Accounting for intangibles and earnings management : the role of useful life

(1)

Accounting for Intangibles and Earnings Management: the Role

of Useful Life

Name: Jeroen Schoemaker Student number: 11298774 Thesis supervisor: Dennis Jullens Date: June 21, 2018

Word count: 12259

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by student Jeroen Schoemaker who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

In this thesis, I analyse the variation in the expected useful life of intangible assets under International Financial Reporting Standards (IFRS). The annual review of the expected useful life (of intangible assets) may signal private information about the underlying economics of the firm or be used as an opportunistic accrual-based earnings management tool. The thesis analyses a hand-collected sample of relatively R&D-intensive European firms from the EUROSTOXX 600 using multiple experimental variables. The results show that, on average, firms have a stable amortization period for their intangible assets. There is a positive relationship between the change in the expected useful life relative to the previous year and avoiding to report negative earnings and beating past year’s earnings. However, the results show that there is no association between the change in the expected useful life and the forecast error or beating analysts’ forecasts. The LIFE is too stable to have an any association with the generic earnings management proxies. The findings are robust to controls for statistical diagnostics, sample composition, and various alternative explanations.

(4)

1 Introduction

Accounting rules provide managers with great discretion on the choices that relate to the classification and presentation of economic information in the financial statements. One such accounting choice, that may shape reporting outcomes, is the decision to determine the expected useful life of intangible assets under International Financial Reporting Standards (IFRS) in Europe. The decision to increase the expected useful life of an intangible asset allows managers to defer the amortization charge to a future period and provisionally inflate the year-end earnings.

The purpose of this master’s thesis will be to investigate if firms use the annual review of the expected useful life of intangible assets to manage earnings under International Financial Reporting Standards (IFRS) in Europe. Much of the extant literature has illustrated that discretion drives the capitalization of intangible assets. For instance, most recently Dinh et al. (2015) investigate if the capitalization of R&D assets is used to meet or beat earnings benchmarks. In the literature there are two perspectives on the accounting choice to capitalize intangible assets on the face of the balance sheet. On the one hand, the capitalization may be used to signal private information to the market or management may use the uncertainty associated with intangible assets for opportunistic behavior. The existing literature has no definitive answer to the discussion. This master thesis will do so similarly for the accounting estimate to review the useful life of intangible assets.

So far, no empirical study that I am aware of has investigated the expected useful life of intangible assets in relation to earnings management. This thesis will follow the literature and investigate if the accounting discretion to review the useful life of intangibles assets is associated with meeting or beating performance thresholds and the variation in analysts’ forecast error as proxies for earnings management. Hence, the empirical question is formulated as follows:

RQ: Do companies use the expected useful life of capitalized intangible assets to manage earnings under IFRS?

(7)

figures to reflect the underlying economics of the entity in order to facilitate optimal economic decision-making.

Moreover, this thesis may conceptually inform the controversial issue of accounting for intangible assets. Intangible assets are associated with uncertainty as the probability that future economic benefits cannot be reliably estimated. This belief reflects the widely-held position on intangible assets in general (see Lev, 2001). Hence, it is important to understand if the information that is communicated to the intended users of financial reports is reliable and informs decision-making. Therefore, the results of this thesis are expected to be of relevance to users of financial reports and standard-setters. Especially given that IFRS is a principled-based standard that allows for more discretion and flexibility in making accounting choices. The results also matter for external auditors that verify the financial statements. Accounting for intangibles is very subjective and requires professional judgment that may not be verifiable against external information. The auditors may have to consider the pattern of changes in the useful life and the economic effects this may have for the users of financial standards.

The analysis is based on the EUROSTOXX 600 firms between 2007 and 2017, the sample represents the largest publicly listed firms in the European market (all IFRS). In the first set, I analyse if adjustments in the expected useful life of intangible assets are positively associated with the forecast error. The results show that the relationship is not statistically significant. Firms, on average, have a stable amortization period for their intangible assets. If the amortization period remains stable, then there is little to no uncertainty for financial analysts. Moreover, LIFE may represent a structural reform of the intangible assets. Such information is generally impounded in the market, therefore it is not an unexpected surprise. Hence, the theorization that LIFE contributes to information uncertainty is incorrect.

Secondly, the empirical results show that the LIFE variable has a positive association with the earnings management proxy. In other words, an increase in the useful life relative to the previous year is positively associated with avoiding to report negative earnings and beating past year’s earnings. However, the LIFE variable is not statistically associated to meeting or beating analysts’ forecasts. The LIFE is too stable to have an any association with a generic earnings management proxy. All the results are robust to sensitivity checks and controls for statistical checks.

(8)

Finally, the results suggest that in the European Union under IFRS the capitalization of intangible assets is used to signal positive information to the market. Namely, an increase in the amount of capitalization reduces the forecast error. Further research is required to investigate this relationship.

The remainder of this thesis is organized as follows: A conceptual model of earnings management is presented in Section 2. Section 3 provides a literature review. The literature view discusses earnings management, financial analysts and intangible accounting. Section 4 presents the hypothesis development. Section 5 explains the variable selection for the empirical model. In section 6 the sample and all main empirical results are provided. The concluding remarks are in Section 7.

(9)

2 Concepts and theories

The accounting for intangible assets is a controversial issue. For instance, while the IASB allows the R&D expenditures in the ‘development’ phase to be capitalized as assets, the standard setter in the US, the FASB, prescribes immediate expenditure of all costs related to R&D. This relates to two different views on how managers may use discretion. Managers may use discretion to signal positive information to the market or use it opportunistically. This thesis is interested in the latter, and therefore will introduce the concepts of earning management and the tools associated with it – accrual-based earnings management, real earnings management and classification shifting. A change in the expected useful life may be classified as accrual-based earnings management. First, this thesis will discuss the concept of agency cost in order to explain that managers may be incentivized to expropriate wealth from shareholders as there is separation of control and ownership. Secondly, this thesis will discuss the accounting concept of earnings management and discuss the model of Cohen and Zarowin (2010), who investigate the use of accrual-based and real-based earnings management to arrive at a cost-benefit analysis. Moreover, this thesis will delve into the incentives of earnings management.

2.1 Agency costs

Agency costs arise from the separation of ownership and control. Therefore, Jensen and Meckling (1976) define an agency relationship as “a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent” (p. 308). The relationship between the stakeholders and management of a firm is characterized by a separation of ownership and control. If both parties are utility maximizers, then the agent (management) may act against the interests of the stakeholders (principals) when it is in the agent’s interest to do so. Jensen and Meckling (1976) argue, therefore, that agency costs may be defined as the sum of monitoring cost by the principal, bonding cost by the agent and the residual loss.

2.2 Earnings management

Healy and Whalen (1999) provide a seminal review of the earnings management literature. Accounting standards define a language that management uses to communicate with the firm’s external stakeholders. Financial reporting, therefore, has to provide a cost-effective means for

(10)

management to separate the firms that perform well from the poor performers to help facilitate efficient resource allocation from investors and stewardship information to stakeholders. The standard-setters have to take into consideration the timeliness and credibility of the information. However, there is a trade-off: standards that emphasize credibility generally provide less relevant and timely information. The standards thus must permit some judgment and subjectivity in financial reporting, which provides opportunities for earnings management. Therefore, they define earnings management as:

“when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers.” (p. 368)

An important element in the definition is that the stakeholders are being misled, which implies that stakeholders may not have access to the same information as management. Management will thus only have an incentive to manage earnings if they expect that stakeholders cannot undo all earnings management. As will be discussed in the following section, this is what predominantly drives the accounting choice between the accrual-based earnings management and the other types of earnings management.

Moreover, this thesis will discuss two motives for earnings management: capital market motivations and contracting motivations (Healy and Wahlen, 1999). The first motive is the capital market where investors and financial analysts use accounting information to value stocks. There is empirical evidence that management may be incentivized to increase short-term stock returns, for instance, firms that have a high percentage of ownership by institutions with trading strategies and portfolio turnover are more likely to cut R&D spending to manage their earnings upward. Firms with high percentage of institutional ownership, on the other, generally do not reduce R&D expenditure to improve the reported earnings.

The second motive to engage in earnings management is related to contracting. The contracts between management and external stakeholders seek to align the interests of both parties. There are several types of contracts that may incentive management to manage earnings (for an

(11)

management to boost the earnings to avoid costly restructuring. However, there is very remote evidence on the frequency and magnitude and which accruals are used to manage earnings. In summary, the extant literature evidences that managers behave opportunistically. The corporate structure that separates control from ownership may create divergence of interests between the agent and the principal. Empirical evidence suggests that the agent indeed is an utility-maximizer and may have multiple motives to engage in earnings management to increase their own welfare at the expense of the principal. Therefore, this thesis will regard earnings management as opportunistic behavior by management that seeks to maximize their own remuneration.

2.3 Earnings management tools

Cohen and Zarowin (2010) investigate the use of accrual-based and real transaction-based earnings management. The key characteristic of accrual-based earnings management is that they do not have any underlying cash flows and therefore accrual-based earnings management is dependent on the choices permitted by the accounting standards. The accounting choices are generally associated with the acceleration of recognizing revenue and a deceleration of recognizing expenses in an accounting period. However, accrual-based earnings management is very costly because, for instance, current expenses are deferred to a future period, but this will also increase the expenses in future periods by the same amount. Moreover, there is a reputational concern for management. As discussed in the previous section, earnings management is only effective if the stakeholders cannot uncover it, otherwise it results in reputational loss for the managers.

The practice of real transaction-based earnings management, on the other, concerns the manipulation of “real” activities by management. For instance, managers may use price discounts to temporarily boost sales or reduce R&D-related expenditure to improve margins. The use of real earnings management activities, however, has an effect on the future performance of the firm and on the future cash flows. Both methods increase the expectation of future earnings and have an effect on future earnings. The key difference is that it is relatively more difficult to detect real transaction-based earnings management, because the activities cannot be separated from daily business routines and strategies.

Lastly, there is classification shifting which is defined as “the deliberate misclassification of items within the income statement” (McVay, 2006). Classification shifting has no effect on the bottom-line earnings, but may portray the information more favorably by reclassifying

(12)

persistent items into transitory items on the income statement. Therefore, all of the earnings management tools share that they increase the expectations about future earnings, but classification shifting and real-based earnings management are more difficult to detect. In sum, managers consider the costs and benefits associated with each earnings management tool. Based on the above, Cohen and Zarowin (2010) identify three types of cost for accrual-based earnings management: a) the scrutiny provided by the capital markets; b) the potential penalty of detection; and c) the difficulty of achieving a given earnings target. Otherwise other forms of earnings management are more likely to be used by management (see Cohen and Zarowin (2010)). This thesis will focus on accrual-based earnings management under IFRS, specifically the use of the accounting choice to annually review the expected useful life of intangible assets under IFRS.

(13)

3 Literature review

3.1 The attributes of intangible assets

Lev (2001) explains the four economic characteristics of intangible assets in his seminal book. The first characteristic of intangible assets is that they are generally nonrival, in other words, “they may be deployed at the same time in multiple uses, where a given deployment does not detract from the usefulness of the asset in other deployments” (pp. 18). For this reason, intangible assets have negligible opportunity cost other than the original investment. An example given is the development of a software product which is characterized by a high initial investment in research and development activities, but the marginal costs are negligible because the production costs of the software is marginal – the software can be downloaded from the cloud. Therefore, many intangible assets do not share the diminishing returns characteristic of physical assets to the same extent. The effect of the nonrivalry attribute of intangible assets is that the scalability is only limited by the size of the market, hence intangibles are generally characterized by increasing returns to scale. Another driver of the scalability is the network effect of intangible assets, which arises when consumers and users appreciate large networks. For instance, the usefulness of a computer operating systems increases with its ability to connect to more other users.

Secondly, intangibles assets are characterized by hazy property rights (partial excludability and spillovers). The risk that non-owners will benefit from the investments is substantial. The most straightforward example is the training of employees whose knowledge will spill-over to competitors when they switch employers. Even when a patent expires, the invention may be used freely by competitors in the market or even before the expiration date expires the competitors may use the re-engineering to subvert the property rights conferred by patents. This is evidenced by the number of patent infringement lawsuits that retaining the benefits of patents is very difficult. Cohen et al. (2000) have used a survey to investigate if the use of patents has strengthened the protection of intellectual capital. Even though the patent protection has strengthened, the U.S. manufacturing firms surveyed relied more extensively on secrecy and first-to-market time to gain returns on investments than on the protection of their patents by legal entities. In fine, the partial excludability gives rise to spillovers that create a tension between the potential to generate value from investments in intangibles and to secure the benefits from the investments in the absence of comprehensive legal control.

(14)

Thirdly, many intangibles, in particular R&D investments, are inputs for the innovation processes of enterprises. A key characteristic of innovation processes is that they are highly risky relative to regular business activities. Many empirical studies attest that the research and development success is concentrated in only small number of projects. Therefore, the third characteristic is that intangible assets are inherently risky relative to physical or even financial assets. In fact, Kothari et al. (1998) conducted a comparative study of earnings volatility (a measure of risk) associated with R&D and property, plant and equipment assets. The volatility for R&D is on average three times larger. Essentially, IFRS defines the R&D into two components: research is the prospect of gaining new knowledge and the development part is when the acquired knowledge or insight is used for commercial purposes. Essentially, the two major risks arise from this process, if discovery will take place and subsequently what the potential is to commercially benefit from any discoveries. Not all intangible assets share these risks to a similar extent, especially since some are developed to be used internally (see p. 35-36).

Fourthly, there is an absence of active and competitive markets in intangible assets. They are not as easily exchanged as financial or physical assets. The intangibles assets therefore share the attribute of nontradability which has consequences for the valuation and measurement, especially as there are no observable market prices. Markets provide its users with information to optimize resource allocation and the producers of goods and services with liquidity. This largely a consequence of what David Teece states, “It is inherent in an industry experiencing rapid technological improvement that a new product, incorporating the most advanced technology, cannot be contracted for by detailed specification of the final product. It is precisely the impossibility of specifying final product characteristics in a well-defined way in advance that renders competitive bidding impossible in the industry” (in Lev 2001, p. 39). This is essentially a prerequisite of an active market, otherwise volume cannot be created. The negligible marginal costs of producing the goods or services and the hazy property rights also complicate the trading of intangible assets.

(15)

In sum, as a business enterprise does not have strict control over most intangibles, the accounting profession has difficulty recognizing intangibles as assets on the balance sheet. The direct expensing of intangible investments results in a deterioration of the usefulness of financial information to stakeholders (Lev and Zarowin (1999). The belief that the prospects of intangible investments are too uncertain results in the immediate expensing of such business activities. Hence, they are associated with information risk.

(16)

3.2 Analysts and the capitalization of intangible assets

The forecasts of analysts reflect information risk and intangible assets may increase information risk (see Healy et al., 1999; Lang et al., 2003). Intangibles assets are associated with greater uncertainty relative to physical and financial asset. In particular, firms that are characterized by high underlying intangibles. Chalmers et al. (2012) provides two examples of operating environments with high underlying intangible assets: R&D firms or large advertising (brand value) firms. Such firms may be particularly difficult to unravel for financial analysts. The accounting choice to capitalize R&D expenditures or intangible assets on the face of the balance sheet, on the one hand, may potentially convey useful information to the users of the financial statements. This will reduce uncertainty, by signaling positive information to the users. The signaling perspective posits that forecast accuracy is positively associated with disclosure of intangible assets (Chalmers et al, 2012, Anagnostopoulou, 2010; Jones, 2007). Wyatt (2005) finds evidence that in the Australian setting, management capitalizes intangible assets when they are more certain – in other words, based on the underlying economics of the firm. In a collaborative study, Matolscy and Wyatt (2006) confirm the previous study. They analyse if firms that capitalize a higher proportion of their underlying intangible assets have an improved analyst information environment relative to other firms. They find that the capitalization of intangibles is negatively associated with the earnings forecast error of a firm.

(17)

to capitalize intangible assets increases information risk, therefore making it more difficult for analysts to interpret the current information and use it to make earnings forecasts for future periods. Their findings suggest that managers use the option to reduce information asymmetry.

Another stream of research, however, finds that the capitalization of intangible assets is used to misrepresent the financials of company. In the United States, Barron et al. (2002) find that analysts make larger earnings forecast errors when firms have higher capitalized intangibles and Barth et al. (2001) find that firms with higher underlying intangible assets have a greater analyst following. Gu and Wang (2003) find a positive relationship between forecast error and the cost of processing information (task complexity) and the levels of underlying intangibles that are capitalized. In the United States, however, the majority of intangible assets are unrecognized. Analysts are interested in firms that outperform other firms, hence they are more likely to cover firms with performance expectations that are higher relative to their peers. If analysts follow well-performing firms, they reduce the risk of losing their clients because they provide more useful recommendations (McNichols and O’Brien, 1997; Francis and Willis, 2001). Matolscy and Wyatt (2006), therefore, posit that analysts prefer firms with higher intangibles because they expect that their performance will be higher than firms with lower intangibles. The choice, however, to capitalize may not reflect the underlying economics of the firm. Prior research suggests a negative association between intangible assets and analyst forecast properties in the United States.

Intangible assets may increase information risk because the accounting choice to capitalize may not reflect the underlying economics of the firm and the benefits that flow to the entity may be relatively more uncertain. However, Gu and Wang (2003) investigate the effect of information complexity on analysts’ forecast errors. One measure for information complexity is investments in intangibles greater than the industry average. Firms that outspend other firms in intangibles are generally associated to be engaged in pioneering innovations. Innovation is understood as creating new products or services fundamentally different from the existing ones. Much of the prior literature finds a positive relation between the complexity of forecasting task and analysts’ forecast errors. Hence, the capitalization of intangibles assets may also be perceived to contribute to the information complexity of the analysts’ information environment.

(18)

In sum, analysts have difficulty unraveling intangible-intensive firms. As the review of the expected useful life is performed ex-post, it is likely that it contributes to the information complexity of intangible assets. Therefore, this master’s thesis expects that the accounting choice to review the expected useful life is negatively associated with forecast errors.

3.3 Intangible assets and earnings management: signaling private information or misrepresenting R&D success

The existing literature has no definitive answer to the discussion. Research and development capitalization, on the one, may be used to signal private information of management to the market. The managers essentially may use the discretion to capitalize research and development assets to signal that future benefits may flow to the entity because of R&D success.1 Interestingly, the setting in which proponents have found that R&D capitalization is used for signaling private information in a way that increases the usefulness of the financial statements, R&D capitalization was allowed on a discretionary basis under domestic GAAP – specifically Australia, France and the UK. On the other hand, R&D development may be used for earnings management. Most recently Dinh. et al. (2015) have illustrated that R&D capitalization is strongly related to benchmark beating. They have analysed 150 of the largest German publicly listed firms between 1998 and 2012. Hence, there are two different perspectives on capitalization of R&D assets – signaling private information to the market (IASB) and discretion used for earnings management (FASB).

However, the IASB has introduced significant restrictions on the capitalization of R&D, especially IAS 38.57 criteria that are used to test the technical and commercial feasibility. The purpose of the standard is to only allow the capitalization of ventures that have a very high probability of succeeding. The recognition criteria, however, still allow for management discretion. There is a significant degree of difficulty in distinguishing research activities from development activities; the development phase is initiated when all the criteria set by the IASB are met.2 In fact, the capitalization of internally generated intangible assets under IAS 38.57 is

(19)

one of the main issues in financial reporting where material misrepresentations have been detected by the German Financial Reporting Enforcement Panel (Meyer & Naumann, 2009). Although the standard-setters have enforced strict criteria on R&D capitalization, the uncertainty associated with intangible assets has created opportunities for earnings management. In other words, the previous literature suggests that intangibles assets are susceptible to misrepresentation. This can be theoretically extended to the useful life of intangible assets.

3.4 Audit quality and earnings management

Meyer & Naumann examination may suggest that external auditors experience difficulty in uncovering any attempts to misrepresent the financial statements. Indeed, very recently Mazzi et al. (2017) have investigated the capitalization of R&D under IFRS and country-level corruption. They find that substantial positive relation between country-level corruption and the amount of development costs capitalized. Additionally, they also have taken into consideration the future profitability of the firms. If the country-level corruption is significantly higher, the contribution of capitalized R&D costs is significantly lower to future profitability. So the enforcement regime and auditor’s quality may inhibit the use of management discretion for earnings management, but in the extant literature there are more determinants for the accounting choice to capitalize R&D expenditure. In particular, the findings of Dinh et al. (2015) in the German setting with one of the lowest corruption ranking in the world (based on Transparency Index), suggests that the subjectivity and judgment in capitalizing internally generated assets may be used not only to signal private information to investors but also to manage earnings by management. Much of the literature has confirmed the discretion that drives the capitalization of internally generated intangible assets (Ciftci, 2010; and Mmohd, 2005). Therefore, this master’s thesis will not expect audit quality to have an effect on earnings management.

(a) The technical feasibility of completing the intangible asset so that it will be available for use or sale. (b) Its intention to complete the intangible asset and use or sell it.

(c) Its ability to use or sell the intangible asset.

(d) How the intangible asset will generate probably future economic benefits. Among other things, the entity can demonstrate the existence of a market for the output of the intangible asset or the intangible asset itself or, if it is to be used internally, the usefulness of the intangible asset.

(e) The availability of adequate technical, financial and other resources to complete the development and to use or sell the intangible asset.

(f) Its ability to measure reliably the expenditure attributable to the intangible asset during its developments

(20)

3.5 Benchmark beating and earnings management

Accounting earnings are considered to be the most important information provided by the financial statements. Habib (2007) explains that the importance of earnings figures suggests that investors evaluate firm performance based on a) reporting positive earnings; b) reporting earnings greater than last year; c) reporting earnings greater than the consensus analyst forecast. The prospect theory has been used to explain why investors rely on such simple heuristics to evaluate firm performance. The theory states that “decision-makers derive value from gains and losses with respect to a reference point, rather than from absolute levels of wealth” (Habib, 2007). Moreover, the experience of loss is asymmetrical to that of gains, i.e. losses are experienced as more displeasing. Academic research is consistent with the proposition that a small gain may result in a premium (see Habib (2007) for an overview). Hence there are managerial incentives to beat benchmarks as discussed in section 2.2. Although the discussion in this section is very limited, I refer to section 2 where the theoretical constructs are designed and I refer to section 5.1 where the empirical design for earnings management proxies are constructed based on the research of Dinh et al. (2015) and the heuristics of Habib (2007).

3.6 Financial analysts and earnings management

Dinh et al. (2015) have summarized the literature about analysts’ ability to uncover earnings management. Analysts are generally aware of earnings management and the prior literature shows that they react negatively to the capitalization of R&D assets in a significant number of markets, but they are not able to discover if earnings management occurred. Burghstahler and Eames (2003) also find evidence that analysts react to signal that may be related to earnings management. It remains an open question to what extent analysts anticipate earnings management and what types of earnings management they effectively consider. If analysts adjust their targets upward or downward based on the expected useful life has not been investigated by the current literature. However, the review of the useful life is performed ex-post. Therefore, it is highly unlikely that analysts are able to predict the economic consequences. Hence, the scrutiny of capital markets may even be less applicable to the expected useful life of intangible assets. In particular, the nature of intangibles allows estimates

(21)

4 Hypothesis development

The longevity of internally generated intangible assets has an effect on the amortization charge in the income statement. The discretion may be used by management to defer incurred costs to a further period by decreasing the amortization charge. Paragraph 90 provides a list of factors that should be considered when determining the useful life of an asset.3 As Picker et al. argue the principles in IAS 38.90 provide management with more discretion but simultaneously give the possibility to provide more relevant information – i.e., signaling or misrepresentation. Moreover, the useful life should be reviewed at least at the end of each annual reporting period, which provides management with the opportunity to partially defer the amortization charge to a future period. Hence, the useful life of capitalized intangible assets is expected to have a direct effect on the earnings of any given firm, which may potentially be used in order to beat analyst benchmarks. The extant literature has documented that management has incentives to maximize their own remuneration at the expense of the stakeholders. In particular, information related to R&D and intangible assets in general is associated with internally-generated information which inhibits market scrutiny. Therefore, the costs applicable to accrual-based earnings management may be substantially lower for the adjustments of the expected useful life and more difficult for an external auditor to uncover. So, the adjustments in useful life of intangible assets may either be a signal of economic benefits that may flow to the entity or be used to manage earnings, however, this thesis expects that there is a greater likelihood that an increase of the useful life is used to manage earnings as determining the useful life a priori is not likely done reliably, hence:

H1: An increase in the expected useful life of intangible assets is positively associated with meeting and/or beating earnings benchmarks

3_{The list of factors that should be considered are:}

a) The expected use of the asset by the entity and whether the asset could be managed efficiently by another management team.

b) Typical product life cycles for the asset, and public information on estimates of useful lives of similar assets that are used in a similar way

c) Technical, technological, commercial or other types of obsolescence d) The stability of the industry, and changes in market demand e) Expected actions by competitors

f) The level of maintenance expenditure required and the entity’s ability and intent to reach such a level g) The period of control over the asset and legal or similar limits on the use of the asset

(22)

In the literature review, I have summarized the analysts’ perspective of earnings management and capitalization of intangible assets in general. As they are not able to uncover earnings management, it is theoretically plausible they also cannot predict what will occur when the review of the useful life will be performed in a future period and what its effect may be on the firm’s earnings. There is information risk for analysts as the disclosed information is not sufficient to accurately translate into future events. Moreover, research has shown that analysts have difficultly interpreting intangible-intensive firms (Barron et al., 2002; Barth et al., 2001; Gu and Wang, 2003). Hence, the adjustments in the expected useful life of R&D and/or intangible assets may reduce the information content of analyst forecasts, hence:

H2: An increase in the expected useful life of intangible assets is associated with a greater analyst forecast error

(23)

5 Empirical design

5.1 Variables and empirical design – H1

In the first hypothesis, this master’s thesis intends to empirically test if the annual review of the useful life of intangible assets is used to meet or beat earnings benchmarks. Recall:

H1: An increase in the expected useful life of intangible assets is positively associated with meeting and/or beating earnings benchmarks

The research design will be based on Dinh et al. (2015) and Chalmers et al. (2012). This thesis, however, will introduce several experimental variables based on the design of Dinh et al. (2015). Equation 1 represents the earnings management proxies and equation 2 represents meeting or beating analysts’ earnings forecasts.

(1) 𝑀𝐸𝐸𝑇&𝐵𝐸𝐴𝑇_𝑖,𝑡 = 𝛽₀ + 𝛽₁𝛥𝐿𝐼𝐹𝐸_𝑖,𝑡+ 𝛽₂𝑆𝑖𝑧𝑒_𝑖,𝑡+ 𝛽₃𝑃𝑟𝑜𝑓𝑖𝑡_𝑖,𝑡+ 𝛽₄𝐿𝐸𝑉_𝑖,𝑡+ 𝛽₅ 𝐼𝐴 𝑀𝑉𝐴𝐷𝑖,𝑡+ 𝑒𝑖,𝑡 (2) 𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡𝑖,𝑡 = 𝛽0+ 𝛽1𝛥𝐿𝐼𝐹𝐸𝑖,𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝛽3𝑃𝑟𝑜𝑓𝑖𝑡𝑖,𝑡+ 𝛽4𝐿𝐸𝑉𝑖,𝑡+ 𝛽5 𝐼𝐴 𝑀𝑉𝐴𝐷𝑖,𝑡+ 𝑒𝑖,𝑡

The control variables for H1 relate to the significance of intangible assets and the uncertainty of earnings. The adjustment of the useful life of intangible assets is a form of accrual-based earnings management. The accounting option to increase or decrease the expected useful life of the intangible assets is dependent on the economic significance of intangible assets for the firm, economic characteristics that are firm-specific, and the economic performance of the firm. If intangible assets are insignificant, its effect on earnings will be negligible. Moreover, well-performing firms will beat or meet earnings benchmarks more often relative to poor well-performing firms. Therefore, several control variables will be used in the regression analysis for H1.

(1) LIFE

The expected useful life of intangible assets is not disclosed in the notes of the annual report. However, the implied useful life of intangible assets may be calculated using the following formula:

[(Adjusted gross amount – (indefinite intangible assets – adjusted goodwill – intangible assets that are not amortized)) / Total amortization charge t]

(24)

The implied useful life variable requires the gross amount so that the implied useful life is calculated using an identical base. The gross amount of the previous year will be used because any year-end additions are amortized the subsequent year. However, the gross amount t-1 of the capitalized intangible assets will have to be adjusted for any acquisitions, impairment charges and disposals that occur in the year t. Appendix A shows the effect of the disposals, impairment charges and additions that occur in year t on the calculation of the implied useful life. The date of acquisition is not reported in the footnotes of the annual report, therefore the precise implied useful life cannot be reliably deduced from the annual report. For example, if the acquisition of an intangible assets occurs at the end of the third quarter, the assets will only be amortized over the fourth quarter of the fiscal-year (see Appendix A). Therefore, this thesis will consistently proxy for any such effects by taking the average gross amount throughout the period. This is calculated as follows:

[(Gross intangible asset t-1 + Gross intangible assetst)/2]

Any intangible assets that have an indefinite useful life, such as goodwill, will be subtracted from the adjusted gross amount. This thesis is interested in the year-on-year variation. The independent variable, therefore, will be LIFE which is calculated as follows:

[Implied useful lifet - Implied useful lifet-1]

(2) MEET&BEAT

The dependent variable for equation 1 will be a dummy variable MEET&BEAT. The variable will be based on the heuristics of Habib (2007), who explains that the importance of earnings figures suggests that investors evaluate firm performance based on a) reporting positive earnings; b) reporting earnings greater than last year; c) reporting earnings greater than the consensus analyst forecast. Hence, in the extant literature the independent variable used is a MEET&BEAT dummy variable that will equal 1 when any of the firm-year observations meets or beats one of the earnings benchmarks, and 0 if otherwise.

(25)

However, in my sample this variable returns 1 for nearly every single firm-year observation. Moreover, the MEET&BEAT variable is empirically hazy as it cannot explain if the increase in the useful life’s effect on the amortization charge was required to meet or beat the earnings benchmark (see theoretical model). Therefore, following Dinh et al. (2015) this thesis has constructed experimental variables.

First, the variable EFFECTEARN will be constructed. EFFECTEARN represents the effect LIFE variable on the earnings through the amortization charge in year t. The variable may be calculated in two ways. The first method, however, is used in this thesis:

[Adjusted gross intangible assets/(Useful lifet - LIFEt) – (Adjusted gross intangible assets/Useful lifet)]

OR

[Amortization charget – (Amortization charget * (1 - %LIFEt)]

PASTBEAT represents 1 for all firm-year observations that meet or beat past year’s earnings and do not meet or beat past year’s earnings after deducting EFFECTEARN from the ΔEARNINGS, and 0 if otherwise.

BeatZERO is a similar variable that represents 1 for all firm-year observations with earnings greater than zero and report earnings below zero after deducting EFFECTEARN from the earnings, and 0 if otherwise.

The two variables will be used to construct the MEET&BEAT variable. MEET&BEAT returns 1 for all firm-year observations that meet or beat one of the below earnings benchmarks, and 0 if otherwise.

1. [(Earnings ≥ 0 – EFFECTEARN)] < 0 (reporting earnings greater than last year) 2. [(Earnings ≥ 0) – EFFECTEARN)] < 0 (management reports earnings greater than

last year)

(3) FORECAST

FORECAST is a dummy variable that will equal 1 when any of the firm-year observations meets or beats analysts’ consensus earnings forecast, and 0 if otherwise.

(26)

1. EPS ≥ EPS Forecast (earnings greater than the consensus forecast)

(4) SIZE

The variable SIZE will be calculated as the natural logarithm of the market value of the company. Dinh et al. (2015) propose that IAS 38 requires firms to have internal management accounting systems to monitor and determine any benefits from R&D projects. As larger firms have more sophisticated control systems, they expect that they may capitalize more relative to small firms. In a similar fashion, this argument may apply to review of the useful life as well.

(5) PROFIT

The variable PROFIT will measure the returns on assets. Firms with lower performance have an incentive to increase their earnings, because any small increment in the reported earnings may result in a significant premium (see Habib, 2007; Healy and Wahlen, 1999).

(6) LEV

The variable LEV will be the leverage calculated as the total liabilities scaled by the market value of equity. This is a measure of the financial health of a firm and may incentivize managers to manage earnings in order to comply with debt covenants (see Dinh et al. 2015).

5.2 Variables and empirical design – H2

The second hypothesis tests if the annual review of the useful life results in a greater forecast error. Recall:

H2: An increase in the expected useful life of intangible assets is associated with a greater analyst forecast error

As the review is performed ex-post, the analysts may experience difficultly to translate the information into future earnings. Therefore, it may increase the information risk. Therefore, the research design follows Matolscy and Wyatt (2006) and the operational uncertainty measures are based on the research of Alford and Berger (1999) and Demers (2002). Equation

(27)

(3) 𝐹𝐶𝐸𝑅𝑅𝑂𝑅_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝛥𝐿𝐼𝐹𝐸_𝑖,𝑡+ 𝛽₂𝑆𝑖𝑧𝑒_𝑖,𝑡+ 𝛽₃𝑃𝑟𝑜𝑓𝑖𝑡_𝑖,𝑡+ 𝛽₄𝐿𝐸𝑉_𝑖,𝑡+ 𝛽5 𝐼𝐴 𝑀𝑉𝐴𝐷𝑖,𝑡+ 𝛽6𝑀𝑉𝐴𝐷𝑖,𝑡+ 𝛽7 𝑀𝑉𝐴𝐷 𝑀𝑉 𝑖,𝑡+ 𝛽8𝑅&𝐷/𝑇𝐴𝑖,𝑡 + 𝛽9𝐹𝑜𝑙𝑙𝑜𝑤𝑖,𝑡+ 𝛽10𝑂𝑃/𝐷𝐸𝐵𝑇𝑖,𝑡+ 𝛽11𝛥𝑃𝑅𝐼𝐶𝐸 + 𝛽12𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑆𝐷𝑖,𝑡+ 𝛽13𝐿𝑂𝑆𝑆𝑖,𝑡+ 𝛽₁₄σRET_𝑖,𝑡+ 𝑒_𝑖,𝑡

The control variables for H2 relate to the significance of intangible assets as previously discussed and the uncertainty of earnings. Prior research has demonstrated that uncertainty and volatility of firm performance influences earnings forecast dispersion and forecast errors (Matolscy and Wyatt, 2006; Lang and Lundholm, 1996; Ang and Ciccone, 2000; Ciccone, 2001).

(7) FCERROR:

The dependent variable will be the forecast error. The forecast error is measured as the absolute value of the earnings forecast minus the mean forecast of earnings for firm i for the fiscal year

t.

(8) IA/MVAD

The variable that proxies for the accounting option to capitalize the underlying intangible investments is IA/MVAD, which is the capitalized intangible assets divided by a proxy for the underlying intangible investments (Matolscy and Wyatt, 2006). Chalmers et al. (2012) also use the MVAD variable to capture the market value of the intangible assets. The market value added (MVAD) is calculated by taking the market value of equity and subtracting it with the book value of equity adjusted for the total amount of intangible assets. IA captures the total amount of intangible assets recognized on the balance sheet. The variable may not be applicable to the European market, however, in the Australian market the higher capitalization of intangibles relative to the underlying intangibles resulted in lower forecast errors (Matolscy and Wyatt, 2006; Chalmers et al. 2012). In addition, Cheong et al. 2010 find that under IFRS that there is a positive association between capitalized intangibles and forecast accuracy. Moreover, Barron et al. (2002) provide evidence that firms with more underlying intangible assets have greater forecast errors and dispersion.

(28)

(9) MVAD/MV

This variable will be used to control for the level of underlying intangible assets and its expected positive association to the earnings forecast error (Matolscy and Wyatt, 2006).

(10) R&D/TA

This variable will capture the volatility (riskiness) of the intangible assets. Research and development may improve current intangible assets or initiate technological obsolescence of others. Chung and Charoenwong (1991) show that intangible-intensive firms have greater upside earnings potential as well as higher downside risk relative to other firms. Therefore, one may expect greater earnings volatility for intangible-intensive companies. Gu and Wang (2003) find that technologies (R&D) that are more innovative are positively associated with forecast errors. Technologies that are more innovative tend to be riskier, hence are more difficult for analysts to process. Although their measures are more comprehensive, this measure will be sufficient to test this hypothesis.

(10) OP/DEBT

OP/DEBT variable is calculated as the operating cash flows divided by the total debt of the firm.

(11) LOSS

A dummy variable LOSS will be created that equals one if the firm experienced earnings loss in a given year and zero if otherwise.

(12) Size

The variable LG(MV) will be used because larger firms tend to have more stable earnings (i.e. less risky) (see also Dinh et al., 2015).

(13) LEV

(29)

appears in the sample, in this master’s thesis the rolling calculation is based on monthly stock returns. Given that in the sample many fiscal-years are missing, the rolling calculation on monthly stock returns provides more accurate estimation of the volatility of stock price.

(15) PRICE

The PRICE variable will control for earnings management incentives that are associated with previous stock price changes. It will be calculated as stock price in time t subtracted by stock price time t-1 and subsequently divided by t-1 price (see Abarbanell and Lehavy, 2000 as referenced in Matolscy and Wyatt, 2006).

(16) Follow

This represents the total number of analysts that issue earnings forecast for the given firm in year t. More analyst following is associated with an improved analyst information environment, given that more effort is expended to understand the financials of the firm.

(30)

6 Sample and Empirical Analysis

6.1 Sample and data

This thesis uses the Wharton Research Data Services (WRDS) database for the control variables, such as company-year level accounting data for European firms that are available through the Compustat Fundaments Annual Database. Moreover, the analyst forecast information is available through the WRDS database, specifically the I/B/E/S database. The Datastream database will also be used for company-year level accounting data. IFRS was officially adopted in the European Union in the year 2005. Hence, I will use a period range of 13 years since the year of adoption. The sample comprises the listed firms of the EURO STOXX 600 with analyst forecast data available between 2005 and 2017 on I/B/E/S database files. The firms of the EURO STOXX 600 are the largest European firms that have to report under IFRS. After removing all the missing observations, a random sample of 100 firms out of the 600 listed firms were selected with a total of 830 firm-year observations out of the total sample of 3607 firm-year observations.

(31)

erroneous firm-years that were theoretically not feasible (e.g. a LIFE of 835). For the forecast regression, an additional 100 outliers have been removed and for the MEET&BEAT regression 31 outliers have been removed. For the latter, the majority concerned LIFE variable with a value greater than 25. Therefore, the final sample for the MEET&BEAT regression has 625 firm-years and 556 firm-years for the forecast regression. In the literature, the intangible regression sample sizes are usually around 600-800 units of observation in the European setting, given the limited availability of company-year information for many variables. Furthermore, the diagnostics on the regressions are summarized in table 2. The DWT stands for the DurbinWatson test which controls for autocorrelation in both regression models. An additional variable was created that assigns a random value to all firm-years so that the statistical language R objectively calculates the DW-statistic. The DW-statistic is near value of 2 for both regressions, this is an excellent result. The mean VIFs are not substantially greater than 1, therefore no problems should arise from multicollinearity. The tolerance of all individual variables has also been performed on all the other regressions, however, only the MEET&BEAT regression is presented to eliminate any concerns about how the experimental variables are determined. The tolerance may not be below 0.2 and the individual VIF may not be greater than 10. This is not the case for any of the variables. The MEET&BEAT, therefore, has a strong validity to represent the phenomenon of earnings management that intends to reflect. The model used for MEET&BEAT proxies for the choice to increase the useful life to meet or beat earnings benchmarks. It is important to note that the model cannot imply causality.

(32)

6.2 Descriptive statistics

The experimental variable LIFE has a mean of 0.27 (Forecast) and 0.30 (MEET&BEAT) and a median of 0.42. The outcomes of the descriptive statistics, thus, demonstrate that in my sample the average firm has a stable amortization period for their intangible assets. The minor adjustments in the implied amortization period are expected as annually new intangible assets are acquired and old ones are disposed (see Appendix A for an illustration). This outcome is in line with the expectations because any adjustments to the amortization period requires a change in accounting estimates. Hence, this generally only occurs when there is a structural change in the underlying economics of the firm. On the one hand, the descriptive statistics show that the

(33)

increase of LIFE proxy is very infrequent, it is expected that there will be no effect on the forecast error. If the amortization period remains stable, then there is little to no uncertainty for financial analysts. Moreover, any structural reform that may be reflected in the LIFE proxy will be available to the market. Hence, the information may not be unexpected as theorized. For the former (consequence a) additional robustness tests will be designed in the regression model for MEET&BEAT. Moreover, the infrequent use of LIFE proxy for earnings management purposes is also reflected in the MEET&BEAT variable. The average value for the latter is 0.03.

There are several other interesting observations, one being the sample represents the 600 largest publicly listed firms in Europe. There are no outliers for the size variable and therefore the Size variable may not have a significant effect on the analysis given that there may be thresholds. In particular, the conjecture that large firms have more sophisticated control systems to monitor and determine benefits (Dinh et al., 2015). Moreover, the sample may reflect a bias as large firms have relatively more stable earnings. The former may increase the LIFE proxy as additional internally-generated information may be used to justify a change in accounting policy. There are several firm-years with a negative MVAD. As Matolscy and Wyatt (2006) argue that the negative MVAD are most likely the result of write-offs of intangible assets that are preceded by falling market prices. The FORECAST variable occurs relatively frequently with an average value of 0.26. Hence, it is unlikely that LIFE proxy will be significantly related to this variable as it is employed infrequently. Finally, the average R&D/TA is relatively high with 0.05 (this represents the largest 25% in the sample). This thesis expected that this would be associated with a relatively larger LIFE proxy. However, the descriptive statistics suggests that this is not the case.

To summarize: adjustments in the LIFE proxy occur infrequently in the sample, this is in line with the expectations because any large increase generally requires a change in accounting policy. Therefore, the LIFE proxy may represent a structural reform of the intangible asset rather than an earnings management decision. Additional robustness tests will be used in the regression model to take this into consideration. Secondly, as any large increase of LIFE proxy is very infrequent, it is expected that there will be no effect on the forecast error. If the amortization period remains stable, then there is little to no uncertainty for financial analysts. Moreover, any structural reform that may be reflected in the LIFE proxy will be available to the market. Hence, the information may not be unexpected as theorized. The relatively large

(34)

R&D expenditures seem to not have any effect on the consistency of the amortization period that is used for the intangible assets. Lastly, given that the sample only contains large publicly listed firms, there may be a bias in the sample.

(35)

6.3 Main empirical results and robustness tests

6.3.1 Main empirical results – Forecast Error

The main empirical results that investigate the association of LIFE in relation to the forecast error (H2) are summarized below in table 4. The LIFE is not statistically significantly related to the forecast error. This is not a surprise, in the descriptive statistics I have identified that the adjustments in the LIFE proxy occur infrequently in the sample, this is in line with the expectations because any large increase generally requires a change in accounting policy. Therefore, I expected that there will be no effect on the forecast error. If the amortization period remains stable, then there is little to no uncertainty for financial analysts. Moreover, any structural reform that may be reflected in the LIFE proxy will likely be available to the market. Thus, this thesis may conclude that the annual review of the useful life of intangible assets is not related to the forecast error. I theorized that the useful life could reflect uncertainty and information complexity for analysts, however, the analysis suggests that managers tend to infrequently revise any assumption that have been used to determine the expected life. The empirical results are conclusive, hence no additional tests are employed.

Moreover, the main empirical results return multiple statistically significant variables as demonstrated in the previous literature (see Lang and Lundholm, 1996; Demers, 2002). Unlike the empirical results of Matolscy and Wyatt (2006) for the full sample regression, my regression additionally reports a statistically significant association on OP/DEBT, IA/MVAD, ΔPRICE and EarningsSD in relation to the forecast error. There is a statistically significant negative association between MVAD/MV and IA/MVAD in the main empirical results. The MVAD/MV variable is used to control for the level of underlying intangible assets and the IA/MVAD variable captures the amount of intangibles that are capitalized (Matolscy and Wyatt, 2006). The results suggest that in the European Union under IFRS the capitalization of intangible assets is used to signal positive information to the market as an increase in the amount of capitalization reduces the forecast error. This is in line with the findings of Matolscy and Wyatt (2006) in the Australian market. Wyatt (2005) finds evidence that in the Australian setting, management capitalizes intangible assets when they are more certain – in other words, based on the underlying economics of the firm. In a collaborative study, Matolscy and Wyatt (2006) confirm the previous study. They analyse if firms that capitalize a higher proportion of their underlying intangible assets have an improved analyst information environment relative to other firms. They find that the capitalization of intangibles is negatively associated with the

(36)

earnings forecast error of a firm. Cheong et al. (2010) similarly investigates the effect of IFRS adoption on the association between reported intangibles and forecast accuracy. He finds a positive association between capitalization of intangible assets and the analyst information environment in Asia. In the United States, Barron et al. (2002) find that analysts make larger earnings forecast errors when firms have higher capitalized intangibles and Barth et al. (2001) find that firms with higher underlying intangible assets have a greater analyst following. Although this thesis does not intend to answer this question, the results are in stark contrast with the US evidence. Further research should investigate this question.

(37)

previous year is positively associated with avoiding to report negative earnings and beating past year’s earnings. It is important to note that the logit regression, however, is based on probability. This outcome holds true for all subsets, however, for the increasers subset the LIFE variable is significant at 0.1 level respectively, two-tailed. The other variables are not statistically significant. The increase of LIFE is infrequently used for earnings management (low R2 0.095and coefficient 0.209), hence the sample size may not be sufficient to find any significant relationships between the control variables and the dependent variable. Profit is as expected negatively associated with MEET&BEAT, because firms do not have to engage in earnings management if their current performance meets or beats analyst forecasts. The IA/MVAD also has a positive association, if more intangible assets are capitalized the amortization is more likely to have a greater effect on the earnings. There are, however, implications to be derived from the former and latter observation. If both variables were statistically significant additional concerns may arise. If IA/MVAD or Profit variables were statistically significant, the empirical proxy MEET&BEAT may likely reflect those relationship (i.e., likely to be not significant). When there are low profits, the LIFE is far more likely to return benchmark beating because its effect on earnings would have to be significantly smaller. Similarly, if IA/MVAD is greater, then the LIFE is far more likely to return benchmark beating because its effect on earnings would be relatively much greater. Hence, the regression model has ruled out other explanations why the LIFE may result in avoiding to report negative earnings and beat past year’s earnings. Additional tests will be discussed in the robustness checks section 6.4.

Panel B reports the main empirical results for the FORECAST model. The LIFE variable is not statistically significant and has a positive association with meeting or beating analysts’ forecasts. The dependent variable in this regression is generic, which confirms that the LIFE is too infrequent to have a statistically significant effect in this test. As expected Profit and Size are statistically significantly related to meeting and or beating earnings benchmarks as demonstrated in the extant literature (see Dinh et al., 2015). The empirical results suggest that in the sample firms who report losses are more likely to beat analysts’ earnings forecasts. The results for our main variable LIFE are conclusive, hence no additional tests are necessary.

(38)

(39)

6.4 Robustness checks

The findings are robust to a number of sensitivity checks. In the previous discussion, this thesis has pointed to several additional events that MEET&BEAT regression may reflect. First, the dependent variable may reflect poor firm performance, therefore an increase in LIFE is more likely to result in avoiding to report negative earnings or beating past year’s earnings. Second, the dependent variable may also reflect aggressive intangible asset capitalization which increases the likelihood that LIFE is far more likely to return benchmark beating because its effect on earnings would be relatively much greater. For both issues the regression model has used control variables.

The third alternative explanation may be that the dependent variable reflects a structural reform of the underlying economics of the firm. The adjustment of the expected useful life of intangible assets occurs infrequently, therefore the major upward observations may confound the relationship with the earnings management proxy. Therefore, the regressions are repeated for a sample where all outliers greater than 9 LIFE are removed. The results are reported in Panel B of table 5. The main empirical results for this regression return a stronger positive association between LIFE and MEET&BEAT and are statistically significant (<0.001). In

(40)

other words, by removing the LIFE observations that may reflect structural reforms of the intangible assets, the coefficient LIFE of 0.209 increases to 0.431. The firm-year observations that have major adjustments in the expected useful life, therefore, are not associated with the earnings management proxy. Thus, the additional regression controls for the confounding effect that the LIFE may reflect a structural reform of the underlying economics.

Theoretically, the empirical results add up. The risk of detection determines how likely it is that management will execute earnings management policies. As discussed in the conceptual framework, earnings management is only effective if the stakeholders cannot uncover it, otherwise it results in reputational loss for the managers. Any major structural reform will have to reflect the underlying economics, otherwise the risk of detection will be accordingly high that any of the potential benefits will be far outweighed by the potential costs. The minor upward adjustments of the LIFE are highly unlikely to reflect improved underlying economics, especially since IAS 38 paragraph 57 is a conservative accounting standard, most commonly used for downward revisions. Neither are they likely to reflect earnings management as their impact on the earnings is insufficient. Therefore, the majority of minor upward adjustments are attributable to year-on-year changes in the composition of intangible assets. Hence, the adjustments that are inside the range of the minor and major increases in the LIFE are theoretically more likely to be associated with earnings management because the risk of detection is significantly lower and the impact is sufficiently large. The empirical model presented in the thesis attests this.

Additionally, an additional regression is performed on only the firm-year observations where LIFE is increased. The purpose of this regression is twofold: a) the firm-year observations that decrease LIFE will not do so to meet or beat earnings benchmarks, therefore they may inflate the empirical results; b) to tests if the increase in the coefficient of LIFE variable in regression that was performed on the LIFE<9 sample is the result of removing positive LIFE observations that have a value greater than 9. The main empirical results for this regression are also reported in Panel B of table 5. For both increaser samples the LIFE is positively associated with MEET&BEAT and are statistically significant. Moreover, the coefficient of

Accounting for intangibles and earnings management : the role of useful life