LIFO method for inventory and the effect on the use of real activities manipulation

LIFO METHOD FOR INVENTORY

AND THE EFFECT ON THE USE

OF REAL ACTIVITIES

MANIPULATION

Wilhelmina B.M. Steman, 10642722

25 June 2017

Supervisor: dhr. dr. W.H.P. (Wim) Janssen

MSc Accountancy & Control – Accountancy track

Amsterdam Business School

Faculty of Economics and Business, University of Amsterdam

Word count: 13,592

Statement of Originality

This document is written by student Wilhelmina B.M. Steman 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.

Abstract

The purpose of this research is to examine the potential relation between the use of LIFO accounting for inventory and the use of real activities manipulation through overproduction. I argue that managers in firms which use LIFO accounting, have more opportunities to engage in real activities manipulation through overproduction and will use these to their own advantage. The assumption that unconditional conservatism (LIFO accounting) provides opportunities for real activities manipulation has been believed to be true without any investigations into this particular subject. Therefore, I would like to contribute to the existing claims made by Penman _ Zhang (2002) by researching this topic.

With this research, I used the method of Roychowdhury (2006) for testing the hypotheses. Consistent with my expectations I found that the use of LIFO accounting for inventory is

positively related to the use of real activities manipulation through overproduction. Furthermore, I found that there were no indications of a joint influence of firms that use LIFO accounting and are active in the manufacturing industry or firm-year observations that are in the post-SOX period on the use of real activities manipulation. Nevertheless, findings suggest that the use of LIFO accounting amplifies the use of real activities manipulation.

Content

2.2.1 Different forms of conservatism 15

2.3 Earnings management 18

2.3.1 Different kinds of earnings management 19 2.3.2 Reasons to use earnings management with respect to 21

accounting conservatism

2.4 Sarbanes-Oxley Act 22

2.5 Hypotheses development 24

3. Research methodology 27

3.1 LIFO accounting for inventories and real activities 27

manipulation 3.2 Estimation models 28 3.3 Descriptive statistics 29 3.3.1 Firm characteristics 29 3.3.2 Regression model 33 4. Results 35

4.1 Comparison of LIFO firms with the rest of the sample 35 4.2 Comparison of LIFO firms in the manufacturing industry 37

and the rest of the sample

4.3 Comparison of LIFO firms pre- and post-SOX 39

5. Summary and Conclusion 42

5.1 Summary 42 5.2 Conclusion 44 5.3 Limitation 44 5.4 Further research 45 References 46 Appendices 51

Appendix A. variable descriptions 51

Appendix B. correlation matrix 52

Appendix B1. correlation matrix various variables 52 Appendix B2. correlation matrix of hypothesized determinants 52

List of Tables

Page Table 1: Descriptive statistics LIFO versus FIFO 30 Table 2: Descriptive statistics pre-SOX versus post-SOX 32

Table 3: Model Parameters 34

Table 4: Comparison of LIFO firms with the rest of the sample 37 Table 5: Comparison of LIFO firms in the manufacturing industry and the 39

rest of the sample

1. Introduction

In this paper, I examine the effect of the use of conservative accounting on the use of real activities manipulation. I specifically look at the effect of the use of LIFO (Last-in, First-out) accounting for inventory on the use of overproduction. Furthermore, I investigate if there are any changes to the effect if a firm is active in the manufacturing industry and if the observation is of the post-SOX (Sarbanes-Oxley Act) implementation period.

Ruch & Taylor (2015) mention in their article that there is little question about the existence of conservatism, but there is an ongoing debate among researchers and standard setters about how costly or beneficial it is to the financial statement users. The Financial Accounting Standards Board (FASB) hasn’t included accounting conservatism as one of the qualitative characteristics of financial reporting in its conceptual framework. This is because it is believed that accounting conservatism biases accounting information and comprises neutrality FASB, 2010). There are also several researchers who agree with this notion and argue as well that conservatism biases financial statement numbers which results in inefficient decision-making by financial statement users (Gigler, Kanodia, Sapra & Venugopalan, 2009; Guay & Verrecchia, 2006). On the other hand, there are researchers who argue that accounting conservatism arises naturally between contracting parties and is actually necessary as an efficient contracting mechanism (Basu, 1997; Watts, 2003a).

Conservatism is defined as reporting the lowest value among possible alternatives for assets and the highest alternative for liabilities. Extending this definition, researchers identified two forms of conservatism that produce the understatement of accounting value. These include: conditional conservatism and unconditional conservatism. The biggest difference between the two forms is that conditional conservatism depends on economic news events. Another difference is that the conditional conservatism in research on accounting conservatism is more prevailing than unconditional conservatism.

Chen, Hemmer & Zhang (2007) show that conservative accounting standards combat the use of earnings management. In their analysis it is presented that employing conservative

accounting standards can reduce earnings management. On the other hand, Watts (2003b) states that it is possible to use earnings management while using conservative accounting, but he does not mention anything about the possibility of the influence of conservative accounting on the use

of earnings management. Furthermore, Jackson & Lui (2010) elaborate that there have been a lot of claims that conservative accounting facilitates earnings management, but there is limited evidence on this matter. Moreover, they mention that it will be unlikely that conditional

conservatism would have a significant relationship with earnings management. This is due to the uncertainty of news events, therefore, managers are not able to effectively rely on the accounting recognition of news events in a given period to meet earnings targets. On the other hand,

unconditional conservatism has the ability to facilitate earnings management through the

accumulation of reserves on the balance sheet that can be released into earnings when an earnings target needs to be met. These reserves are accumulated over multiple accounting periods. If managers use these reserves to manipulate earnings, they are opportunistically exploiting conservative accounting to achieve earnings targets in future periods. Some examples of unconditional conservatism are: accelerated depreciation methods, expensing R&D costs, expensing advertising costs, accumulated reserves in excess of expected future costs (e.g. allowance for doubtful accounts), and LIFO inventory (Ruch & Taylor, 2015).

In this paper I will follow-up on a research proposal made by Ruch & Taylor (2015), they stated that there could be an investigation into the relationship between unconditional

conservatism methods and the use of real activities manipulation. Penman & Zhang (2002) already suggested in their investigation that unconditional conservatism could facilitate real activities manipulation, but this has not been researched yet. In order to investigate this, I will look at the reserves created by using the LIFO accounting convention for inventory, which can be released into earnings by manipulating the operations of the firm (Ruch & Taylor, 2015). These reserves arise because of the difference in the value of inventory based on LIFO accounting or another measure (FIFO [First-in, First-out] or average cost). LIFO accounting carries inventories on the balance sheet at lower amounts than FIFO or average cost if inventory prices have risen in the past. In that case, it is more conservative than the alternatives under conditions of rising inventory prices. Earnings are not affected by LIFO if dollar inventories are unchanged for the earnings period: cost of goods sold is equal to the purchase for the period, as it is with FIFO. Accordingly, the LIFO reserve is unaffected. When dollar inventories increase (so higher LIFO inventory costs brought into cost of goods sold), either through physical inventory growth or inventory price changes, earnings are lower under LIFO. This leads to an increase in the LIFO

reserve. If dollar inventories decline, the LIFO reserve decreases and this decreases is brought into earnings (Penman & Zhang, 2002).

In my investigation, I will explore if managers in firms that use LIFO accounting for inventory are more likely to engage in real activities manipulation via overproduction than managers in firms that use FIFO or average cost methods. Overproducing means that managers are allocating more overhead to inventory and less to cost of goods sold, which leads to lower cost of goods sold and increased operating margins. Managers can increase production more than necessary in order to increase earnings when dollar inventories are low. When this is the case and the company keeps producing products they are able to diminish the LIFO reserve which leads to more earnings. Generating more earnings through overproducing is the third method of real activities manipulation according to the model of Roychowdhury (2006). In order to investigate this relationship my main research question will be:

For answering my hypotheses I will use the agency theory, because in prior research is found that conservatism contributes to lowering agency costs through which it helps solve this problem (Ahmed, Billings, Morton & Stanford-Harris, 2002; Jayaraman & Shivakumar, 2013; Watts, 1977, 2003a, 2003b; Watts & Zimmerman, 1986; Ball, 2001; Holthausen & Watts, 2001; Ball & Shivakumar, 2005). They all claim that conservatism helps to alleviate agency conflicts. Although most of this research is based on conditional conservatism, in my opinion, this would also apply for unconditional conservatism. This is because the aim of using unconditional conservatism is also to make the reported information more informative and improve the credibility, which lead to less miscommunication and therefore reduce agency costs (Jensen & Meckling, 1976).

Furthermore, in my dataset I will only include U.S. firms, because in the U.S. companies are obligated to report the value of the LIFO reserve in the footnotes of the financial statements if they use LIFO accounting for inventory. Moreover, I would also like to see if there are any

Are managers in firms that use LIFO accounting for inventory more

likely to engage in real activities manipulation through

changes in the effect of the use of LIFO accounting for inventory on the use of real activities manipulation through overproduction after the implementation of SOX (Sarbanes-Oxley Act). Graham, Harvey & Rajgopal (2005) found evidence that managers prefer real activities manipulation over accrual-based earnings management, because real activities manipulation methods are less likely to be scrutinized by auditors and regulators. Moreover, Graham, Harvey & Rajgopal (2005) and Cohen, Dey & Lys (2008) have observed a switch from accrual-based earnings management to real activities manipulation in the post-SOX period. To see if this effect is also present in this particular case I will investigate the following sub-question:

The research method used in this paper is the model of Roychowdhury (2006), which measures real activities manipulation. The data will be collected from COMPUSTAT

incorporating U.S. firms for the period 1998-2013. This data range is selected to make the

research as most recent and complete as possible. Not all necessary variables where present in the database for all different companies and years until now, therefore I selected the time period with the most complete dataset available to include as many observations as possible.

Findings do provide an indication for the positive relation between the use of LIFO accounting for inventory and the use of real activities manipulation through overproduction. Contrary to my predictions, there was no indication that this effect was stronger for companies that where active in the manufacturing industry. Moreover, there is also no indication that the use of real activities manipulation through the use of overproduction has increased in the post-SOX implementation period. This result was contradicting with my expectations.

This research contributes to prior literature in several ways. Firstly, as there have not been many investigations into the effects of unconditional conservatism this research will shed some light on the proposed issue regarding the possibility of earnings management. Secondly, as the use of accrual-based earnings management seems to have decreased after the implementation of SOX, the question arises if this implies an increase in the use of real activities manipulation. Thirdly, since the prior literature already assumes the positive effect of unconditional

Is there a change in the use of real activities manipulation for firms

that use LIFO accounting for inventory after the implementation of

conservatism on the use of real activities manipulation, it might be interesting to see some results of an investigation on this subject to get more insight in the actual consequences. However I will investigate only one aspect of unconditional conservatism and one aspect of real activities

manipulation which will not lead to a conclusive answer about this question. Further research into this area is necessary to assure the assumption made is trustworthy.

The rest of the paper is organized as followed. In the second section, I will explain the prior studies that relate to the investigated topic. I will start with explaining the agency theory and how it relates to conservatism. After that I will elaborate on accounting conservatism and the different forms that exist. Further I will explain what earnings management is and elaborate on the different kinds of earnings management. I will continue with analysing the changes made when the Sarbanes-Oxley Act was implemented and how this could have led to a change in the use of earnings management. After discussing all the literature, I will form my hypotheses. Then I will elaborate on my research method and the sample collection after which I will perform the regressions and form the answers to my hypotheses. After this I will summarize my research and conclude on the results of the findings. Finally, I will end this paper with a limitation and some ideas for further research.

2. Literature Review

In this chapter I will discuss prior literature that relates to the investigated subject. I will start with explaining the agency theory. Next I will explain accounting conservatism and elaborate on the different forms of accounting conservatism. After that I will continue with discussing the literature on earnings management and mention the different kinds of earnings management. Further, I will discuss the implementation of the Sarbanes-Oxley Act and how this could influence the use of earnings management. Finally, in the last subparagraph, I will form my hypotheses.

2.1 Agency theory

During the 1960s and early 1970s economists explored risk sharing among individuals (Arrow, 1971; Wilson, 1968). This literature described the risk-sharing problem as one that arises when cooperating parties have different attitudes toward risk. Agency theory broadened this risk-sharing literature to include the so-called agency problem that occurs when cooperating parties have different goals. The agency theory is specifically directed at the agency relationship (Eisenhardt, 1989). Pepper & Gore (2015) elaborate that the agency theory has been a major component of the economic theory of the firm since the publication of formative work by Spence & Zeckhauser (1971), Alchain & Demsetz (1972), Ross (1973), and Jensen and Meckling (1976).

Jensen & Meckling (1976) define the agency relationship as a contract under which one or more principals engage the agent to perform some service on their behalf, which involves delegating some decision-rights to the agent. If both parties in this relationship are maximizing their utility they see that there is a good reason to believe that the agent will not always act in the best interests of the principal. The principal can limit these urges of self-interest by establishing appropriate incentives for the agent and by incurring monitoring costs to check the activities of the agent. In some situations the principal will pay the agent to expend resources (bonding costs) to guarantee that he will not take certain actions which would harm the principle or to ensure that the principal will be compensated if he does take such actions. Jensen & Meckling (1976)

however claim that it is generally impossible for the principal or the agent to ensure that the agent will make optimal decisions from the principal’s point of view at zero costs. They continue with saying that in most agency relationships the principal and the agent will incur positive monitoring and bonding costs, and in addition there will be some divergence between the agent’s decisions and those decisions which would maximize the welfare of the principal. This is also a cost of the agency relationship and is referred to as ‘residual loss’.

The relationship between stockholders and managers of a company fit the definition of a pure agency relationship and therefore the issues associated with the separation of ownership and control are associated with the general problem of agency. The problem that arises from this agency relationship is less severe if the manager of the firm is also a 100% shareholder of the firm. In this case, the manager experiences all the benefits of his decisions but also all of the costs associated with the decisions. On the other hand, if the manager of the firm does not hold 100%

of the shares of the company he or she does experience all of the benefits but only a part of the costs which could lead to less efficient decision-making. In this situation the managers could make decisions that only benefit themselves, because the costs that are associated with the decision do not fully damage the welfare of the manager (Jensen & Meckling, 1976).

According to Eisenhardt, the agency theory is concerned with solving two problems that can occur in an agency relationship (a) the desires or goals of the principal and agent conflict and (b) it is difficult or expensive for the principal to verify what the agent is actually doing. She mentions that a problem exists for the principal which cannot verify that the agent is behaving appropriately. In this relationship there is also a problem about different risk preferences. If the agent and the principal prefer different amounts of risk there might be an issue. The focus of the theory is to determine the most efficient contract, governing the principal-agent relationship, given assumptions about people (e.g., self-interest, bounded rationality, risk aversion), organizations (e.g., goal conflict among members) and information (e.g., information is a commodity which can be purchased). The most interesting question in her opinion is if a behaviour-oriented contract (e.g., salaries, hierarchical governance) is more efficient than an outcome-oriented contract (e.g., commissions, stock options, transfer of property rights, market governance). Overall, the domain of agency theory is relationships that mirror the basic agency structure of a principal and an agent who are engaged in cooperative behaviour, but have differing goals and differing attitudes toward risk (1989).

The agency theory re-establishes the importance of incentives and self-interest in

organizational thinking. It reminds us that much of organizational life, whether we like it or not, is based on self-interest (Perrow, 1986). The agency theory also makes two contributions to organizational thinking. The first is the treatment of information, which has a cost and can be purchased, according to the agency theory. The implication is that organizations can invest in information systems in order to control agent opportunism. The second is the theory’s risk implications. Organizations are assumed to have uncertain futures. Agency theory extends organizational thinking on this subject by pushing the ramifications of outcome uncertainty to their implications for creating risk. Uncertainty is viewed in terms of risk/reward trade-offs, not just in terms of inability to pre-plan. The implication is that the outcome uncertainty coupled with differences in willingness to accept risk should influence contracts between principal and agent (Eisenhardt, 1989).

Contracts between parties to the firm use accounting numbers to reduce agency costs (Watts & Zimmerman, 1986). Agency cost-reducing contracts include debt contracts between the firm and holders of the firm’s debt, management compensation contracts, employment contracts, and cost-plus sales contracts. Contracting parties demand timely measures of performance and net asset values for compensation and debt contract purposes. They want this, because for

example managerial performance measures in compensation contracts, such as earnings, are more effective when they are timely and reflect the effects of the managers’ actions on firm value in the period in which the actions are taken. Timeliness avoids dysfunctional outcomes associated with managers’ limited tenure with the firm. This will lead to the conclusion that earnings-based management compensation contracts and accounting-based debt contracts generate a demand for timely earnings and net asset measures, in other words accounting conservatism (Watts, 2003a). Furthermore, Eisenhardt (1989) mentions in her study that when boards provide richer

information, timely earnings and net asset measures, top executives are more likely to engage in behaviours that are consistent with stockholders’ interests. When companies use more

conservative accounting methods and provide timely earnings and net asset measures, there will be less agency costs.

In summary, the agency theory has been a large component of the economic theory of the firm for a long time. The agency relationship is defined as ‘a contract under which one or more principals engage the agent to perform a service on their behalf which involves delegating some decision-rights to the agent’. There is a good reason to believe that the agent will not always act in the interest of the principle, because the agent will have urges of self-interest. Therefore, in order to align the incentives of the agent and the principle there are some agency costs involved. The agency theory is concerned with solving two problems (a) the desires or goals of the

principal and agent conflict and (b) it is difficult or expensive for the principal to verify what the agent is actually doing. The principle cannot verify that the agent is behaving appropriate. The agency theory reminds us that much of organizational life is based on self-interest.

2.2 Conservatism

This paragraph consists of an explanation of accounting conservatism. Firstly, I will elaborate on the term conservatism after which I will discuss the different types of conservatism. The focus of the subparagraph will be on unconditional conservatism.

Basu (1997) argues that conservatism has influenced accounting practice for at least 500 years. Furthermore, Sterling (1970) states that conservatism is the most influential principle of valuation in accounting. Recent evidence from empirical research suggests not only that accounting practice is conservative, but also that the practice has become more conservative in the last 30 years (Watts, 2003a). Moreover, evidence shows that listed firms’ financial statements in the U.S. and other Anglo-American law countries are conservative (Basu, 1997; Watts 2003b; Ball, Kothari & Robin, 2000). As stated in the accounting conservatism literature there are a lot of different definitions of the term ‘conservatism’. Bliss (1924) elaborates that accounting

conservatism is traditionally defined by the saying “anticipate no profit, but anticipate all losses”. Moreover, Beaver & Ryan (2005) define conservatism as the on average understatement of the book value of net assets relative to their market value. Furthermore, Feltham & Ohlson (1995) define accounting conservatism as the persistence in underestimating the book value of the firm, implying that book value is less than market value. According to Watts (2003a) anticipating profits means recognizing profits before there is legal claim to the revenues generating them and that the revenues are verifiable. In the accounting conservatism literature the saying is interpreted as representing the accountant’s tendency to require a higher degree of verification to recognize good news as gains than to recognize bad news as losses (Basu, 1997).

Again, conservatism is the asymmetrical verification requirements for gains and losses. Through this interpretation there exists a difference in the level of conservatism. The greater the difference in degree of verification required for gains versus losses, the greater the level of accounting conservatism. An important effect of the asymmetric treatment of gains and losses of accounting conservatism is the ongoing understatement of net asset values. Capital market regulators, standard setters, and academics criticize conservatism because this understatement in the current period can lead to overstatement of earnings in future periods and cause an

understatement of future expenses (Watts, 2003a). Penman & Zhang (2002) argue that

sheet, but also has an effect on the quality of earnings in the income statement. According to them, when a firm increases its investments conservative accounting leads to lower reported earnings than they would be if management used a more liberal accounting choice. These lower earnings create unrecorded reserves which provide managers with flexibility to report more income in the future. Management has the opportunity to increase those reserves, and thereby reduce earnings, if they increase the amount of investments of the firm. On the other hand, management can also reduce the reserves and release them into earnings, by subsequently reducing investments or reducing the rate of growth in investments. Penman & Zhang’s arguments suggest that managers are able to use the joint effect of real activity and accounting policy to manage their earnings.

Lafond & Watts (2008) state that conservatism occurs because it reduces agency costs associated with asymmetries in information and loss functions among contracting parties and because of the inability to verify the more informed parties’ private information. According to them, accounting conservatism is a governance mechanism that reduces the manager’s ability to manipulate and overstate the financial performance and increases the firm’s cash flows and value. They explain this by the fact that manager’s welfare is affected by financial statements and therefore his incentives in other ways as well. The firm’s stock price, the manager’s likelihood of keeping his position, his informally determined compensation, the value of his stock options and the resources under his control are all affected by the firm’s financial statements even in the absence of formal accounting-based contracts. This asymmetric loss function gives the manager incentives to use private information to his own benefit and transfer wealth from investors to himself. This could be done by overstating financial performance and consequently stock prices. Thereby, they acknowledge that the market recognizes that managers have these incentives to use private information to their own benefit. Therefore, uninformed buyers keep the information asymmetry and the adverse selection problem in mind while bidding on shares. This leads to lower bid prices for the shares. Moreover, the managers’ time spent on manipulation and overstatement diverts their efforts from increasing firm value, thereby generating agency costs and reducing the firm value even more. Lafond & Watts conclude that firms with greater conservatism have lower information asymmetry between managers and outside investors. Furthermore, equity investors will demand more conservative earnings as a means of mitigating agency problems (2008). On the contrary, Jackson & Lui (2010) find in their investigation that

conservative accounting could facilitate earnings management. They conclude that this effect of conservatism may partially counterbalance some of its claimed benefits.

In conclusion, recent evidence suggests that the accounting profession is becoming more and more conservative. Moreover, there is also evidence that the financial statements of U.S. firms are conservative. This is helpful for my investigation which will focus on U.S. firms only. As mentioned in the accounting conservatism literature, there is a variety in the explanations of the term ‘conservatism’. The traditional definition, originating from 1924, states that

conservatism means “anticipate no profit, but anticipate all losses”. An important effect of the asymmetric treatment of gains and losses under accounting conservatism is the ongoing

understatement of net asset values. This understatement could lead to overstatement of earnings in future periods. Moreover, conservatism in combination with investment opportunities could enable managers to use this joint effect and manage their earnings. Which leads to lower quality earnings. On the other hand, there are also positive effects of conservatism. According to the Lafond & Watts (2008) paper, conservative reporting leads to lower information asymmetry between managers and outside investors. This results in equity investors demanding more conservative earnings as a means of mitigating agency problems. Finally, most evidence questions the overall benefits of conservatism, as a lot of counterbalancing costs have been found.

2.2.1. Different forms of accounting conservatism

In the literature, accounting conservatism is manifested in two distinct ways (Beaver & Ryan, 2005). The first is conditional conservatism and the second unconditional conservatism. The foremost difference between these is that the application of conditional conservatism depends on economic news events, while the application of unconditional conservatism does not. When negative economic news is recognized in a timelier manner in accounting earnings than positive economic news, conditional conservatism exists. Therefore, it can be said that conditional conservatism is characterized by the asymmetric recognition of positive and negative economic news (Ruch & Taylor, 2015). Furthermore, Beaver & Ryan (2005) define conditional

conservatism as not writing up book values under favourable circumstances, but do write down book values under unfavourable circumstances.

According to Ruch & Taylor, unconditional conservatism occurs through consistent under-recognition of accounting net assets. Dissimilar to conditional conservatism, unconditional conservatism does not depend on economic news events. Examples of unconditional

conservatism are: accelerated depreciation methods, expensing R&D costs, expensing advertising costs, accumulated reserves in excess of expected future costs (e.g., allowance for doubtful accounts, warranty allowance) and the use of LIFO inventory accounting (2015). They further emphasize the importance to distinguish between these two kinds of conservatism for three main reasons. The first being that these two forms have different effects on the financial statements. The accounting policies consistent with unconditional conservatism are expected to have a relatively consistent impact on the income statement from period to period, while the application of conditional conservatism is more likely to be transitory on the income statement because of fluctuations in the content and timing of economic news across periods (Chen, Folsom, Paek & Sami, 2013). Conditional and unconditional conservatism both result in understated net assets on the balance sheet. However, they do have different effects on the timing of income statement recognition, and in turn, different effects on the timing of balance sheet recognition. Moreover, the application of one type of conservatism affects the application of the other type. This is investigated by Beaver & Ryan (2005), they found that unconditional conservatism creates “accounting slack” that may exclude the use of conditional conservatism. Through the

unconditional understatement of assets the extent of write-downs recognized in the presence of bad news events is limited. Therefore, reduces observed asymmetric timeliness in earnings. As a consequence, conclusions made regarding the presence of conditional conservatism may be entangled by the presence of unconditional conservatism. This result is also found in another investigation which concludes that unconditional conservatism decreases conditional

conservatism and that the two forms meet different needs of firms (Qiang, 2007). Furthermore, the conditions that give rise to conditional conservatism may differ from those of unconditional conservatism (Ruch & Taylor, 2015). An investigation of Qiang (2007) showed that conditional conservatism arises in settings where contracting and litigation costs are high, other than

unconditional conservatism, which arises in settings where litigation, regulatory, and tax costs are high. Nevertheless, Beaver & Ryan elaborate in their investigation that the two types of

conservatism have many of the same purposes, including capturing investors and other’s perceived asymmetric loss functions, minimizing firm’s litigation, tax, or regulatory costs, and enabling accounting and industry regulators to minimize economic instability and avoid criticism (2005).

In this investigation I will focus on unconditional conservatism, which hasn’t had as much attention in empirical research as conditional conservatism. The use and effect of unconditional conservatism is easiest to observe in the use of LIFO accounting for inventories. Accounting rates of return are usually higher under LIFO, because of rising inventory prices. When using this method the company builds a LIFO reserve over time, which consists of the difference between LIFO and FIFO carrying values. The difference between LIFO and FIFO costs of goods sold is equal to the change in the value of the LIFO reserve if purchases of inventories increase relative to sales (with continuing rising purchase prices). This also results in lower earnings under the LIFO method than using the FIFO method for inventory. The build-up of the LIFO inventory creates the reserve, and the decline in inventory, known as LIFO dipping, is a liquidation of the reserve. The same thing happens, although less transparently, for all forms of unconditional conservatism (Penman & Zhang, 2002). According to Ruch & Taylor (2015), unconditional conservatism has the ability to facilitate earnings management through this accumulation of reserves on the balance sheet which can be released into earnings when an earnings target needs to be met. Managers have the capability to increase production more than necessary in order to increase earnings when dollar inventories are lower. When dollar inventories are low and the company keeps producing products they are able to diminish the LIFO reserve which leads to more earnings, because the inventory amount decreases. My focus in this investigation is to see if managers make use of this capability and actually manage their earnings using a conservative accounting method.

In brief, the literature defines two distinct forms of accounting conservatism, namely conditional and unconditional conservatism. The biggest difference between the two is the dependence on news events of conditional conservatism, which unconditional conservatism does not. There are three reasons why it is important to distinguish between these two forms of

conservatism. First of all, is that they have different effects on the financial statements. Secondly, the application of one type of conservatism affects the application of the other type. Lastly, the

conditions that give rise to conditional conservatism may differ from those of unconditional conservatism. The focus of this investigation will be on unconditional conservatism and will look at the use of LIFO accounting for inventory in combination with the use of earnings management. The use of LIFO accounting results in a LIFO reserve on the balance sheet which represents the difference between the carrying values of LIFO and FIFO inventory. Through overproduction and the use of this reserve, managers are able to manage their earnings.

2.3 Earnings management

The following paragraph consists of an explanation of earnings management as a whole. After that description I will elaborate more about the different kinds of earnings management and focus on the real activities manipulation.

Managers have the opportunity to use their knowledge about the business and its opportunities to choose the best reporting methods, estimates, and disclosures that match the firms’ business economics and potentially increase the value of accounting as a form of

communication. On the other hand, auditing is imperfect, therefore managers are also able to use this judgement to create opportunities for earnings management. In this case, managers choose reporting methods and estimates that do not accurately reflect their firm’s underlying economics (Healy & Wahlen, 1999). Healy & Wahlen define earnings management as: the use of managerial judgement 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. They mention that there are many ways in which managers can exercise judgement in financial reporting. This is because management must choose among acceptable accounting methods for reporting economic transactions, such as the straight-line or accelerated depreciation methods or the LIFO, FIFO, or weighted-average inventory valuation methods (1999).

However, Stein (1989) declares that this definition frames the objective of earnings management which is to mislead stakeholders about the underlying performance of the firm. This can be the case if managers believe that stakeholders do not undo earnings management.

Furthermore, this could also happen if managers have access to information which is not

outsiders. According to Stein, stakeholders are then likely to anticipate a certain amount of earnings management.

There are a few reasons for managers why they would consider using earnings

management. These include: capital market expectations and valuation, contracts written in terms of accounting numbers, and anti-trust or other government regulation. In order for managers to decide if they want to use earnings management they will first have to consider the costs and benefits which are involved in this decision. The costs include the potential misallocation of resources that arises from earnings management. The benefits comprise of improvements in management’s credible communication of private information to external stakeholders, and improvements in resource allocation decisions (Healy & Wahlen, 1999).

In short, it is stated that managers have the opportunity to use their knowledge in a good or bad way. They have the chance to choose the right reporting methods for their company or use their judgement to create opportunities for earnings management. Earnings management is most commonly used to mislead stakeholders about the underlying economic performance of the company or in order to influence contractual outcomes that depend on reported accounting numbers. However, it is only possible for managers to mislead stakeholders if they believe that there is no use of earnings management within the company. If this is not the case, stakeholders will anticipate on the use of earnings management. The reasons why managers would use earnings management are mostly to uphold analysts’ expectations, don’t make any contract breaches or improve their own welfare within these contracts, or to be able to comply with anti-trust or other governmental regulation.

2.3.1 Different kinds of earnings management

Gunny mentioned in her article that earnings management can be classified into three categories: fraudulent accounting, accruals management and real activities manipulation. First, fraudulent accounting involves accounting choices that violate GAAP (2005). Second, accruals management involves within-GAAP choices that try to obscure or mask true economic performance(Dechow & Skinner, 2000). Third, real activities manipulation occurs when managers undertake actions that deviate from the first best practice to increase reported earnings (Gunny, 2005). Schipper

(1989)was the first one to include real activities manipulation into the definition of earnings management. She described earnings management as “a purposeful intervention in the external

financial reporting process, with the intention of obtaining some private gain, minor extension of this definition would encompass ‘real’ earnings management, accomplished by timing investment or financing decisions to alter reported earnings or some subset of it”. Furthermore,

Roychowdhury (2006) defines real activities manipulation, as “departures from normal

operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations”. He

mentions that these departures of normal operational practices do not necessarily contribute to firm value but they do help managers to meet reporting goals.

Although accruals management is less costly, with respect to future firm value, there are several reasons why managers engage in real activities manipulation. Firstly, aggressive

accounting choices with respect to accruals are at higher risk for scrutiny of accounting practice post-SOX, in comparison to real activities manipulation which will draw less attention to auditors or regulators. Secondly, through limited accounting flexibility because of accrual manipulation in prior years and shorter operating cycles, managers use real activities manipulation to a greater extent. Thirdly, relying on accrual manipulation alone entails a risk. Moreover, operating decisions are controlled by the managers, whereas accounting treatments must meet the requirements of the auditor (Gunny, 2005; Roychowdhury, 2006; Zang, 2011). Moreover, Graham et al. (2005) also find in a survey that financial executives attach a high importance to meeting earnings targets and that they are willing to manipulate real activities to meet these targets, even though the manipulation potentially reduces firm value. Roychowdhury elaborates that real activities manipulation can reduce firm value because actions taken in the current period to increase earnings can have a negative effect on cash flows in future periods. He also mentions that even though there are certain costs associated with real activities manipulation, executives are unlikely to rely solely on accrual manipulation to manage earnings (2006). On the other hand, managers will use more accrual-based earnings management and less real activities manipulation when the latter is more costly for them, due to having a less competitive status in the industry, being in a less healthy financial condition, experiencing higher level of monitoring form institutional investors, and incurring greater tax expenses in the current period (Zang, 2011).

Referring to the different reasons managers have to use either accrual-based earnings management, real activities manipulation or a combination of both, Zang (2011) found that there is a direct substitutive relation between real activities manipulation and accrual-based earnings management. She shows that real manipulation is positively correlated with the costs of accrual manipulation, and that accrual and real manipulations are negatively correlated.

Summarizing, earnings management consists of three categories: fraudulent accounting, accruals management and real activities manipulation. Real activities manipulation is described as a departure from normal operating activities to mislead at least some stakeholders into believing that certain financial reporting goals have been met. Even though accrual-based earnings management is less costly with respect to firm value, managers do have reasons to use real activities manipulation. The foremost reasons is that it draws less attention from auditors and regulators with respect to accounting scrutiny and violating accounting standard rules. Managers have the choice of using accrual-based earnings management, real activities manipulation or a combination of both if they are looking for ways to boost their earnings. However, a study concluded that a combination of both is not likely to occur due to the fact that accrual-based earnings management and real activities manipulation are negatively correlated.

2.3.2 Reasons to use earnings management with respect to accounting conservatism

Chen, Hemmer & Zhang mention that conservatism introduces additional noise in the accounting numbers, which reduces the stewardship value of accounting data and therefore increases the cost of motivating the manager. However, they show that under reasonable conditions, the

conservative standard actually reduces earnings management so much that the noise introduced by the conservative standard is more than offset by the reduction in earnings management. This overall net reduction in noise makes the reported numbers more informative about managerial actions (2007). On the other hand, critics of conservatism elaborate that conservatism can

facilitate earnings management (AICPA, 1939; Devine, 1963; FASB, 1980; Levitt, 1998; Penman & Penman, 2007). Furthermore, Jackson & Lui (2010) mention that conservatism induces

problems that may partially counterbalance some of its claimed benefits. They have investigated this through looking at bad debt expenses. Also, Penman & Zhang (2002) investigated the

relation between accounting conservatism, the quality of earnings, and stock returns. They elaborate that the quality of earnings depends on the interaction between real activity and accounting policy, not merely on changes in accounting methods and estimates. This suggests that managers can use the joint effect of real activity and accounting policy to manage earnings. But, Jackson & Lui (2010) comment that there is little evidence supporting the claim made by the critics of conservatism. Therefore, I would like to contribute to the existing evidence by

investigating the relation between real activities manipulation through overproduction and the conservative use of LIFO accounting.

2.4 Sarbanes-Oxley Act

Hereafter, I will elaborate on what the Sarbanes-Oxley Act is. Furthermore, I will explain when and why it was implemented and what the main objectives where of the implementation.

A wave of corporate governance failures raised concerns about the integrity of the accounting information provided to investors and resulted in a drop in investor confidence (Jain et al., 2003; Jain and Rezaee, 2006; Rezaee, 2004). These failures were highly publicized and ultimately led to the passage of the Sarbanes-Oxley Act (SOX), which was initially designed to fix the auditing of U.S. public companies. This is consistent with the official name of the law: the Public Company Accounting Reform and Investor Protection Act of 2002 (Coates, 2007).

Congress passed SOX in July 2002, in response to a number of high-profile scandals starting in late 2001. According to Romano, the act has been widely considered the most far-reaching securities legislation since the Securities Acts of 1933 and 1934. It not only imposes additional disclosure requirements, but more importantly, proposes substantive corporate governance mandates, a practice that is unprecedented in the history of federal securities legislation (2004). Sarbanes-Oxley created a unique, quasi-public institution to oversee and regulate auditing, the Public Company Accounting Oversight Board (PCAOB). Their first task was to implement a second core goal, which was to enlist auditors to enforce existing laws against theft and fraud by corporate officers. New rules concerning the auditor-firm relationship, auditor rotation, auditor provision of non-audit services, and corporate whistle-blowers reinforce this core. Through the implementation of SOX there were new incentives for firms to spend money on internal controls, above and beyond the increases in audit costs that would have occurred after the corporate

scandals of the early 2000s. In exchange for these higher costs, Sarbanes-Oxley promises a variety of long-term benefits. Investors will face a lower risk of losses from fraud and theft, and benefit from more reliable financial reporting, greater transparency, and accountability.

Moreover, public companies will pay a lower cost of capital, and the economy will benefit because of a better allocation of resources and faster growth. However, the law’s full costs are hard to quantify, and the benefits even harder (Coates, 2007).

According to Zhang (2007), SOX aims to prevent deceptive accounting and management misbehaviour by requiring more oversight, imposing greater penalties for managerial conduct, and dealing with potential conflicts of interest. Moreover, SOX exposes managers and directors to greater litigation risks and stiffer penalties. Ribstein argues that CEOs allegedly will take less risky actions, consequently changing their business strategies and potentially reducing firm value (2002). One of the main objectives of SOX was to restore the integrity of financial statements by reducing earnings management and accounting fraud. Therefore, the extent of earnings

management prior to SOX and the effect of SOX on earnings management is an important research topic (Cohen, Dey & Lys, 2008).

Prior literature documents that managerial tendency to manage earnings and to avoid negative earnings surprises has increased significantly over time (Brown, 2001; Bartov et al., 2002; Lopez & Rees, 2001; Matsumoto, 2002; Brown & Caylor, 2005). Although the main purpose of the implementation of SOX was to reduce earnings management, Cohen, Dey & Lys found that the use of earnings management increased steadily over the period 1987-2005. Furthermore, meeting or beating prior year’s earnings numbers, consensus analysts’ forecasts, and avoiding losses continued to be important incentives to manage earnings. They conclude that the use of earnings management increased after the implementation of SOX. Even though the use of accrual-based earnings management declined significantly after the implementation of SOX, real activities management increased significantly according to Cohen, Dey & Lys (2008). A survey of Graham et al. (2005) showed that firms switched to managing earnings using real methods, possibly because these techniques, while more costly, are likely to be harder to detect.

In conclusion, the scandals regarding corporate governance failures in the early 2000s raised concerns about the integrity of accounting information which resulted in a drop of investor confidence. Deceptive accounting and management misbehaviour should be reduced through the

implementation of SOX. Moreover, the implementation created new incentives for firms to spend money, which would result in a variety of long-term benefits. These include: a lower risk of losses from fraud and theft, more reliable financial reporting, greater transparency and

accountability, lower costs of capital, and improvement in the overall economy through a better allocation of resources and faster growth. The main goal of SOX is to restore the integrity of the financial statements through a reduction in earnings management and accounting fraud. Even though this was the main purpose, there is evidence that the use of earnings management increased after the passage of SOX.

2.5 Hypotheses development

In this paragraph I will form my hypotheses based on the discussed prior literature. To detect real activities manipulation, I will investigate patterns in production costs for firms that use LIFO accounting for inventory.

According to Ruch & Taylor (2015) and Penman & Zhang (2002), LIFO accounting for inventories is more conservative relative to FIFO accounting. If firms use LIFO accounting a reserve will be formed to represent the value difference between using the LIFO and FIFO accounting method. LIFO accounting carries inventories on the balance sheet at lower amounts than FIFO if inventory prices have risen in the past. In that case, it is more conservative than the alternatives. Earnings are not affected by LIFO if dollar inventories are unchanged for the earnings period: cost of goods sold is equal to the purchase for the period, as it is with FIFO. On the other hand, if dollar inventories increase either through physical inventory growth or

inventory price changes, earnings are lower under LIFO. Therefore, the LIFO reserve increases. If dollar inventories decline, earnings are higher under LIFO. This leads to a decrease in the LIFO reserve and earnings are increased as the result of a decline in inventory. Managers in firms that use LIFO accounting for inventory could engage in real activities manipulation using this reserve and the development of it. This could be done by overproducing.

Overproduction is the manipulation method that I will investigate. Overproduction is measured by looking at the production costs of a company. Production costs are defined as the sum of COGS and the change in inventory during the fiscal year. This definition generates “production” costs for non-manufacturing firms, although the terminology does not literally

apply to such firms. Examining production costs instead of COGS has an advantage which entails that accrual manipulation to lower reported COGS through the inventory account, for instance by delaying write-offs of obsolete inventory, should not affect production costs. Consequently, production costs should primarily reflect the effects of real activities.

Overproducing means the reporting of lower cost of goods sold through increased production. Managers are able to increase production more than necessary and could use this method to manage earnings upward. When managers produce more units, they can spread the fixed overhead costs over a larger number of units, thus lowering fixed costs per unit.

Nevertheless, the firm incurs production and holding costs on the over-produced items that are not recovered in the same period through sales. The incremental marginal costs incurred in producing the additional inventories result in higher annual production costs relative to sales (Roychowdhury, 2006). In the case of lowering prices of inventory managers can use this when applying the LIFO accounting method to overproduce which leads to bigger differences in the LIFO and FIFO method and causes the LIFO reserve to shrink. This amount is added into earnings for the company. In order to see if this really happens in firms as is suspected by Ruch & Taylor (2015), I would like to test the following hypothesis:

H1A: Managers in firms that report their inventory according to the LIFO

method use overproduction relative to firms that report their

inventory according to other accounting methods

Moreover, according to Roychowdhury (2006) overproduction and price discounts can both generate high production costs relative to sales. Both manufacturing and non-manufacturing companies can offer price discounts to boost sales but overproduction as an earnings management strategy is only available to firms in manufacturing industries. Therefore, the evidence of

abnormal production costs should be driven to a greater extent by LIFO firm-years that belong to manufacturing industries. This will be true if price discounting and overproduction by LIFO manufacturing firms have a greater effect on production costs than price discounting by LIFO non-manufacturing firms. To investigate this effect the follow-up of my first hypothesis will be:

Furthermore, as discussed in the literature there has been an apparent change in the use of earnings management after the implementation of SOX. Although the intention was to reduce earnings management after the implementation, this is not always observed. In multiple investigations is seen that there has been a change in the kind of earnings management that is primarily used by managers. Before the implementation the most common form of earnings management used was accrual-based earnings management, but after the implementation this changed into real activities manipulation. In order to investigate if the amount of real activities manipulation used by managers in firms that apply LIFO accounting has also changed after the implementation of SOX, I would like to test the following hypothesis:

To test my hypotheses I will use corporate information about U.S. companies, which are obliged to report the value of the LIFO reserve in the footnotes of the financial statements if they use the LIFO accounting method for inventory. Up until now, there has been no study to my knowledge which has investigated the use of LIFO accounting for inventories on the likelihood of engaging in real activities manipulation activities. The next section discusses the empirical methodology used in this study.

H1B: The effect on overproduction of the use of LIFO accounting for

inventories is more severe in manufacturing industries.

H2: Overproduction by managers in firms that use LIFO accounting

3. Research methodology

3.1 LIFO accounting for inventories and real activities manipulation

The definitions for earnings management that are mentioned in prior literature indicate that managers make use of judgement in order to manage earnings:

“ Earnings management occurs when managers use judgement 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.” (Healy & Wahlen, 1999)

This definition makes clear that managers have some opportunity to exercise judgement within the application of accounting standards. There are three methods of using earnings management: fraudulent accounting, accrual-based earnings management and real activities manipulation (Gunny, 2005). In this paper I will investigate the third method looking at overproduction. Real activities manipulation is defined as:

“Departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the

normal course of operations.” (Roychowdhury, 2006)

Using conservative accounting methods should limit the use of this discretion (Chen, Hemmer & Zhang, 2007). However, prior research indicates that this should not be the case for the use of LIFO accounting as indicated by Ruch & Taylor (2015). Although this has never been investigated by anyone before, it is believed to be true. In order to see if this believe has not been misplaced, I will investigate if the use of LIFO accounting will amplify the use of overproduction to manipulate earnings.

In order to answer my previously mentioned hypotheses, I will use the model of

Roychowdhury (2006) to measure the use of real activities manipulation. I will collect my sample from the COMPUSTAT database for the period 1998-2013. I require market value to be available on COMPUSTAT, this restricts my sample to the post-1998 period. The sample will be divided into two main groups: firm-years with a LIFO reserve as an indicator for the use of conservative accounting and firm-years which do not have a LIFO reserve. Furthermore, for the test of hypotheses 1B I will add an extra variable which will indicate if firms belong to the

period pre-SOX (1998-2001) and post-SOX (2002-2013). I will use annual data for my investigation.

I eliminate firms which do not have all necessary data available, firms in regulated industries (SIC codes between 4400 and 5000) and banks and financial institutions (SIC codes between 6000 and 6500). Imposing all data-availability requirements yields 49,569 firm-years over the period 1998-2013, of which 5,268 are observations from firms that use LIFO accounting for inventory and 44,301 observations from other firms.

3.2 Estimation models

Following Dechow, Kothari & Watts (1998), expenses are expressed as a linear function of contemporaneous sales. The model for normal COGS is estimated as:

COGSt/At-1= α0 + α(1/At-1) + β(St/At-1) + ԑt. (1)

Similarly, following Dechow, Kothari & Watts (1998), the model for ‘normal’ inventory growth is estimated by the following regression:

∆INVt/At-1 = α0 + α(1/At-1) + β1(∆St/At-1) + β2(∆St-1/At-1) + ԑt. (2)

The production costs are defined as PRODt = COGSt + ∆INVt. Using (1) and (2), the

normal production costs are estimated from the following regression:

PRODt/At-1 = α0 + α1(1/At-1 ) + β1 (St/At-1 ) + β2(∆St/At-1 ) + β3(∆St-1/At-1) + εt. (3)

According to Roychowdhury (2006) it is general convention in the literature to include a scaled intercept, α(1/A_t-1 ), when estimating non-discretionary accruals. Eliminating the unscaled intercept does not materially affect the results, nor does retaining the unscaled intercept. At-1

represents the total value of assets at the end of the period t, St is the sales for the period, ∆S_t-1 the

Regression (3) is used for the calculation of the normal production costs which is the outcome of the formula. The normal production costs are then used to estimate the abnormal production costs which will be used to calculate the overproduction of companies. The normal production costs are deducted from the total production costs (PRODt) of companies which

results in the abnormal production costs.

For the data collection for this regression model I started with the collection of the value of total assets of the previous year-end to use as the variable At-1. For the change in sales I

collected the amount of sales of the current year and the one of the previous year and subtracted that to come to the ∆St variable. For the ∆St-1 I collected the amount of sales of the previous year

and the year before that and subtracted that amount.

3.3 Descriptive Statistics 3.3.1 Firm Characteristics

Table 1 presents descriptive statistics comparing the LIFO firms to the rest of the sample which are defined as FIFO firms. The mean market capitalization of the LIFO firms, around $5.8 billion, is more than double the market capitalization of the full sample, which is around $1.9 billion. This implies that firms that use LIFO accounting for inventory have a much higher value of equity on the market than firms that use another method to account for inventory. Furthermore, the amount of total assets of around $6.7 billion is more than four times the amount of total assets of the full sample ($1.5 billion). The LIFO firms have a significantly higher ratio of market value of equity to book value of equity than the overall sample (4.36 versus 2.43). The only explanation I could think of that could cause these differing results is that firms that use LIFO accounting for inventory tend to be larger than firms that use other methods to account for their inventory.

The production costs are significantly higher for LIFO firms compared to the full sample as can be seen in Table 1. Furthermore, the production costs compared to sales are equal to 75% for LIFO firms and 69% for the rest of the sample which is also an indicator for overproduction according to Roychowdhury (2006), because high production costs relative to sales imply high holding costs on overproduced inventory which are not recovered in the same period through sales. Furthermore, mean inventory turnover ratio is significantly lower for LIFO firms at 9.66%, consistent with the idea that LIFO firms lower their reported COGS by overproduction.

Table 1

Descriptive statistics LIFO versus FIFO

LIFO firms FIFO firms

Mean Mean Difference in Means (t-stat) MVE ($ million) 5790.64 1888.06 -3902.58 (-24.22)*** MVE/BVE 4.36 2.43 -1.93 (-1.51)

Total assets ($ million) 6660.64 1456.87 -5203.78 (-35.87)*** Sales ($ million) 6918.75 1338.85 -5579.91 (-47.53)*** IBEI ($ million) 314.55 69.53 -245.02 (-24.97)*** Production costs ($ million) 5190.94 929.66 -4261.28 (-47.47)*** Inventory turnover 9.66 24.81 15.15 (12.71)*** Sales/A 1.41 1.20 -0.21 (-13.47)*** Production costs/A (%) 128.51 122.40 -6.11 (-0.80) IBEI/A (%) 4.54 -22.02 -26.56 (-7.78)*** Number of observations 49569 LIFO firms 5268 FIFO firms 44301

***Significant at the 1% level.

The sample period spans 1998-2013. LIFO firms are firm-years with a reported LIFO reserve and FIFO firms are all other firms in the sample. The numbers in parentheses are t-statistics for the differences in means. All descriptive statistics are reported for the full sample of 49,677 firm-years. Please see Appendix A for variable descriptions.

Table 2 presents descriptive statistics comparing LIFO firms before the implementation of SOX to the period after the implementation of SOX. The mean capitalization of Post-SOX firms-years is significantly higher than that of Pre-SOX firm firms-years. Furthermore, the total amount of assets is more than double for the period after the implementation of SOX. But the proportion of sales compared to total assets has remained the same, which implies that the growth of sales was the same as the growth of total assets. Also, the production costs are significantly higher in the post-SOX period, but the percentage of the production costs of assets is significantly lower. This implies that the change in the amount of total assets has changed more than the amount of production costs, which is not complying with the expectations of more use of real activities manipulation after the implementation of SOX. On the other hand, mean inventory turnover is significantly lower for the Post-SOX firm-years, which is consistent with the idea that after the implementation of SOX, firms relied more on real earnings manipulation. The higher mean inventory turnover rate implies that firms lowered their COGS through overproduction after the implementation of SOX. These results are contradicting and need to be examined further.

Table 2

Descriptive Statistics pre-SOX versus post-SOX

Pre-SOX Post-SOX Mean Mean Difference in Means (t-stat) MVE ($ million) 3681.30 6953.38 -3272.08 (-6.90)*** MVE/BVE 0.14 6.68 -6.54 (-1.98)**

Total assets ($ million) 3463.48 8423.04 -4959.56 (-6.85)*** Sales ($ million) 3943.06 8559.06 -4616 (-8.38)*** IBEI ($ million) 142.45 409.42 -266.97 (-7.29)*** Production costs ($ million) 4356.07 5651.15 -1295.07 (-3.05)*** Inventory turnover 12.13 8.30 3.84 (9.08)*** Sales/A 1.41 1.41 - (-0.21) Production costs/A (%) 168.80 106.30 62.50 (13.75)*** IBEI/A (%) 3.79 4.95 -1.16 (-2.98)*** Number of observations 5268 Pre-SOX firm-years 1872 Post-SOX firm-years 3396

***Significant at the 1% level. **Significant at the 5% level.

The sample period spans 1998-2013. Pre-SOX years are years between 1998-2001 and Post-SOX firm-years are between 2002-2013. The numbers in parentheses are t-statistics form t-tests for the differences in means. All descriptive statistics are reported for the LIFO firms sample, consistent of 5,268 firm-years.

3.3.2 Regression model

Table 3 reports the regression coefficients for the key regression used to estimate “normal” levels of production (see section 3.2). I estimated these models using the entire sample for the tests of LIFO firms versus FIFO firms and for the second test between the pre-SOX and post-SOX period. The table reports the coefficients of the models used in the calculation of the “normal” levels of production. The results of this model are further used to calculate the abnormal production costs of the firm-year observations which are the dependent variable in the main regressions of my investigation. These numbers are used in the following regressions to see the effect of the use of LIFO inventory on the use of real activities manipulation through

overproduction. As can be seen in Table 3, all coefficients of the regression model are significant and the adjusted R2 is equal to 0.85, which implies that 85% of the variances are explained by the variables used in the model. Appendix B1 includes the correlation matrix for the various

variables. No high correlation is present.

The effect of the sales of the current year and the effect of the change in the sales of prior year are positive on the “normal” level of production costs, as can be seen in Table 3.

Furthermore, the effect of the change of sales from the current year is negative. These

coefficients are different from the study that Roychowdhury (2006) performed. In his study the effects of the current year sales and the change of sales in the current year had positive effects on the “normal” level of production costs. Moreover, only the effect of the change of sales in the prior year had a negative effect. Although my results appear to be different from his

Table 3 Model Parameters Intercept 1.0389 (34.28)*** 3.4423 (165.11)*** 0.0526 (17.10)*** -0.3398 (-47.13)*** 0.0071 (113.25)*** 0.85

***Significant at the 1% level.

This table reports the estimated parameters in the following regression:

(a) PROD_t/A_t-1 = α₀ + α₁(1/A_t-1 ) + β₁ (S_t/A_t-1 ) + β₂(∆S_t/A_t-1 ) + β₃(∆S_t-1/A_t-1) + ε_t.

The regression is estimated for the full sample of 49,569 firm-years. Please see Appendix A for variable descriptions.