variations, instead of using absolute variations; c) not scaled the variables by a constant; d) computed the specific normal stock return of each company, as a basis to determine the abnormal earnings per share (ABEPS). Moreover, to interpret the final results, we have considered the p-value rather than the change in R2 index. In this way, we have tried to contribute to increasing the “external validity” of the research results in the field of the non-financial indicator value relevance. The general conclusion that we can draw from our results confirms the importance of non-financial indicator disclosure and communication. The hypothesis that these measures – at least the ones which are the key-measures of an industry – have a great informative content for investors can be confirmed. As a consequence, the lack of communication of these indicators causes significant information asymmetries which lower market efficiency. Consequently, the ongoing regulation process concerning the non-financial information disclosure is moving along the right track. Nevertheless, even if the “change in mono-brand stores” indicator seems to be an effective synthesis of a fashion company’s value drivers, we have to admit that other variables, with the same or a superior predictive capacity, exist. In addition to this, in order to validate our results, we could test the value relevance of the “change in mono-brand stores” indicator, considering a different measuring period (window) for market performance, i.e. using shorter periods (three/four-month returns instead of annual returns) or looking at different information announcement events (the publication of the analysts’ report instead of the company’s annual report). Finally, to confirm and reinforce the predictive capacity of the model that we have developed, our model could be applied to the datasets that were used in the models of the existing literature which we have drawn on. Unfortunately, the datasets which are used in that literature come from sources that are difficult to access. The future developments of our research could test our model validity applying it to other industries. Following in the footsteps of the Gartner/EBRC project, supported by AICPA (Gartner and EBRC, 2010), 22
The value relevance of non-financial performance indicators
the use of different non-financial indicators could help the national and international standard setters to pinpoint the measures that really matter for different industries and standardize their disclosure.
References AICPA (1994). Improving business reporting. A customer focus, American Institute od Certified Public Accountants, New York. Ali A., Hwang L.S., Trombley M.A (2001). Residual-Income-Based Valuation predicts future stock returns: Evidence on mispricing vs. risk explanations, The Accounting Review, 78, 2: 377-396. Altan E.C. (2004). La moda allo specchio. Comunicare la moda: strategie e professioni, Milano: FrancoAngeli. Amir E., Lev B. (1996). Value relevance of nonfinancial information: the wireless communications industry, Journal of Accounting and Economics, 22, 1-3: 3-30. Barth M.E., Beaver W.H., Landsman W.R. (1992). The market valuation implications of net periodic pension cost components, Journal of Accounting and Economics, 15, 2-62. Barth M.E., Beaver W.H., Landsman W.R. (2001). The relevance of the value relevance literature for financial accounting standard setting: another view, Journal of Accounting and Economics, 31, 77-104. Barth M.E., Clement M.B., Foster G., Kasznik R. (1998). Brand values and capital market valuation, Review of Accounting Studies, 3, 1-2: 41-68. Behn B.K., Riley R.A. (1999). Using nonfinancial information to predict financial performance: the case of the U.S. airline industry, Journal of Accounting, Auditing & Finance, 14, 1: 29-56. Birchard B. (1994). Mastering the new metrics, CFO: The Magazine for Chief Financial Officers, 10, 10: 30-38. Breton G., Taffler R. (2001). Accounting information and analyst stock recommendation decisions: a content analysis approach, Accounting and Business Research, 31, 2: 91-101. Brown S., Lo K., Lys T.Z. (1999). Use of R2 in accounting research: measuring changes in value relevance over the last four decades, Journal of Accounting and Economics, 28, 83115. Chan L., Karceski J., Lakonishok J. (2003). The level and persistence of growth rates, Journal of Finance, 58, 2: 643-684. Collins W.A., Maydew E., Weiss I.S., (1997), Changes in the value-relevance of earnings and book values over the past forty years, Journal of Accounting and Economics, 24, 1: 39-67. Connoly R.A., Hirschey M. (1988). Market value and patents: a bayesian approach, Economics Letters, 27, 1: 83-87. Coram P.J. (2010). The effect of investor sophistication on the influence of nonfinancial performance indicators on investors' judgment, Accounting & Finance, 50, 263-280. Edvinsson L., Malone M.S. (1997). Intellectual Capital, London: Piatckus. Eustace C. (2000). The intangible economy impact and policy issued: report of the european high level expert group on the intangible economy, European Commission, October. Francis J., Schipper K. (1999). Have financial statements lost their relevance?, Journal
23
Dainelli F. - Giunta F.
of Accounting Research, 37, 2: 319-352. Gartner-EBRC (2010), available at http://www.wici-global.com/kpi.php, 20-9-2011. Hand J., Lev B. (2003). Intangible Assets: Values, Measures and Risks, Oxford: Oxford University Press. Healy P.M., Palepu K.G. (2001). Information Asymmetry, Corporate Disclosure, and the Capital Markets: A Review of the Empirical Disclosure Literature, Journal of Accounting & Economics, 31, 1-3: 405-440. Healy P.M., Wahlen J.M. (1999). A review of the earnings management literature and its implications for standard setting, Accounting Horizons, 13, 4: 365-383. Hirschey M., Richardson V.J., Scholz S. (2001). Value relevance of nonfinancial information: the case of patent data, Review of Quantitative Finance and Accounting, 17, 223-235. Hofmann J. (2008). How intellectual capital creates value: towards the strategic management of intangibles, Frankfurt am Main: Deutsche Bank Research. Holthausen R.W., Watts R.L. (2001). The relevance of the value-relevance literature for financial accounting standard setting, Journal of Accounting and Economics, 31, 3-75. ICAEW (2009). Developments In New Reporting Models, London: ICAEW. Ittner C.D., Larcker D.F. (1998). Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction, Journal of Accounting Research, 36, 1-35. Kaplan R.S., Norton D.P. (1992). The balanced scorecard: measures that drive performance, Harvard Business Review, (January-February): 71-79. Kothari S.P., Zimmerman J.L. (1995). Price and return models, Journal of Accounting and Economics, 20, 2: 155-192. Kotler P. (2004). Marketing management, Milano: Pearson Education Italia. Labory S., Zanni L. (2002). Il sistema moda in Toscana, Firenze: IRPET. Lambert R.A. (1998). Customer satisfaction and future financial performance. Discussion of: 'Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction', Journal of Accounting Research, 36, 37-46. Lev B., (2001). Intangibles Management, Measurement, and Reporting rules: a strategic guide to the network economy, Washington, D.C.: Brookings Institution Press. Lev B., Sougiannis D.F. (1996). The capitalization, amortization and value-relevance of R&D, Journal of Accounting and Economics, 21, 107-137. Lev B., Thiagarajan S.R. (1993), Fundamental information analysis, Journal of Accounting Research, 31, 2: 190-215. Lev B., Zambon S. (2003). Intangibles and intellectual capital: an introduction to a special issue, European Accounting Review, 12, 4: 597-603. Lev B., Zarowin P. (1999). The boundaries of financial reporting and how to extend them, Journal of Accounting Research, 37, 2: 353-386. Liedtka S.L. (2002). The information content of nonfinancial performance measures in the airline industry, Journal of Business Finance & Accounting, 29, 7: 1105-1121. Modina S. (2004). Il business della moda, Milano: FrancoAngeli. Ohlson J. (1995). Earnings, book values and dividends in security valuation, Contemporary Accounting Research, 11, 2: 661-687. Oliveira L., Rodrigues L.L., Craig R. (2010). Intangible assets and value relevance: Evidence from the Portuguese stock exchange, The British Accounting Review, 42, 241–252. Previts G., Bricker R., Robinson T., Young S. (1994). A content analysis of sell-side financial analyst company reports, Accounting Horizon, 8, 2: 55-70. Rappaport A., Mauboussin M.J. (2001). Expectations investing: reading stock prices for
24
The value relevance of non-financial performance indicators
better returns, Boston: Harvard Business School Press. Riley R.A., Pearson T.A., Trompeter G. (2003). The value relevance of non-financial performance variables and accounting information: the case of the airline industry, Journal of Accounting and Public Policy, 22, 3: 231-254. Rogers R.K., Grant J. (1997). Content analysis of information cited in reports of sellside financial analysts, Journal of Financial Statement Analysis, 3, 1: 17-31. Seethamraju C. (2000). The value relevance of trademarks, Ph.D. dissertation, New York University. Shapiro C., Varian H.R. (1999). Information rules: a strategic guide to the network economy, Boston: Harvard Business School Press. Skinner D.J. (2008). Accounting for intangibles: a critical review of policy recommendations, Accounting and Business Research, 38, 3: 191-204. Stewart T. (1997). Il capitale intellettuale: la nuova ricchezza, Milano: Ponte alle Grazie. Sveiby K.E. (1997). The intangible assets monitor, Journal of Human Resource Costing & Accounting, 2, 1: 73-97. Taub S. (2003). MVP's of MVA: Measuring how much market value companies created, CFO: The Magazine for Chief Financial Officers, 19, 9: 59-66. Wallman S.M.H (1995). The future of accounting and disclosure in an evolving world: the need for dramatic change, Accounting Horizons, 9, 3: 81-91. Wooldridge M.J. (2010). Econometric Analysis of Cross Section and Panel Data, 2nd Edition, Cambridge: The MIT Press. Zambon S. (2003). Study on the measurement of intangible assets and associated reporting practices, prepared for the Commission of the European Communities Enterprise Directorate General, April 2003. Zambon S. (2003). Study on the measurement of intangible assets and associated reporting practices, prepared for the Commission of the European Communities Enterprise Directorate General, April 2003.
25