The Value Relevance of Intangible Assets in Korean Firms. Hoejun Min

Transcription

1 The Value Relevance of Intangible Assets in Korean Firms Hoejun Min 29th May, 2012

2 The Value Relevance of Intangible Assets in Korean Firms Master thesis Department Accountancy Faculty of Economics and Business Studies Tilburg University Supervised by Yachang Zeng ANR Hoejun Min Completed on 29th May, 2012 Date of Graduation: 12th June, 2012

3 Summary Despite the increasing importance and role of intangible assets in today s economic environment, accounting practice for intangible assets is one of the most controversial issues in that they are not fully reflected in financial statements. As an attempt to explore the relevance of reported intangible assets, this study examines the market value relevance of different intangible asset classes (goodwill, intellectual properties, capitalized development expenditure, capitalized software expenditure and other intangible assets) reported by listed Korean firms during the period of 2001~2010, using the popular Ohlson (1995) framework. Result indicates that some intangible assets have positive associations with the market value of firms, while some others have negative associations. When this analysis is extended to the difference between industries, some intangible assets are found to be more value relevant for information technology firms and some to be less value relevant. 1. Introduction As the structure of modern economy rapidly changes with the unprecedented technological progress in most industries, the main source of corporate value is also changing from classical tangible assets such as land or capital to intangible assets based on human knowledge and ideas. Today intangible assets are generally recognized as the primary resource which strengthen the firm s competitive position and ensure its future viability (Cañibano et al., 2000), and even further, as the most important factors to economic growth and societal wealth (Blair and Wallman, 2001). On the contrary, current accounting standards mainly focus on tangible assets, not sufficiently reflecting the value of intangible assets. The current situation that intangible assets are not adequately demonstrated or quantified in financial statements despite their significant contribution to value creation is referred as the Value Paradox (Blaug and Lekhi, 2009). Considering the inherent uncertainties regarding the future economic benefit and economic life of intangible assets, the difficulties in the valuation (Barth et al., 2001) and proper financial 1

4 reporting of intangible assets are in a sense unavoidable. However, in many industries where firms make significant value-enhancing investments on intangibles, conventional accounting standards may lead to distorted values of earnings and assets, making the financial statements inadequate to their users (Amir and Lev, 1996). With the recognition of these issues, this study examines the value relevance of reported intangible assets during the period of 2001~2010 under Korean setting. The primary environmental factor that justifies an interest in this Korean setting is that the accounting practice for reporting intangible assets was relatively more generous. For example, Korean accounting standard allows capitalization of qualified development expenditure 1 and therefore lots of firms report related item on financial statements, which is not allowed in many accounting settings including US. Korean GAAP also allows capitalization of internally generated software regardless of industry, but this is only limitedly allowed in US. This accounting environment, combined with the data availability of detailed intangible asset class accounts, offers a good opportunity to explore intangible assets. Most prior studies that examined specific intangible assets in other settings (e.g., Aboody and Lev 1998; Kallapur and Kwan 2004; Oliveira et al. 2010) relied on hand-collected data, since global database such as Datastream and Compustat provide only aggregated intangible assets. But this is not the case in Korea, where data of every individual intangible asset class as reported by firms is available via a local database. More specifically, this study tests the hypothesis that different intangible asset classes such as goodwill, intellectual properties, capitalized development expenditure, capitalized software expenditure and other intangible assets would be relevant to firm s market value. In addition, whether the value relevance of intangible assets is different among industries is also tested. In line with the argument of Blair and Wallman (2001) that the explanatory power of accounting information of reported intangible assets for firm value is significantly decreasing in high- 1 Conditions for capitalization are as follows: the technical feasibility of completing the production of the intangible asset, firm s intention to complete and use or sell, firm s ability to use or sell, highly probable future economic benefits, the availability of adequate technical or financial resource for completion, a reliable measurement of expenditure. 2

5 technology firms, the difference in value relevance of reported intangible assets for IT(information technology) firms and non-it firms is examined. Both tests are performed by multiple regression models based on the Ohlson (1995) framework, which is widely used by prior studies. Results from regression models which incorporate individual intangible asset classes separately and 5,242 sample firm-years indicate that all intangible asset classes have significant associations with the market value of the firm. While intellectual properties, capitalized development expenditure and software expenditure showed positive associations, goodwill and other intangible assets showed negative associations. Some of these findings are in consistent with prior studies but others contradict them. Negative association of goodwill may have been resulted from investor s concern for an excess acquisition cost, or the mandatory amortization of goodwill under old standard which may not be adequate for many firms according to the finding of Jennings et al. (1996). For information technology (IT) firms, all intangible asset classes except goodwill showed differences in the level of value relevance when compared to non-it firms, with relatively lower value relevance of capitalized development expenditure and software expenditure and relatively higher value relevance of intellectual properties and other intangible assets. These results may imply investors different perceptions for each intangible asset reported by IT-firms based on the relative objectiveness of its recognition. In other words, investors may have critical attitudes for the future benefits of capitalized development and software expenditures, because recognizing these assets are open to managerial judgement to a large extent. This study would contribute to the literature in that it examines the value relevance of multiple intangible asset classes. As will be outlined later, majority of prior studies focus on one specific item or aggregated total intangible assets. Further, this differentiation of multiple intangible asset classes value relevance is extended to different industries, raising a question about the differences in market value associations among industries. While extensive explorations for the reasons of different value relevance are left to future studies, these findings suggest that investors evaluations for intangible assets do not coincide indeed but on the 3

6 contrary differ substantially. From the viewpoint of the relevance in accounting information, this may leave a room for a consideration of alternative accounting policies either in the standards or in the practical applications of standards. Specifically, if investors critical attitudes for capitalized development and software expenditures in IT firms are resulted from problematic accounting practice in the industry, regulatory actions are needed to improve the relevance of accounting information. This paper is structured as follows. In the following section some brief backgrounds to the definition and classification of intangible assets, along with accounting treatments for intangible assets, are presented. This is followed by a summary of the previous literature that led to the research questions. The research methods, including variable definitions are presented with the research questions, and data collection and the findings follow. Finally, conclusion and limitations of the study are outlined. 2. Accounting for intangibles & prior studies 1) Accounting standards for intangible assets Despite the absence of a universal definition which is resulted from different perspectives of various stakeholders (Ali et al., 2010), intangible assets in accounting context are often defined as assets which lack a physical substance, but result from legal or contractual rights and are likely to produce future benefits (Cañibano et al., 2000). According to International Accounting Standards (IAS) 38 which defines intangible asset as an identifiable non-monetary asset without physical substance, an asset is reported as intangible asset only when it meets the conditions of identifiability, control over a resource and the existence of future economic benefits. If it fails to meet this definition, then related expenditure should be expensed unless part of a business combination, when it should be treated as part of goodwill. As intangible assets differ in their fundamental characteristics, they are often classified into sub-categories. The classification of intangible assets is also a matter in need of further 4

7 agreement. However, several national and international standard setting bodies and institutions have made significant efforts in order to develop a classification of intangibles (Cañibano et al., 2000). For example, Financial Accounting Standards Board (FASB) classifies intangible assets according to related economic activities such as marketing, customer relationship, contraction (FAS 141), while EU (2003) takes a broader perspective by using the class of separately identifiable intangible assets in which many kinds of intangible assets other than intellectual property and goodwill are included. Table 1 shows more details about the classifications of FASB and EU. It is noted that goodwill is often excluded in many classifications (including that of FASB) of Table 1. Classification of intangible assets Classification & specific items (examples) under classes FASB (statement 141) EU (2003) 1. Marketing-related intangible assets Trademarks, Service marks, Trade dress, Newspaper mastheads, Internet domain names, Non-competition agreements 2. Customer-related intangible assets Customer lists, Order or production backlog, Customer contracts and related customer relationships, Non-contractual customer relationships 3. Artistic-related intangible assets Plays, Books, Musical works, Pictures, Video and audiovisual material 4. Contract-based intangible assets Licensing, Advertising, Lease agreements, Construction permits, Franchise agreements, Operating and broadcasting rights, Use rights, Servicing contracts, Employment contracts 5. Technology-based intangible assets Patented technology, Computer software and mask works, Unpatented technology, Database, Trade secrets 1. Intellectual property : Intangible assets with legal or contractual rights including Patents, Trademarks, Designs, Licenses, Copyrights, Film rights, Mastheads 2. Separately identifiable intangible assets Information systems, Networks, Administrative structures and process, Market and technical knowledge, Human capital (if embodied in a codified form), Brands, Intangibles embodied in capital equipment, Trade secrets, Internally generated software, Drawings 3. Goodwill (non-separable intangible assets) Prior intangible investment embodied in organizations, Management expertise, Geographic position, Monopoly market niche 5

8 intangible assets. In this case, goodwill is reported separately on financial statements and distinguished from identifiable intangible assets, in that goodwill is not identifiable. This is indeed the same for current Korean standard which took effect as of with the introduction of the International Financial Reporting Standards (IFRS). However, this study includes goodwill as a class of intangible assets because goodwill was indeed reported as such in Korea until 2010, and this inclusion has no practical effect on the value relevance of other intangible assets than goodwill. Before the introduction of IFRS, Korean accounting standards 3 for intangible assets were consistent during the period of focus in this study (2001~2010). In the course of the International Monetary Fund (IMF) aid program which was introduced in Korea as a result of East Asian financial crisis of 1997, Korean accounting standards were significantly reformed in a direction closer to Western standards (Han and Manry, 2004). This new set of standards, including the introduction of PP&E impairment, immediate recognition of foreign currency translation gain/loss in the income statement, tightened capitalization conditions for R&D related expenditure 4, among others, took effect in In Korean standards, intangible asset was defined as identifiable non-monetary asset without physical substance but has future economic benefit, which firms possess with the purpose to utilize in the course of manufacturing goods, providing services or lease to others or manage. 5 A Classification method based on characteristics and benefits for business was generally recommended, while examples of sub-classes such as industrial properties, license & franchise, copy right, software, capitalized development expenditure, premium in renting a building, mining/fishing rights were suggested. 6 2 Major changes with the introduction of K-IFRS (Korea IFRS) in accounting for intangible assets are 1) the definition of intangible asset, 2) the impairment & amortization of intangible assets. These will be addressed in corresponding contexts. 3 For the convenience of understanding, Korean accounting standards (or Korean standards) henceforth means the standards during the period of 1999~2010, unless otherwise stated like K-IFRS which is the official name of the new standard. 4 It was no longer allowed to capitalize research expenditure under the new standard. 5 The definition in K-IFRS is just identifiable non-monetary asset without physical substance, in accordance with IAS In K-IFRS, mining right was excluded from the Statement 1038 (intangible assets) of Korea Accounting 6

9 The most significant difference between Korean standards and International standards was probably the treatment regarding impairment and/or amortization of intangible assets. While the finiteness of an intangible asset s useful life is assessed and amortization takes place only for intangible assets with finite useful life when IAS 38 is applied, Korean standard required that any intangible asset had a finite useful life (typically not longer than 20 years 7 ) and the amortization took place once an intangible asset was ready to use. An evaluation for an asset s recoverable amount was required at least annually if its useful life was longer than 20 years. Goodwill was also amortized over its useful life not exceeding 20 years under Korean standard, while IFRS 3 prohibits the amortization of goodwill but requires the impairment test in accord with IAS As a result, it is highly probable that intangible assets reported by Korean firms had quite different book value if international standards were applied. If this is the case, it naturally follows that there also could be differences in value relevance of reported intangible assets. As previously noted, there are more generous aspects in Korean accounting practice regarding the recognition of intangible assets when compared to major accounting environments. Unlike US, Japan and Germany where development expenditure should always be expensed 9, Korean standards allow capitalization of qualified expenditure and therefore are on the same side with IFRS in this issue, probably one of the most controversial issues in accounting for intangible assets. Even though a presence of highly probable future economic benefits 10 was required by Korean standards as one of the conditions for capitalizing development expenditure during the period of 1999~2010, it should be noted that this might not necessarily resulted in a more conservative recognition of development expenditure asset by Korean firms, and there was Standards which is in accordance with IAS 38. However, it is still allowed to report mining right as an intangible asset. 7 Except when a monopolistic or exclusive right was given by related laws or contracts. 8 These discrepancies were all eliminated with the introduction of K-IFRS. 9 In UK and Australia, capitalizing of only successful development costs was allowed until 2005 when the IFRS was introduced. 10 The corresponding implicit probability was 80%, while the requirement of IAS 38 ( probable future economic benefits ) implicitly assumes the probability of 50%. However, this standard also changed in accordance with IAS 38 in K-IFRS: that is, highly was removed in this phrase. 7

10 still significant flexibility in the capitalization of development costs (Han & Manry, 2004). 11 Korean standards also allow capitalization of internally generated software in all industries and therefore quite a number firms report this item 12, but in US only firms operating in software industry are allowed to recognize software expenditure. Further, capitalization is not allowed at all in Japan and Germany. 13 In addition, an upward revaluation of intangible assets is a possible option in Korea when objective evidence of increased value is available. But this is strictly prohibited in US, Japan, and Germany. Considering these features, it can be concluded that in certain aspects Korean accounting practice was less conservative and therefore reporting intangible assets was more generously allowed, when compared to some major accounting environments. The standard setting bodies in most leading capital markets are endeavoring to internationally harmonize accounting standards nowadays, and as a part of this process the subject of intangible assets and the proper accounting treatment for intangible assets is being heavily debated (Dahmash et al., 2009). It is likely that this debate will continue in the near future, because accounting for intangible assets is very controversial issue that standard setters have come to confront (Jennings et al., 1996). This suggests that, for the foreseeable future, accountants and financial economists are likely to be concerned with finding better ways to recognize and value intangible assets (Chauvin and Hirschey, 1994; Dahmash et al., 2009). 2) Review of prior studies With the recognition that accounting information without important intangibles-related items may not provides correct information to market participants, various studies examined the value relevance of intangible assets as a way of evaluating the properness of reported accounting information. Majority of these studies can be classified according to the specific intangible assets on which those studies focused. More specifically, many studies examined R&D assets, 11 In the total sample firm-year of 5,242 (as is shown in later), capitalized development expenditure was reported by 77.1% (4,039 firm-years). 12 Capitalized software expenditure was reported by 26.7% (1,402 firm-years) of total sample. 13 It is of note that software was reported as a tangible asset in UK and Australia until

11 goodwill and brand. Unlike Korean or international standards, the US GAAP does not allow capitalizing the expenses of R&D, a deliberate economic investment activity that drives corporate growth, innovations/inventions and technological advances (Zhao, 2002). This practice motivated several researchers to investigate the value relevance of R&D expenditure by assuming the capitalization of those expenditures. Lev and Sougiannis (1996) examined the value relevance of capitalized R&D and concluded that R&D capitalization is value relevant to U.S. investors by yielding statistically reliable and economically relevant information. Healy et al. (2002) also found that capitalizing R&D expenditure could explain economic returns better than fully expensing in pharmaceutical industry. Zhao (2002) showed that while the reporting of R&D costs provides additional information to accounting earnings and book value in US, the allocation of R&D costs between capitalization and expense further increases the value relevance in France and the UK, where capitalization is allowed. The software capitalization, which is the major exception of immediate expensing in US, were also found to be significantly associated with capital market variables such as stock price or return (Aboody and Lev, 1998). Ballester et al. (2003) also argued that market participants behave as if R&D expenditures have significant future economic benefits to the firm covered by the Compustat database, while there are significant differences in industries between the timeseries and the cross-sectional estimates of the parameters and the economic value of the R&D asset. In Korean context, Han and Manry (2004) showed that R&D expenditures were positively associated with stock price and the association is stronger for the portion of R&D expenditure that was capitalized, rather than expensed. The main focus of these arguments is that, capitalization enables management to provide more useful information about the percentage of R&D outlays capitalized versus expensed and about the period of amortization. The users of financial statements thus know better the performance of projects and management s expectation of their probable future benefits (Ang et al., 2008). However, some studies found different evidences. Using firm-specific model 9

12 parameters, Callen and Morel (2005) found weak empirical support for the value relevance of capitalized R&D expenditure at the firm level. In a French context, Cazavan-Jeny and Jeanjean (2005) found that capitalized R&D is negatively associated with stock prices and returns. They interpreted this result as implying that investors are concerned with and react negatively to capitalization of R&D, and suggested that French managements opportunistic approach to the use of R&D capitalization could explain the non-relevance of R&D capitalization in their setting. Another important intangible asset that is often focused in itself is goodwill. With ever increasing business scope and area of firms, recent decades have seen lots of M&A in most industries in all regions over the world. As a consequence more and more firms are reporting goodwill in their financial statements, and sometimes goodwill has a very significant portion in total assets. Among many researchers who paid attention to the value relevance of goodwill, Chauvin and Hirschey (1994) reported a consistently positive market-value influence of accounting goodwill numbers and McCarthy and Schneider (1995) found that the market perceived goodwill as an asset and incorporated the information in the valuation of a firm. While these studies examined goodwill as it was actually reported, Jennings et al. (1996) assumed a hypothetic reporting without the amortization of goodwill, to find that its value relevance was higher than the value relevance of the actual reporting with the amortization. In UK setting, Jifri and Citron (2009) found that both recognized and note-reported goodwill are significantly associated with stock price. Even though brand is not universally recognized under most current accounting standards, it is indeed a quite valuable asset that significantly contributes to the performance of the firm. Brand values of global firms such as Coca cola or IBM are even comparable to their market capitalization. 14 Barth et al. (1998) examined the value relevance of brand assets using the brand value published by Financial World as the value of brand assets, to find that the brand 14 The brand value of Coca cola is $71.9Billion(2011, Interbrand) and the market capitalization as of 31th DEC, 2011 was $158.4B, and IBM s brand value is $69.9B while the market capitalization was $215.9B. 10

13 value was significantly associated with the stock price and stock return. Under UK setting, which is one of the few exceptions of brand recognition as reporting brand assets including selfgenerated ones is allowed, Kallapur and Kwan (2004) found that reported brand assets in 33 firms are positively associated with stock prices. While these studies focused on one specific intangible asset, some other studies looked into the total intangible assets as a whole or multiple intangible asset classes. Ely and Waymire (1999) investigated the relation between intangible assets, earnings and stock prices under a reporting regime which permitted considerable flexibility for managers to capitalize intangible assets (that is, the pre-sec era in the US). They found the evidence that the coefficient relating earnings to stock price decreased with the level of capitalized intangibles, and interpreted this result as being consistent with the perception by investors that managers might be overstating earnings through capitalizing what indeed should be expensed. The study by Klock and Megna (2000) is also unique in the sense that they examined industry-specific intangibles. They investigated the measurement and valuation of intangible capital in the US wireless telecommunications industry and found that the spectrum license explained 60% of the variation in Tobin s q, while advertising expenditure, customer base and brand royalty also contributed to variation in Tobin s q. Studies focusing on multiple intangible items appear to be more prevalent in countries where capitalization of intangible-related expenditure is more generously allowed, and Australia is a representative example of those countries. In Australian context, Ritter and Wells (2006) found evidences that voluntarily recognized & disclosed identifiable intangible assets, along with goodwill, have positive associations with stock prices. Goodwin and Ahmed (2006) reported that the value relevance of earnings increased for intangible capitalizing firms compared with non-capitalizers. Dahmash et al. (2009) found that the information presented by the average Australian company with respect to goodwill and identifiable intangible assets is value relevant but biased, in that goodwill tended to be reported conservatively and identifiable intangible assets aggressively. 11

14 Some studies in other countries also reported similar findings. Oliveira et al. (2010) found that reported goodwill and other intangible assets are highly significantly associated with stock price in Portugal, and Ali et al. (2010) reported the value relevance of intangible assets in Malaysia s top 50 firms. In Korean context, reported intangible assets were found to be not less value relevant than other assets in the studies of Lee and Kim (2003) and Chung and Cho (2004) for the period of 1991~2002 and 1991~2001, respectively. However, it is noteworthy that few studies explicitly compared the value relevance of specific reported intangible assets. Except the study of wireless telecommunication industry by Klock and Megna (2000) which might not be generalized, only Oliveira et al. (2010), Lee and Kim (2003) and Chung and Cho (2004) looked into specific intangible items other than goodwill (Intellectual property, capitalized R&D expenditure and other intangible assets for Oliveira et al., capitalized R&D expenditure and other intangible assets for Lee and Kim, Chung and Cho). Likewise, explicit differences in value relevance between industries have not attracted much attention, with few exceptions such as Ballester et al. (2003) (which focused on economic value of capitalized R&D asset in different industries) and Dahmash et al. (2009) (which examined the value relevance of goodwill and identifiable intangibles in 8 different industries in the robustness test for the aggregated sample). As a consequence, an approach focusing on both multiple specific intangible items and different industries is hard to find. By using this approach, this study aims to examine the different value relevance of different reported intangible asset classes, not only for the overall firms but for firms in different industries. 3. Research questions & methods 1) Hypotheses and research models Based on prior empirical studies, the value relevance of intangible assets reported by Korean firms is to be tested. The first research question examines whether intangible assets help explain the market values, that is, are value relevant, once the effects of book value and earnings are 12

15 controlled for. H1 : Reported intangible assets are value relevant in explaining market equity value. Despite the widely-observed positive value relevance, for specific reported intangible assets there are also different findings such as Callen and Morel (2005) and Cazavan-Jeny and Jeanjean (2005) as already outlined. If investors expect different future benefits for different reported intangible assets, their resultant value relevance would also be different. Moreover, if a reported intangible asset is positively associated with the market value while another asset is negatively associated, aggregating them together may result in a loss of explanatory power. The findings of Ely and Waymire (1999) that the relations of intangibles to stock price differ with the characteristics of the intangible assets (e.g., rights-baseness or subjection to periodic amortization or downward revaluations), which was contributed to investors different perceptions of the reliability of intangibles carrying values, also support the conjecture that different reported intangible assets may show different associations with the market value. Consequently, total intangible asset is divided into goodwill, intellectual properties, capitalized development expenditure and software expenditure, and other intangible assets in this study. These items are the most frequently observed ones in the financial statements of Korean firms. Therefore, above classification (4 individual asset classes and aggregated other intangible assets) can be considered as a compromising choice between a detailed classification to include as many different intangible asset classes as possible and the need to secure minimum sizes of samples for specific items. In addition, a dummy variable is added to distinguish lossreporting firms, to reflect the finding of Collins et al. (1999) that loss-reporting firms show different price-earnings relation. First item of focus is goodwill, which has been found to be value relevant in the majority of prior studies such as Chauvin and Hirschey (1994), McCarthy and Schneider (1995) and Jifri and Citron (2009), among others. Consistently, it is expected that goodwill is also value relevant in this study and therefore the first subset of the first hypothesis is: H1a : Goodwill is value relevant in explaining market equity value. 13

16 Intellectual properties, as policy tools for encouraging innovation, were also found to be generally value relevant in the literature review of Greenhalgh and Rogers (2007) who focused on patents, trademarks and copyrights. 15 Consistently, intellectual properties are expected to be value relevant and this yields the second subset: H1b : Intellectual properties are value relevant in explaining market equity value. For capitalized development expenditure, most prior studies are also reporting value relevance (e.g., Zhao 2002; Healy et al. 2002; Ballester et al. 2003; Han and Manry 2004) with few exceptions such as Cazavan-Jeny and Jeanjean (2005). Consequently, capitalized development expenditure is also expected to be value relevant in Korean context and therefore the third subset is: H1c : Capitalized development expenditure is value relevant in explaining market equity value. As previously noted, capitalized software expenditure has not been yet popularly covered in literature. However, in consistent with Aboody and Lev (1998), capitalized software expenditure is expected to be value relevant in Korean context: H1d : Capitalized software expenditure is value relevant in explaining market equity value. The fifth and last subset includes other intangible assets. Even though Oliveira et al. (2010) found a positive association between stock price and other intangible assets, it should be noted that this item aggregates all other intangible assets and therefore is a residual. This means that it should not be treated as the same variable as that of other studies. However, in the context of the first hypothesis, it is expected that other intangible assets are also value relevant: H1e : Other intangible assets are value relevant in explaining market equity value. Like most of prior studies that examined value relevance of accounting numbers, the test of value relevance in this study is based on the Ohlson (1995) model, in which a firm s market value is a function of the book value of equity and earnings. The main idea is that if accounting 15 Some studies (e.g., Bosworth and Rogers; 2001) took alternative approach by taking the frequencies of application activities for patent or trademarks as independent variables, and also found significantly positive associations between these variables and market value of firms. 14

17 data are good summary measures of the events incorporated in security prices, they are valuerelevant because they provide a value of the firm which is close to its market value (Oliveira et al., 2010). Consequently, the model to test above hypotheses is as follows. Model 1 : P t + d t = a 0 + a 1 (BVE IA) t + a 2 NI t + a 3 NI t LOSS + a 4 GW t + a 5 IP t + a 6 DV t + a 7 SW t + a 8 OI t + e t where, P t = Stock price as of 31st March of the year t+1. d t = dividend per share of year t. (BVE IA) t = book value of equity minus the amount of intangible assets (including goodwill), per share at the end of year t. NI t = net income per share of year t. LOSS = 1 for loss-reporting firms (that is, NI t < 0) and 0 for others. GW t = goodwill per share at the end of year t. IP t = intellectual properties per share at the end of year t. DV t = capitalized development expenditure per share at the end of year t. SW t = capitalized software expenditure per share at the end of year t. OI t = all other intangible assets per share at the end of year t. e t = residuals. All variables are per-share values, consistent with many of previous studies such as Oliveira et al. (2010), Ritter and Wells (2006) and Zhao (2002). With this model, whether each independent variable coefficient differs significantly from zero is to be tested. The second hypothesis is about the difference of value relevance between industries. According to Chan et al. (1990), stock price changes after announcements of R&D spending were different between high-technology firms and low-technology firms : high-technology firms that announced increases in R&D spending experienced positive abnormal returns, where lowtechnology firms experienced negative abnormal returns. Meanwhile, Lev and Zarowin (1999) showed that the informativeness of accounting earnings decreased during the period of 15

18 1976~1995 and that this decrease was abnormally steep for R&D-intense firms, arguing that immediate expensing of R&D expenditure that would contribute to future earnings resulted in the decreased informativeness. These findings, despite the difference between their contexts, motivate a question about the value relevance of reported intangible assets between industries. If investors expectations of future economic benefit from intangible assets are different between industries, it is also possible that there are differences in the value relevance of intangible assets. In this study, in a similar context with prior studies, the difference between information technology (IT) firms and other firms is examined. If the argument of Blair and Wallman (2001) about the decreasing explanatory power for firm value of reported intangible assets in high-technology firms is extended to Korean setting, it is expected that value relevance of intangible assets would be lower in Korean high-technology firms, which IT-firms are proxies for, when compared to non- IT firms. This leads to the second hypothesis: H2 : The value relevance of reported intangible assets in IT firms is different from that of non- IT firms. To test this hypothesis, model 1 is revised with an additional dummy variable of ITF multiplied by each intangible asset variable. Model 2 : P t + d t = b 0 + b 1 (BVE IA) t + b 2 NI t + b 3 NI t LOSS + b 4 GW t + b 5 IP t + b 6 DV t + b 7 SW t + b 8 OI t + b 9 GW t ITF + b 10 IP t ITF + b 11 DV t ITF + b 12 SW t ITF + b 13 OI t ITF + n t where, ITF = 1 for firms classified as IT firms (by Korean industry classification code 45000) and 0 for other firms. 16 n t = residuals. With this model, whether the value relevance of each item is different between IT firms and non-it firms is to be tested. 16 The sub-categories of IT industry include software & service, hardware & equipment, semiconductor equipment industries. 16

19 Table 2 : Sample selection 2001~2010 Starting number of firm-years 26,420 Less : Financial firms or firms with other than 31st December year end - 5,559 Firms without stock prices - 6,430 Firms with audit opinions other than unqualified Firms with negative book values of equity Firms with intangible assets less than 1% of total assets - 8,912 Final number of firm-years = 5,242 2) Data Retrieval As previously noted, this study depends on the Korean Investor Service database (KIS-Value), the most popular local database in Korea because global database such as Compustat or Datastream are not providing detailed intangible asset accounts reported by firms. The data includes firms listed on Korean stock market during the period of 2001~2010. The final sample of 5,242 observations is derived from a potential sample of 26,420 observations following a filtering process described in Table 2. 5,559 observations were excluded at first because they were for firms in financial industry or who do not have a 31st December fiscal year end (as is usual in Korea). Financial firms are excluded following Ahmed et al. (2000), because they usually have a minimal level of operating assets and are subject to additional regulatory requirements that potentially affect the relation between their accounting numbers and stock market values. In addition, firms with fiscal year ends other than 31st December were excluded to ensure that all sample firms are at the same stage in the financial reporting process for any given valuation date, just like the study of Dahmash et al. (2009). Also excluded are : 6,430 observations without stock prices; 163 observations with the auditor opinions other than unqualified ; 114 observations with negative book value of equity; 8,912 observations with less than 1% of intangible assets (compared to total assets). Of final sample of 5,242 firm-years, which is used to test both the 1st and 2nd hypothesis, 2,169 firmyears are for IT firms and 3,073 are for non-it firms, while 3,341 firm-years include net profit 17

20 and 1,901 net loss. 4. Findings 1) Descriptive statistics and correlation analysis Table 3 shows the descriptive statistics of sample data. While the sample indeed covers a wide range of listed firms in Korea, it is noteworthy that the data is highly skewed and the minimum values are zero for all 5 individual intangible asset classes. Furthermore, many of the quartile values are also zero. This is mainly because many firms do not have multiple intangible assets, even after filtering our firms with intangible assets less than 1% of total assets. As these skewness might indicate a potential heteroskedasticity problem in the regression analysis, an inspection of the residual errors would be necessary. Table 3. Descriptive Statistics (in KRW) Variable Mean Std. Dev. Min. 1Q Med. 3Q Max. P + d 11,565 38, ,540 3,450 8,608 1,050,997 BVE IA 7,147 20,463-2,791 1,046 2,395 5, ,327 NI 500 3,607-57, ,946 GW 144 1, ,259 IP ,439 DV ,612 SW ,245 OI 208 1, ,281 Frequency Percent Total 5, IT 0 3, , LOSS 0 3, , # EUR 1 = 1, KRW (Korean Won) ( ) 18

21 Table 4. Correlations (Pearson) matrix P + d BVE IA NI GW IP DV SW OI P + d 1 BVE IA ** 1 NI ** ** 1 GW ** ** ** 1 IP ** ** ** ** 1 DV ** ** ** ** 1 SW ** ** ** ** ** OI ** ** ** ** * ** 1 # significant at the 0.05(*), 0.01(**) level (2-tailed) Table 4 shows the Pearson correlation coefficients between variables. The coefficients show that all the independent variables are correlated positively with stock price added by dividend (henceforth market value ). Many of independent variables are also correlated between themselves, while most of the values range between 0.15 and ) Regression results Table 5 shows the regression results for both models, which are derived by using SPSS (version 19). 17 For model 1, which is designed to examine different intangible items value relevance, all coefficients for independent variables are significant but there are quite differences among their signs. Book value of equity subtracted by total intangible assets, along with net profit, shows significantly positive associations with market value, and the association between profit and market value significantly decreases for firms with net loss. These results are all in line with prior studies such as Collins et al. (1999), Zhao (2002), Ballester et al. (2003), Han and Manry (2004), Jifri and Citron (2009), Oliveira et al. (2010), among others. For separately examined 5 intangible asset classes, intellectual property, capitalized development and software expenditure are positively associated to market value, while goodwill and other intangible assets are 17 The results of checking residual errors are presented in Appendix. Results suggest that there are not serious heteroskedasticity problems. 19

22 Table 5. Regression results Model 0 1) Model 1 Model 2 (coefficient) (t-statistics) (coefficient) (t-statistics) (coefficient) (t-statistics) BVE IA *** *** *** NI *** *** *** NI LOSS *** *** *** GW *** GW ITF IP *** IP ITF *** DV *** *** DV ITF ** SW *** *** SW ITF *** OI *** *** OI ITF *** Adj. R Observations 5,242 5,242 5,242 F-statistics 4, *** 1, *** 1, *** Note : * p < 0.1, ** p < 0.05, *** p < ) For the comparison of adjusted-r 2 : Intangible asset is not separated here. That is, model 0 is P t + d t = c 0 + c 1 BVE t + c 2 NI t + c 3 NI t LOSS negatively associated. It is also noted that there are significant differences in the magnitudes of positive coefficients, which range from (development expenditure) to (software expenditure). This model has an adjusted R 2 of 0.738, explaining 73.8% of the total variance in the market value. Compared to the adjusted R 2 of when intangible asset is not separated, segregating individual intangible assets improves R 2 by or 3.2%p. 18 The negative value relevance of goodwill (coefficient = , p < 0.01) is not consistent with many prior studies which found positive associations. However, in Korean context, results 18 Alternative analyses in which each intangible asset is dropped yield different values of R 2 in consistent with Table 5; Dropping each intangible asset decreases R 2 from to (without GW), (without IP and DV), (without SW), and (without OI), respectively. It is verified that corresponding decrease(s) in R 2 is biggest for SW which shows the largest absolute value of t-statistics in Table 5, and smallest for IP and DV which show the smallest absolute values of t-statistics in Table 5. 20

23 are conflicting in that Lee and Kim (2003) found a positive association but Chung and Cho (2004) found a negative (but insignificant) association of goodwill. The result of this study might be contributed to the different time frame of sample data (1991~2001 or 1991~2002 for prior studies vs. 2001~2010 for this study), with the conclusion that the stock market did not acknowledge the future benefits of goodwill, but on the contrary, reacted negatively to reported goodwill. One probable way to explain the negative association of goodwill is to assume that investors considered the reported goodwill as a result of an excess payment over the fair value of the acquired firm in an acquisition. Alternatively, the mandatory amortization of goodwill during the sample period might have a substantial impact on the association with a systematic decrease in the amount of reported goodwill, which no longer happens under the new standard since This conjecture is supported by Jennings et al. (1996) who found evidence which suggests that goodwill may not be declining in value for many firms. Jifri and Citron (2009) also argued that goodwill subject to impairment reviews would be valued more realistically, and therefore be associated more closely with market prices. Other intangible assets also shows a negative value relevance (coefficient = , p < 0.01), which is not consistent with a positive association in Oliveira et al. (2010) or positive but insignificant associations in Lee and Kim (2003) and Chung and Cho (2004). However, as previously noted, it may not have much meaning in direct comparison with other studies where the components of other intangible assets are all different. Value relevance of intellectual properties, capitalized development and software expenditure are all significantly positive, which can be interpreted as an acknowledgement for the future benefits of these individual assets by market participants. Value relevance of capitalized development expenditure (coefficient = 2.632, p < 0.01) is consistent with many prior studies that reported positive associations between stock price and development expenditure under various settings. Similar to the finding of Aboody and Lev (1998), capitalized software expenditure also showed a positive association (coefficient = , p < 0.01) with the market value. Value relevance of intellectual properties (coefficient = 7.698, p < 0.01) is consistent with 21

24 Greenhalgh and Rogers (2007), but different from Oliveira et al. (2010) who found an insignificantly negative coefficient and contributed it to the undervaluation of intellectual properties on financial statements. In model 2 which is for examining the differences in value relevance of intangible asset classes between IT firms and non-it firms, the differences were significant for all individual asset classes except goodwill (coefficient = , t-statistics = ). Specifically, market value associations of capitalized development and software expenditure are significantly smaller for IT firms compared to non-it firms (coefficient = , p < 0.05 for DV; coefficient = , p < 0.01 for SW). This result suggests that investors positive expectations for future benefits of capitalized development and software expenditure significantly decreased for IT firms. Even though a decrease in explanatory power for firm value of accounting information has already been noted (e.g., Blair and Wallman 2001), an explanation of this negative effect of capitalized development and software expenditure in IT firms may need further arguments like that of Cazavan-Jeny and Jeanjean (2005). They suggested investors concerns about earnings management through arbitrary capitalization as a reason for negative value relevance of capitalized R&D expenditure under French setting with relatively weaker legal enforcement. Alternatively, investors might not expect future benefits of capitalized development and software expenditures for IT firms because of significant uncertainties stemming from the rapid technological changes in the industry. On the contrary, market value associations of intellectual properties (coefficient = , p < 0.01) and other intangible assets (coefficient = 5.631, p < 0.01) were significantly larger for IT firms. Relatively high value relevance of intellectual properties implies investors recognition that intellectual properties in IT industry have more value than those in non-it industries. Compared to capitalized development and software expenditure of which recognition require management s decisions regarding conditions of capitalization, many intellectual properties such as patent or trademarks are reported at their acquisition cost and are legal rights granted by authorized institutions. Like the suggestion of Ely and Waymire (1999), this difference may 22

25 lead to investors different perceptions on the credibility of these individual intangible assets carrying values. Finally, adjusted R 2 of this model is 0.794, which is quite larger than that of first model (0.738). This enhanced R 2 suggests that examining different value relevance of intangible assets between IT firms and non-it firms indeed helps to improve the explanatory power of the regression model by or 5.6%p. 5. Summary, conclusion, limitations and future research This study examined the value relevance of reported intangible assets in a sample of Korean listed firms. The results indicate that all intangible asset classes have significant associations with the market value of the firm. Intellectual properties, capitalized development expenditure and software expenditure have positive associations, while goodwill and other intangible assets have negative associations. The negative association of goodwill may have been resulted from investor s concern of an excess acquisition cost, or the mandatory amortization of goodwill under old standard which may not be adequate for many firms according to Jennings et al. (1996). If the latter is the case, the introduction of new standard as of 2011 will offer an opportunity to future research for a comparison of value relevance when sufficient sample data is available, like the enhanced value relevance of goodwill in Portugal after introduction of IFRS (Oliveira et al., 2010). For information technology (IT) firms, all intangible asset classes except goodwill showed differences in the level of value relevance when compared to non-it firms. But the directions of these differences are also different among individual classes; value relevance of capitalized development expenditure and software expenditure are lower for IT firms than non-it firms, while value relevance of intellectual properties and other intangible assets are higher for IT firms than non-it firms. While investors concern about earnings management through arbitrary capitalization and/or pessimistic expectations for future benefits in IT industry may have 23