The role of conditional conservatism in firm investment efficiency : evidence from European high-tech companies

The Role of Conditional Conservatism in Firm Investment

Efficiency—Evidence from European High-Tech Companies

Name: Xuting Xiao

Student number: 11758279

Thesis supervisor: Dr. Sanjay Bissessur Date: June 23, 2018

Word count: 12,876

MSc Accountancy & Control, Accountancy

Statement of Originality

This document is written by student Xuting Xiao 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

I argue that conservatism improves investment efficiency for high-tech firms in Europe. In particular, I predict that it resolves debt-equity conflicts, facilitating a firm's access to debt financing and limiting both overinvestment and underinvestment. This permits the financing of prudent investments that otherwise might not be pursued. The empirical results partly confirm these predictions. I find that more conservative firms reduce investment in settings prone to overinvestment, and continue to reduce investment in settings prone to underinvestment, and they issue more debt in both settings.

Keywords:

Conservatism, asymmetric timeliness, investment efficiency, under-investment, over-investment

Contents

2.4 Prior literature on the relation between conservatism and investment efficiency ... 14

3. Hypothesis development ... 15

4. Research design ... 17

4.1 Conservatism and investment efficiency ... 17

4.2 Samples and data sources ... 20

5 Results ... 24

5.1 Conservatism and investment efficiency ... 24

5.2 Conservatism and debt financing ... 27

6 Conclusions ... 29

1. Introduction

FASB excluded conservatism as an attribute to the quality of financial reporting because it is believed to dampen the credibility of accounting information (FASB, 2010). Proponents of this view contend that this invalid information could hinder management from making the right choices. (Gigler, Kanodia, Sapra, & Venugopalan, 2009; Guay & Verrecchia, 2006). The opponents of this view, however, suggest that its development is necessary among misaligned parties and could serve as a contracting mechanism (Basu, 1997; Watts, 2003a). The influence of conservatism on accounting practice is a major area of findings among extant literature. Of those literature, the main theory tested in this paper is the work of Watts (2003) proposing explanations for conservatism. Watts articulated four explanations for conservatism: contracting, shareholder litigation, taxation and accounting regulation, the former two playing a major role. Ruch and Taylor (2015) interpret the debate on the cost and benefits of accounting conservatism as the existence of different perspectives on the function of conservatism. They argue that there are two main functions of conservatism—valuation and contracting. The former focus on the adequacy of conservative information to reflect market value of a firm and serve decision-making process, while the latter maintains conservatism as a way of assessing the fulfillment of obligations of management and resolve agency conflicts among contracting parties. Investment efficiency is one of the fundamental questions of finance literature. According to Modigliani and Miller (1958) paradigm, optimal investment decision is reached when all positive net present value projects are pursued, presuming that firms are likely to acquire financing for these projects and will continue investing until marginal benefits equal marginal costs.

In this paper, I try to examine the relation between conditional conservatism and investment efficiency.

RQ: Does conditional conservatism contributes to investment efficiency?

Specifically, this thesis explores the effect of conditional conservatism (hereafter CC) on investment efficiency in the following ways: (1) by reducing both underinvestment and overinvestment; (2) by facilitating debt financing. According to Basu (1997), more verification is required to recognized gains than losses, thus causing a timing asymmetry in a manner that bad news are recognized sooner in earnings than good news. I hypothesize that accounting conservatism influence investment efficiency in two aspects. First, Conservatism is negatively associated with information asymmetry and negative NPV projects, and

conservatism is negatively associated with agency conflicts and overinvestment. Results from prior studies indicate that the 'informational role' of conservative reporting may be a reaction to information asymmetry (La Fond and Watts, 2008; Khan and Watts, 2009) and to some extent mitigate information asymmetry, thereby aligning interests through restricting the self-interest behaviors of contracting parties (La Fond and Watts, 2008). To put more specifically, early signal reflected from a timelier disclosure on bad news indicating declining debt value strengthens controllability of debt holders while the possibility of rents extracting behaviors of shareholders and managers is significantly minimized (Ball and Shivakumar 2006; Nikolaev 2010). Moreover, conservative accounting is demonstrated to show efficiency in resolving agency conflicts between managers and shareholders. Ball and Shivakumar (2005) contend that since losses are recognized sooner, managers realize the possibility of delaying the losses until the tenure of their successors is minimal, and it follows that more cautious considerations are taken into account to avoid choosing negative NPV projects which would harm firm value (e.g. 'empire building strategies', 'pet' projects or 'trophy' acquisitions). Second, conservatism is positively associated with debt issuance. As mentioned above, the informational role of conservatism mitigates agency conflicts between debt holders and managers, and as a result, a more prudent accounting mechanism may build trust in debt investors that they are more informed about the underlying risk of debt value of the firm. Furthermore, it is argued that through facilitating the issuance of debt, conservatism brings in more investment opportunities which may be abandoned without the enhanced financing opportunities.

I study the association between conditional conservatism and investment efficiency using an European sample of 491 firm-year observations for the period 1997-2017. I follow the methodology in Biddle et. al (2009) and analyze the association between investment efficiency and a proxy of firm-level conservative reporting CONS, which is based on the work of Khan and Watts (2009). The analysis yields two findings. First, as predicted, conservatism contributes to reducing overinvestment, which is also consistent with findings in other literature. Seen from this result, during the period from 1997 to 2017, high-tech firms in Europe which were prone to over-invest, or managers prone to make aggressive investment decisions, were highly likely to cut back on investment after adopting conservatism accounting. However, contrary to prediction and findings in other literature, adopting conservatism seems to continue reducing investment when firms tend to under-invest. One possible explanation could be that the deviation from prediction might be due to relative

small set of control variables. For example, accrual quality (FRQ) and governance quality (GOV) are two factors that could influence investment according to Beatriz et. al (2010) and Juan et. al (2016). Besides, there are many other optional proxies for measuring conservatism, such as a firm-specific measurement of conservatism, which is also widely adopted in other literature. Moreover, since the samples of investigation are based in Europe, another explanation for this controversy could be due to policy reasons—difference between the Continental Law System and the Anglo-American Law System. Finally, as predicted, conservatism positively influence debt issuance, which is consistent with findings in other literature. For example, according to Juan et. al (2016), it is discovered that there is a positive association between conservatism and future change in debt issuance in settings prone to underinvestment, and in terms of economic significance, a one-decile change in conservatism results in an increase in future debt issuance of 2.6%, relative to average total debt, among firms that are under-investing. This finding validates that adopting conservatism contributes to the conflicts on asymmetric pay-offs between debt holders and management since adopting conservative accounting means that debt holders are more and better informed about the underlying economic performance and whether the obligations of management are well performed.

Providing an answer to this research question is important. First, in the work of FASB and IASB, the role of accounting conservatism was questioned as a desirable quality of financial information (FASB 2008), and this opinion23 actuate the debate on the economic implications of accounting conservatism and the potential issues absent conservatism. This thesis attempts to inform on the debate and provide a cost-benefit analysis and contribute to the role of conservative accounting. Finally, if conservatism indeed contributes to enhanced investment efficiency, then knowing how it works might shed light on firms seeking this accounting mechanism as a way of achieving persistent growth by gaining a solid reputation among investors.

Second, payoffs among various parties within an organization tend to be imbalanced and asymmetric, agency conflicts occurs when managers have private information and misaligned interest against that of outside investors. When managers have self-interest to make value-destroying decisions (e.g. taking on negative NPV projects to build their own 'empire' while jeopardizing firm value) or prefer to disclose information on good news sooner while hide or delay disclosures on bad news, investors withdraw their resources or stop providing funds to the firm due to IA and their uncertainty on future expectations of the firm. Eventually,

managers shoulder all costs resulted from failures to attract investment, and one countermeasure they adopt to avoid this situation is providing more conservative accounting information. This thesis aim to investigate whether adopting accounting conservatism would achieve aligned interests of different parties by examining the causality between CC and investment efficiency.

Moreover, an understanding on the role of conservatism might inform investors on how to direct their resources flow more efficiently and inform standard setters on the consequences of eliminating conservative accounting information.

The remainder of the paper proceeds as follows. In section 2, I discuss theory and prior research on accounting conservatism, investment efficiency and the relationship between the two constructs. Section 3 describes hypothesis development; section 4 presents research design and section 5 explains empirical findings, and section 6 concludes the paper.

2. Literature review

2.1 Accounting conservatism

The thesis is based on the accounting conservatism literature. According to Basu (1997), more verification is required to recognized gains than losses, thus causing a timing asymmetry in a manner that bad news are recognized sooner in earnings than good news. Moreover, it is discovered that decreased earnings related to bad news are less persistent than increased earnings related to good news. Based on the work of Beaver and Ryan (2005), conservatism can be developed into two types—conditional conservatism and unconditional conservatism. While unconditional conservatism tend to happen at the inception of an asset, conditional conservatism appears to happen depending on news of different type. This paper focus on conditional conservatism since it allow the inclusion of more unforeseeable factors, only on which the relation between the two constructs in the research question is built to be meaningful. Otherwise, the research loses its meaning since unconditional conservatism is more of a choice of accounting method and is fixed irrespective of certain influencing factors. The influence of conservatism on accounting practice is a major area of findings among extant literature. Of those literature, the main theory tested in this paper is the work of Watts (2003) proposing explanations for conservatism. Watts articulated four explanations for conservatism: contracting, shareholder litigation, taxation and accounting regulation, the former two playing a major role. Besides, the paper offered a new argument that “even

without contracting considerations, an information perspective produces conservatism once the information costs of changed managerial behavior are introduced”. First, under contracting explanation, three attributes of accounting measure—timeliness, verifiability and asymmetric verifiability are taken into account, since contracting parties from earnings-based management compensation contracts, accounting–based debt contracts demand for timely earnings and net asset measures; these earnings or cash flow measures have to be verifiable for the contract to be enforceable; the asymmetric payoffs from the contracts require asymmetric verifiability. According to Watts (2003), contracting explanation is summarized in three aspects:

(1) In debt contracts, conservatism restricts managers from forgoing positive net present value projects, overstating earnings and assets and making liquidating dividend payment to shareholders at the expense of shareholders;

(2) In compensations contracts, conservatism restricts managers from overstating net assets and cumulative earnings for incentive purposes instead of enhancing firm value by taking positive net present value projects;

(3) In corporate governance, conservatism timely indicates the existence of negative net present value projects and facilitates investigations and adoptions of proper counteractions, in a way protecting the shareholders’ property rights.

In what’s following, the contracting explanation is to become the main testing object of this paper. Second, the information perspective relates to the contracting explanation, since the conservative earnings numbers provide investors with control over other sources of information when unverifiable future cash flows supplied by analysts can be evaluated against subsequent conservative earnings numbers and are therefore proved of higher quality. In the contrary, the quality of other information decreases if accounting earnings are unverifiable. Third, the litigation explanation applies recently in the U.S. since the 1966 changes in the rules for bringing class action suits. As a consequence, the expected litigation costs of overstatement are higher than those of understatement, for which reason management and auditors have incentives to report conservatively. Fourth, the income tax explanation states that the connection between taxable and reported income provides an incentive to defer income to reduce the present value of taxes. Last, regulatory explanation arises when losses from overvalued assets and overstated income are more observable and usable on the political

process than forgone gains due to undervalued assets or understated income. This phenomenon provides incentives for regulators and standard setters to be conservative.

Other areas of conservatism investigated in prior literature include the construct validity of the various empirical measures of conservatism (Wang, Ho´ gartaigh, & van Zijl, 2009), the legal and political determinants of conservatism (Habib, 2007; Watts, 2003a, 2003b), and the relationship between timely loss recognition (i.e., conditional conservatism) and earnings quality (Dechow, Ge, & Schrand, 2010). Despite those researches focusing on certain specific measures or reasons of conservatism, the cost and benefits of conservative accounting and the relative effects of the two types of conservatism on financial reports and users of the statements are explored in a broader sense in the work of Ruch and Taylor (2015).

2.2 Investment efficiency

Investment efficiency, the dependent variable in question, is one of the fundamental questions of finance literature. According to Modigliani and Miller (1958) paradigm, optimal investment decision is reached when all positive net present value projects are pursued, presuming that firms are likely to acquire financing for these projects and will continue investing until marginal benefits equal marginal costs. The important assumptions under this theory are frictionless market and the only driver of investment being investment opportunities. However, in reality capital market frictions does exist and firms make sub-optimal investment decisions, resulting in either over- or under-investment. These frictions can happen because of conflicts of interests or financial constraints. To be precise, prior literature and empirical work suggest two decisive frictions causing investment inefficiency, namely information asymmetry and agency problems.

Information asymmetry is aligned with under-investment. When managers have private information that is unavailable to outside capital providers about the expected profitability of the investment, the external investors have difficulty evaluating the investment performance of the managers and thus making investing decisions. If it is hard for investors to assess the profitability of the project, they tend to limit the degree to which managers can pursue the positive net present value projects in the form of financial constraints, leading to under-investment. Jensen and Meckling (1976), Myers and Majluf (1984) and Gomariz and Bellesta (2014) develop a framework for analyzing the role of the asymmetric information in investment efficiency through information problems, such as moral hazard and adverse selection.

The principal-agent conflict or moral hazard problem attributes distortions in investment decisions to the misalignment between managerial incentives and shareholder interests. Instead of maximizing shareholders' wealth, managers have other private objectives that might lead them to inefficiently invest the firm's capital. For example, managers make over- or under-investment decisions for empire building, perquisite consumptions, and management entrenchment, as suggested by Jensen (1986) and Aggarwal and Samwick (2006). Blanchard et al. (1994) argue that managers tend to use the firm's cash windfalls to invest in unattractive projects for the purposes of self-gratification or to maintain the independence of their firms with themselves at the helm. Some behavioral traits of managers can also induce over- or under-investments, including managers' career concerns, the incongruity in risk preferences between managers and shareholders (Holmstrom, 1999), managers' preference for a “quiet life” (Bertrand and Mullainathan, 2003), and managers' hubris in overestimating their own abilities to identify good investment opportunities. Another consequence of severe moral hazard problems is a higher cost of capital for firms that will cause under-investment.

The adverse selection problem suggests that firms' insiders possess superior information as to the firm's true value and have the incentives to issue overpriced capital or issue capital when the firm is overpriced. If they are successful in raising funds, excess capital enables over-investment. However, outside capital providers, with informational disadvantages, may detect or suspect such behaviors by firm insiders and respond by increasing the firm's cost of external financing. This can be a pervasive effect that affects all firms including those that do not overprice new issues of shares. As a result, managers who are inclined to protect the existing shareholders might forego profitable investment opportunities, rather than selling under-valued securities, or rely more on internally generated capital for investments. The adverse selection problem affects not only equity financing (e.g., Easley and O'Hara, 2004), but also debt financing, for which lenders, who possess less information than borrowers, will require a higher cost of debt that squeezes out “good borrowers” (Fazzari et al., 1988). Lambert et al. (2012) and Bhattacharya et al. (2011) suggest that providing more information to more investors can reduce a firm's cost of capital.

In contrast to the information asymmetry view, which suggest that managers act in the interests of shareholders, the agency view, occurring as a result of the separation of ownership and control, argues that managers are self-interested and they tend to maximize their welfare by choosing sub-optimal projects at the expense of shareholders in order to

expropriate some of the firm resources at their disposal. For instance, Jensen (1986) predicts that empire building induces managers with free cash flow to overinvest; this is especially true when managers are not monitored by shareholders. Blanchard et al. (1994) and Lang et al. (1991), empirically investigate the agency view and confirm that it is a principal source of investment inefficiency. As a consequence, over-investment is resulted from poor project selection. On the other hand, the cost of raising funds may also increase since investors are probably able to anticipate potential resource expropriation.

From the argument above, probably it is not hard to infer reversely that investment efficiency is related with reduced external financing costs, followed by enhanced investment opportunities, external financing referring to both debt financing and equity financing. Beatriz et. al (2010) argues that reduced cost of debt financing contributes to less risk-taking debts, resulting in reduced debt overhang negative effects on investment efficiency. (Myers, 1977; 1984). To clarify, Myers’ debt overhang theory (1977) refers to a causal relationship between increased leverage and future under-investments. A supporting evidence of this theory can be found in the work of Cai, Zhang et al (2010), who found a negative relation between leverage ratio and future investment. On the other hand, investment efficiency also includes reduced equity financing cost. (Beatriz et al, 2010; Guay and Verrecchia, 2007; Suijs, 2008).

2.3 Conservatism—functionality defined

FASB excluded conservatism as an attribute to the quality of financial reporting because it is believed to dampen the credibility of accounting information (FASB, 2010). Proponents of this view contend that this invalid information could hinder management from making the right choices. (Gigler, Kanodia, Sapra, & Venugopalan, 2009; Guay & Verrecchia, 2006). The opponents of this view, however, suggest that its development is necessary among misaligned parties and could serve as a contracting mechanism (Basu, 1997; Watts, 2003a). This argument is based on asymmetric payoffs of contracting parties and recognizing certain news sooner in earnings could be of crucial interest to a certain party. (e.g. a timelier reflection of loss could be deemed as a source of credible information by debt holders). Ruch and Taylor (2015) interpret the above debate on the cost and benefits of accounting conservatism as the existence of different perspectives on the function of conservatism. They argue that there are two main functions of conservatism—valuation and contracting. The former focus on the adequacy of conservative information to reflect market value of a firm

and serve decision-making process, while the latter maintains conservatism as a way of assessing the fulfillment of obligations of management and resolve agency conflicts among contracting parties.

First of all, accounting conservatism is one method of mitigating agency problems via contracting mechanism through debt, compensation, or other types of contracts (Watts, 2003). Creditors, owners of debt covenants, seek stability in corporate operations and income, and view conservative accounting practices as helping them manage their debt rights. The consensus among scholars is that accounting conservatism mitigates the agency problems associated with the conflicting interests of creditors and both shareholders and managers (Ahmed et al., 2002; Ball, Robin, & Sadka, 2008; Bushman & Piotroski, 2006; Givoly & Hayn, 2000; LaFond & Watts, 2008; Watts, 2003; Cho-Min Lin et al., 2018). García, Manuel, Osma, and Fernando (2009) assert that as a governance mechanism, accounting conservatism mitigates agency problems attributable to the asymmetric temporal recognition of gains and losses. Although some assert that conservative accounting procedures represent a substitute for corporate governance, others think that the managers of well-governed firms naturally prefer conservative accounting principles (Chi, Liu, & Wang, 2009).

Second, another function of conservatism lies in its impact on valuations, or its information predictability on firm value. The study conducted by Sohn (2011) is the first empirical research that examines directly the incorporation of accounting conservatism into analysts forecast, results showing that analysts incorporates accounting conservatism into their earnings forecasts and that forecasting earnings is more difficult for less conservative firms, the two findings leading to the third that the return predictability of the V/P ratio is stronger for more conservative firms. One implication discovered from this study is that accounting conservatism as a firm(manager) attribute plays an incremental role in determining forecast accuracy. The empirical findings of this research are considered of value since they indicate that conservatism-incorporated analysts forecast can influence managers’ choice of target to meet or beat analysts forecast and that accounting conservatism restricts managers’ discretion to manipulate earnings to an unpredictable level, which comforts analysts’ job and enhances forecasts accuracy. The return predictability of V/P ratio lies in the fact that V represents a firm’s fundamental value more accurately than the current mispriced P. Thus a more accurately estimated V contributes to the efficiency of V/P ratio in its return predictability, given that the main component of V is analysts forecast (F), and F contains less noise with the incorporation of conservatism which narrows down the unpredictable range of earnings.

Some other areas of research, by comparison, deal with the impact of conservatism on valuations directly instead of analysts forecast. Using a LIM-based valuation model that accommodates conservative accounting effects positively corrects the negatively biased intrinsic firm value derived from conventional residual income valuation model and LIM (linear information model) (Choi et al., 2006). Within a valuation framework, unconditional and conditional conservatism have indistinguishable impact due to their origin of a desire to be cautious in the recognition of increases in economic wealth, and homogeneity is required in the degree of accounting conservatism and the persistence of abnormal earnings when for the coefficients used in the cross-section regressions( David Ashton & Pengguo Wang, 2015). From the discussion above, the valuation and contracting functions of accounting conservatism facilitate the resolution of information asymmetry and agency conflicts, which are also two frictions causing investment inefficiency. Hence, it could be a fact that adopting accounting conservatism improves investment efficiency, which deserves investigation and leads to the research question of this paper.

2.4 Prior literature on the relation between conservatism and

investment efficiency

Admittedly, the research on the relation between conservatism and investment efficiency is nothing new but an area receiving constant attention. And it seems to be a consent among scholars that there is a positive relation between conservatism and investment efficiency. (Rajan et. al 2007, García et. al 2010, Ahmed et.al 2011, Haw et. al 2014, Juan et. al 2016). For firms more prone to underinvest, those adopting conservatism tend to invest more in prudent projects and issue more debt, and these new investments are also financed with new debt, providing new evidence that confirm prior literature (Ahmed et. al, 2002; Zhang 2008; Wittenberg-Moerman, 2008) on conservatism helping lower cost of debt. (García Lara et. al, 2016). Besides, evidence shows that conservatism mitigates over-investment problem (Francis and Martin, 2010). Relatedly, prior accounting studies document a significant association between accounting conservatism and level of debt (Ball and Shivakumar 2005; Khan and Watts 2009), issuance of new debt (Nikolaev 2010; Basu et. al 2011), cost of debt (Ahmed et. al. 2002; Zhang 2008), and nonpricing terms of debt contracts (Ball et. al 2008a; Nikolaev 2010; Li 2010). In agency view, empirical work document the effect of accounting conservatism on governance, or monitoring over managers’ investment decisions. Conservatism plays a vital role in providing managers with ex ante incentives to avoid taking

negative net present value projects and in ex post monitoring managers’ investment decisions ( Ahmed et. al, 2011; Watts, 2003; Ball and Shivakumar, 2005 ).

Nonetheless, to my knowledge, literature listed above were almost exclusively conducted outside Europe, and the manner in which sample was collected was based on the categorization of listed or private firms. Therefore, the main task of this paper is to investigate whether the positive influence of conservatism on investment efficiency still stands true by providing empirical results based specifically on a recent sample of high-tech firms in Europe, a novel setting compared with prior literature. Moreover, with the intended purpose mentioned above, this thesis attempts to include a different combination of constructs to evaluate the concept of investment efficiency from prior literature.

3. Hypothesis development

Information frictions exist among vested interest parties, specifically bondholders, shareholders and management. This information asymmetry signals misaligned interest (agency problem) which could lead to rents extracting behaviors of contracting parties (e.g. managers making suboptimal decisions in their own interest against that of the company, shareholders prefer high-risk projects enjoying upward potential due to limited liability at the expense at debt holders, who would shoulder the risk of not getting the principal back) (Jenson and Meckling, 1976) . Results from prior studies indicate that the 'informational role' of conservative reporting may be a reaction to information asymmetry (La Fond and Watts, 2008; Khan and Watts, 2009) and to some extent mitigate information asymmetry, thereby aligning interests through restricting the self-interest behaviors of contracting parties (La Fond and Watts, 2008). Studies from Ahmed and Duellman (2007b) and Garcia Lara et. al (2009) suggest a link between information asymmetry and more stringent corporate governance, which limits managerial power and enhance monitoring quality. To put more specifically, early signal reflected from a timelier disclosure on bad news indicating declining debt value strengthens controllability of debt holders while the possibility of rents extracting behaviors of shareholders and managers is significantly minimized(Ball and Shivakumar 2006; Nikolaev 2010). Moreover, conservative accounting is demonstrated to show efficiency in resolving agency conflicts between managers and shareholders. Ball and Shivakumar (2005) contend that since losses are recognized sooner, managers realize the possibility of delaying the losses until the tenure of their successors is minimal, and it follows that more cautious considerations are taken into account to avoid choosing negative NPV projects

which would harm firm value (e.g. 'empire building strategies', 'pet' projects or 'trophy' acquisitions). In a similar sense, conservatism facilitates the discovering, abandoning and discontinuing negative NPV projects(e.g. overinvestment) (Francis and Martin, 2010; Bushman et. al, 2011).

The arguments above lead to the following hypotheses:

H1a: Conservatism is negatively associated with information asymmetry and negative NPV projects.

H1b: Conservatism is negatively associated with agency conflicts and overinvestment.

In prior literature, the other role of conservatism in financing has also been paid lots of attention. As mentioned above, the informational role of conservatism mitigates agency conflicts between debt holders and managers, and as a result, a more prudent accounting mechanism may build trust in debt investors that they are more informed about the underlying risk of debt value of the firm. This line of reasoning is consistent with prior studies on the effect of adopting conservative accounting on debt structure (Ball and Shivakumar 2005; Khan and Watts 2009), attracting debt financing (Nikolaev 2010; Basu et. al 2011), cost of debt capital (Ahmed 2002; Zhang 2008), non-pricing terms of debt contracts (Ball et.al 2008a; Nikolaev 2010; Li 2010), the positive effect of conservatism on mitigating financial constraints for firms going through financial difficulties (Juan et. al 2016), the effect of conservatism on public debt financing in privately-held firms and how does the agency conflicts between debt holders and shareholders affect this effect (Haw et. al 2014). Furthermore, it is argued that through facilitating the issuance of debt, conservatism brings in more investment opportunities which may be abandoned without the enhanced financing opportunities. Alternatively, there is limited evidence on the role of conservatism on equity financing. The work based on Yongtae et.al (2013) reports that 'issuers with a greater degree of conservatism experience fewer negative market reactions to SEO announcements', and it is further shown that 'an important mechanism through which conservatism affects SEO announcement returns is by mitigating the negative impact of information asymmetry'. Similarly, several other studies (Beatriz et. al, 2010; Guay and Verrecchia, 2007; Suijs, 2008) contends that accounting conservatism contributes to reducing cost of equity capital.

The arguments above lead to the following hypotheses: H2: Conservatism is positively associated with debt issuance.

4. Research design

4.1 Conservatism and investment efficiency

Biddle et. al (2009) contends that measures of investment-cash flow sensitivities can reflect either financing constraints or a cash surplus. The test on hypotheses(1) and (2) is thus based on the measurement proposed by Biddle et. al (2009), as adopted also in Beatriz et. al (2010), which permits analyzing the effects of accounting choices in reducing over- and under-investment, as well as the net effect. This thesis follows the model adopted in Beatriz et. al (2010) to capture the effects of conditional conservatism on investment efficiency as follows: Investmentt+1=αi +βi +δ1CONSt + δ2CONSt*OverInvt+1 + δ3Overinvt+1

+ γControlst+ γYear + εt (1)

where Investment is a measure of future investment in both capital and non-capital goods, CONS is a firm-year-specific measure of conservatism, increasing in conservatism, OverInv is a ranked variable capturing settings where over- or under-investment is more likely. Controls is a vector of control variables that affect the level of investment and conservatism. Firm size (Size) is measured as the log of market value of equity. Depreciation method (AccDep) is an indicator variable that equals one if the firm uses accelerated depreciation, and zero otherwise. Recent work by Jackson et. al (2009) suggests firms that use accelerated depreciation have larger capital investments. Volatility of cash flow from operations (StdCFO) is the firm-specific standard deviation of the cash flow from operations scaled by average total assets, measured in the five-year period ending in the previous fiscal year (t-5 to t-1). Volatility of sales (StdSales) is the firm-specific standard deviation of annual sales deflated by average total assets, for years t-5 to t-1. Tangibility is the ratio of property, plant and equipment to total assets. Capital structure (Leverage) is ratio of short-term plus long-term debt scaled by market value of equity. Equation (1) is estimated with a fixed effect model that includes year indicator variables to control for year-specific shocks to investment.

Similar to Biddle et. al (2009), the investment proxy, Investment, is a measure of total investment defined as capital expenditures plus research and development plus acquisition expenditures less cash receipts from sales of property plant and equipment, multiplied by 100 and scaled by average total assets. OverInv takes values between 0 and 1, where 0 (or values close to 0) indicates under-investment and 1 (or values close to 1) indicates over-investment. In the above regression (model 1) the coefficients of interest are δ 1 andδ 2. The main

hypothesis is that conditional conservatism improves investment efficiency; that is, conservatism reduces both under- and over-investment. Therefore, when underinvestment is present (i.e., OverInv = 0) I expect coefficient β 1 to be positive. A positive 1 indicates that

conditional conservatism increases capital investment in settings where under-investment is most likely. On the contrary, when over-investment is present (i.e., OverInv = 1) I expect coefficient β 2 to be negative and greater in absolute value than 1 (i.e., β 1 + β 2 < 0),

indicating that conservatism decreases investment in settings where over-investment is most likely.

Clearly, the key element in model (1) is the definition of OverInv: the proxy to detect settings in which there is under- or over-investment. Following Biddle et. al (2009) and Beatriz et. al (2010), I define OverInv at the economy-wide level identifying years in which over-investment is most likely in the whole economy. It is defined as the decile ranks of the residuals from a time-series regression of annual average future capital expenditures on annual average current sales growth. This regression is estimated in time-series fashion as follows:

Investmentt+1 = b0 + b1 SalesGrowtht + μt+1 t = 1997, … 2017 (2)

where Investment is the average future investment for each sample year, and SalesGrowth is a proxy of firm investment opportunities calculated as the average change in sales from year t-1 to t for each sample year. To obtain OverInv at the aggregate economy-wide level, I rank the residuals of regression (2) into deciles and rescale the ranks from 0 to 1 to facilitate the interpretation of the coefficients of regression (1). Finally, I assign this annual measure to each firm based on its year. Thus, sample years with large positive (negative) residuals will be considered as years of average over-investment (under-investment) at the economy-wide level, and they will have values of OverInv close to 1 (0).

To test hypothesis H2, I use the following model adopted in Juan et. al (2016): Δ Debt issuancet+1= βt + δ1 CONSt + δ2 CONSt *OverInvt + δ3 OverInvt

+γ Controlst + γYear + εt (2)

The dependent variable, ΔDebt issuance t+1, is defined as the future change in new debt

issuance scaled by current sales. Debt issuance equals long-term debt issuance, minus the reduction in long-term debt, plus changes in current debt. If more conservative firms in

settings prone to underinvestment issue more debt than less conservative firms, I expect the sum ofδ 1 and δ2 to be positive and significant.

To estimate models (1) and (2) I need a firm-specific measure of conservatism (CONS). To construct this proxy, I follow the work of Khan and Watts (2009) (Model 4), who estimate a measure of firm-year measure of conservatism drawing from the Basu (1997) model. Prior literature demonstrates that the Basu (1997) model (Model 3) is able to capture cross-sectional variation in conditional conservatism. The models are as follow:

Earni = β0 + β1Negi + β2Reti + β3RetiNegi + μi (3)

where Earn is net income before extraordinary items deflated by market value of equity at the beginning of the period, Ret is the annual stock rate of return of the firm, measured compounding twelve monthly CRSP stock returns ending at fiscal year end, Neg is a dummy variable that equals 1 in the case of bad news (negative or zero stock rate of return) and 0 in the case of good news (positive stock rate of return) and i indexes the firm. In model (3), the β 2 coefficient is the good news timeliness measure, β 3 captures the incremental timeliness

of earnings to bad news, and the total bad news timeliness is β 2 +β 3. Under conservative

accounting, β 3 is predicted to be positive and significant. Largerβ 3 coefficients indicate

more pronounced conditional conservatism.

Xi=β 1+β 1Di+Ri (μ1+μ2Sizei+μ3M/Bi+μ4Levi)+DiRi(λ 1+λ 2Sizei+λ 3M/Bi+λ 4Levi)+δ 1

Sizei+δ 2 M/Bi+δ 3 Levi+δ 4Di Sizei+δ 5Di M/Bi+δ 6 Di Levi）+ε i (4)

where i indexes the firm, X is earnings, R is returns (measuring news), D is a dummy variable equal to 1 when R<0 and equal to 0 otherwise. Earnings is net income before extraordinary items, scaled by lagged market value of equity. Size is the natural log of market value of equity. M/B is the market-to-book ratio. Lev is leverage, defined as long-term and short-term debt deflated by market value of equity.

Detailed definitions of variables can be find in table 1 as follows.

Table 1 Definitions of variables

Investment Investment is a measure of total investment defined as capital expenditures plus research and development plus acquisition expenditures less cash receipts from sales of property plant and equipment, multiplied by 100 and scaled by average total assets

Debt issuance ΔDebt issuance is defined as the future change in new debt issuance scaled by

current sales. Debt issuance equals long-term debt issuance, minus the reduction in long-term debt, plus changes in current debt

CONS CONS is the Kahn and Watts (2009) firm-year measure of accounting conservatism. Higher values of CONS are associated to higher conservatism

OverInv OverInv is a ranked variable based on the annual unexplained aggregate investment for all firms in the economy. It takes values from 0 to 1; values closer to 0 (1) indicate settings in which under-investment (over-investment) is most likely

Size Size is measured as the log of market value of equity.

AccDep Depreciation method (AccDep) is an indicator variable that equals one if the firm uses accelerated depreciation, and zero otherwise. Recent work by Jackson et al. (2009) suggests firms that use accelerated depreciation have larger capital investments

StdCFO Volatility of cash flow from operations (StdCFO) is the firm-specific standard deviation of the cash flow from operations scaled by average total assets, measured in the five-year period ending in the previous fiscal year (t-5 to t-1)

StdSALE Volatility of sales (StdSales) is the firm-specific standard deviation of annual sales deflated by average total assets, for years t-5 to t-1

Tangibility Tangibility is the ratio of property, plant and equipment to total assets

Leverage Leverage is ratio of short-term plus long-term debt scaled by market value of equity

SalesGrowth SalesGrowth is a proxy of firm investment opportunities calculated as the average change in sales from year t-1 to t for each sample year

Earn Earn is net income before extraordinary items deflated by market value of equity at the beginning of the period

Ret Ret is the annual stock rate of return of the firm, measured compounding twelve monthly CRSP stock returns ending at fiscal year end

Neg Neg is a dummy variable that equals 1 in the case of bad news (negative or zero stock rate of return) and 0 in the case of good news (positive stock rate of return) and i indexes the firm

M/B M/B is the market-to-book ratio

4.2 Samples and data sources

Data are available through databases within the Wharton Research Data Services (WRDS) system to which the university subscribes. Firm-specific measure of conservatism data for European high-tech companies are available through Compustat Global. Data on other operationalized dependent variables — Investment, ΔDebt issuance t+1, and data on the

computation of the related variables and certain control variables are available through databases such as Compustat Global, CRSP.

The sample period will start from 1997 because my intention is to include as much as possible the data so as to enhance generality and to examine on a broader range of time the research question in the setting of European high-tech companies — a setting absent attention from prior literature from my knowledge of the relevant literature. Besides, I specifically focus on high-tech industry in Europe since one of the main dependent variable, Investment, is measured as capital expenditures plus research and development plus acquisition expenditures less cash receipts from sales of property plant and equipment, multiplied by 100 and scaled by average total assets, a number which can be reasonably assumed occupying a large portion of expenditures of high-tech companies. After taking out observations with

missing values, it leads to 491 firm-year observations. Detailed descriptive statistics can be found in table 2 and table 3. The sample will run through the most recent year with data available in all databases.

Table 2

Descriptive statistics on European high-tech firms Current ISO Country

Code - Incorporation

Freq. Percent Cum.

AUT 2 0.41 0.41 BEL 7 1.43 1.83 BGR 2 0.41 2,24 CHE 4 0.81 3.05 CHL 2 0.41 3.46 CYM 2 0.41 3.87 CYP 2 0.41 4.28 DEU 58 11.81 16.09 DNK 11 2.24 18.33 ESP 5 1.02 19.35 FIN 12 2.44 21.79 FRA 53 10.79 32.59 GBR 105 21.38 53.97 GIB 1 0.20 54.18 GRC 14 2.85 57.03 HUN 2 0.41 57.43 IRL 1 0.20 57.64 ISL 1 0.20 57.84 ISR 22 4.48 62.32 ITA 16 3.26 65.58 JEY 3 0.61 66.19 LTU 1 0.20 66.40 LUX 1 0.20 66.60 LVA 1 0.20 66.80 NGA 5 1.02 67.82 NLD 19 3.87 71.69 NOR 10 2.04 73.73 POL 71 14.46 88.19 PRT 2 0.41 88.59 ROU 2 0.41 89.00 RUS 4 0.81 89.82 SVK 1 0.20 90.02 SWE 49 9.98 100.00 Total 491 100.00 Table 3 Descriptive statistics

stats mean std P50 max min

Investment 12.4091 14.8361 9.2287 118.1110 -11.6686 CONS 0.1502 0.0853 0.1533 0.2999 0.0001 Size 4.4360 2.0477 4.3056 14.1149 -4.0174 AccDep 0.6519 0.4765 1.0000 1.0000 0.0000 StdCFO 0.1217 0.0936 0.0995 0.2320 0.0201 StdSALE 0.3256 0.7532 0.1867 20.9112 0.0000 Tangibility 0.0998 0.1315 0.0469 0.9733 0.0000 Leverage 0.2136 4.0118 0.0654 129.3750 -126.6667 OverInv 0.9309 0.2537 1.0000 1.0000 0.0000 Debt issuance -556.6612 708.1690 -556.6604 26434.6300 -4282.2000

I utilize a sample of Compustat and CRSP firms over 1997–2017. CONS is the Kahn and Watts (2009) firm-year measure of accounting conservatism. Higher values of CONS are associated to higher conservatism. OverInv is a ranked variable based on the annual unexplained aggregate investment for all firms in the economy. It takes values from 0 to 1; values closer to 0 (1) indicate settings in which under-investment (over-investment) is most likely.Investment is a measure of total investment defined as capital expenditures plus research and development plus acquisition expenditures less cash receipts from sales of property plant and equipment, multiplied by 100 and scaled by average total assets. Δ Debt issuance is defined as the future change in new debt issuance scaled by current sales. Debt issuance equals long-term debt issuance, minus the reduction in long-term debt, plus changes in current debt. Size is measured as the log of market value of equity. Depreciation method (AccDep) is an indicator variable that equals one if the firm uses accelerated depreciation, and zero otherwise. Recent work by Jackson et al. (2009) suggests firms that use accelerated depreciation have larger capital investments. Volatility of cash flow from operations (StdCFO) is the firm-specific standard deviation of the cash flow from operations scaled by average total assets, measured in the five-year period ending in the previous fiscal year (t-5 to t-1).Volatility of sales (StdSales) is the firm-specific standard deviation of annual sales deflated by average total assets, for years t-5 to t-1. Tangibility is the ratio of property, plant and equipment to total assets. Leverage is ratio of short-term plus long-term debt scaled by market value of equity.

Table 4

Pearson Correlations between Investment, CONS, OverInv, Debt issuance and control variables Investme

nt

CONS Size AccDep StdCFO StdSALE Tangibili

ty

Leverage OverInv Debt issuance Investm ent 1 CONS -0.054 1 Size -0.218*** 0.009 1 AccDep 0.164** -0.014 0.056** * 1 StdCFO -0.142 0.133 0.255** * -0.039** 1 StdSAL E 0.045 -0.022 -0.206** * -0.042* -0.028 1 Tangibil ity 0.111* 0.0300 0.106** * 0.080*** 0.078*** -0.063*** 1 Leverag e -0.190*** -0.009 0.008 0.043** -0.001 -0.001 -0.086*** 1 OverInv 0.563*** -0.006 -0.120** * 0.046** -0.085*** 0.045** 0.003 -0.001 1 Debt issuance -0.039 -0.035* 0.143** * -0.022 0.419*** -0.011 0.003 -0.001 0.016 1

I utilize a sample of Compustat and CRSP firms over 1997–2017. CONS is the Kahn and Watts (2009) firm-year measure of accounting conservatism. Higher values of CONS are associated to higher conservatism. OverInv is a ranked variable based on the annual unexplained aggregate investment for all firms in the economy. It takes values from 0 to 1; values closer to 0 (1) indicate settings in which under-investment (over-investment) is most likely. Investment is a measure of total investment defined as capital expenditures plus research and development plus acquisition expenditures less cash receipts from sales of property plant and equipment, multiplied by 100 and scaled by average total assets. Δ Debt issuance is defined as the future change in new debt issuance scaled by current sales. Debt issuance equals long-term debt issuance, minus the reduction in long-term debt, plus changes in current debt. Size is measured as the log of market value of equity. Depreciation method (AccDep) is an indicator variable that equals one if the firm uses accelerated depreciation, and zero otherwise. Recent work by Jackson et al. (2009) suggests firms that use accelerated depreciation have larger capital investments. Volatility of cash flow from operations (StdCFO) is the firm-specific standard deviation of the cash flow from operations scaled by average total assets, measured in the five-year period ending in the previous fiscal year (t-5 to t-1).Volatility of sales (StdSales) is the firm-specific standard deviation of annual sales deflated by average total assets, for years t-5 to t-1. Tangibility is the

ratio of property, plant and equipment to total assets. Leverage is ratio of short-term plus long-term debt scaled by market value of equity.

As for the selection of industry classification code in order to sample high-tech firms, I base all selection on four-digit NAICS (North American Industry Classification System) taken directly from Charles and Mary (2009) as a guidance. The NAICS project is a North American joint venture designed to facilitate industry comparisons and reflect changes in the economy. The production-based framework of NAICS, compared with the product-based SIC, is designed to bring together industries that have similar activities and processes as well as similar products. Although Standardized Industry Classification (SIC) codes was dominantly adopted in previous research, recent studies have documented limitations and potential deficiencies of SIC codes as a methodology to identify and segment firms for empirical research purposes (Clarke [1989]; Amit and Livnat [1990]; Guenther and Rosman [1994]; Kahle and Walkling [1996]; Fan and Lang [2000]; Walker and Murphy [2001]; Bhojraj, Lee, and Oler [2003]). Moreover, SIC codes are product based, as opposed to process-based, and reflect a focus limited to the products and services delivered by a firm without consideration of markets or methods to market products and services (Fan and Lang [2000]). The more recently developed NAICS and GICS classifications offer altematives to SIC codes. With an emphasis on the service sector and emerging industries, the NAICS system in particular, shows promise for use in future research. Besides, The increased refinement of the four-digit codes provides sufficient benefit over three-digit NAICS sampling to warrant recommending using four-digit codes when using NAICS codes to sample high-tech firms. (Charles and Mary [ 2009]). Since GICS is currently absent in WRDS, I chose NAICS to sample high-tech firms. Detailed descriptive statistics on the four-digit NAICS can be found in table 5 as follow.

Table 5

Optimal Four-Digit NAICS Code Combination for Sampling High-Technology Firms

NAICS Code Industry Name

3254 Pharmaceutical and Medicine Manufacturing

3341 Computer and Peripheral Equipment Manufacturing

3342 Communications Equipment Manufacturing

3344 Semiconductor and Other Elec. Component Mfg.

3345 Navigational, Measuring, and Control Instrument Mfg.

3346 Manufacturing and Reproducing Magnetic and Optical Media

3353 Electrical Equipment Manufacturing

3391 Medical Equipment and Supplies Manufacturing

5112 Software Publishers

5133 Telecommunications (1997)

5172 Wireless Telecommunications Carriers (except Satellite)

5 173 Telecommunications Resellers

5179 Other Telecommunications

5181 Internet Service Providers and Web Search Portals

5182 Data Processing, Hosting, and Related Services

5412 Accounting, Tax, Bookkeeping, and Payroll Services

5415 Computer Systems Design and Related Services

5416 Management, Scientific, and Technical Consulting Services

5417 Scientific Research and Development Services

5418 Advertising and Related Services

5613 Employment Services

5614 Business Support Services

5 Results

5.1 Conservatism and investment efficiency

As a first analysis, I study the association between conditional conservatism and investment efficiency in situations where firms deviate from the optimal level of investment. Table 1 presents results based on the time-series aggregate measure. However, contrary to prediction, conservatism is negatively associated with investment in circumstances prone to underinvestment (i.e, OverInv=0 or close to 0). The coefficient on CONS is negative and significant (CONS=-0.0358), indicating conservative accounting reduces investment among firms that are under-investing. In terms of economic significance, a one-decile change in conservatism results in a decrease in investment of 3.58%, relative to average total investment, among firms that are under-investing. As can be seen from this result, during the period from 1997 to 2017, high-tech firms in Europe which were already prone to under-invest were highly possible to continue reducing under-investment after adopting conservatism accounting. Unfortunately, this finding is inconsistent with the dominant finding in other literature, which shows a positive relation between conservatism and investment under circumstances where firms are prone to under-invest. The deviation from prediction might be due to relative small set of control variables. For example, accrual quality (FRQ) and governance quality (GOV) are two factors that could influence investment according to Beatriz et. al (2010) and Juan et.al (2016). Besides, there are many other optional proxies for measuring conservatism such as a firm-specific measurement of conservatism, which is also widely adopted in other literature. Moreover, since the samples of investigation are based in Europe, another explanation for this controversy could be due to policy reasons—difference between the Continental Law System and the Anglo-American Law System. The main coefficient of interest is the interaction between conditional conservatism and

over-investment in years when there are signs of over-over-investment in the economy (i.e, OverInv = 1 or close to 1). The coefficient on CONS*OverInv is negative and significant (CONS*OverInv=-0.0552). In terms of economic significance, a one-decile change in conservatism results in a decrease in investment of 5.52%, relative to average total investment, among firms that are over-investing. As can be seen from this result, during the period from 1997 to 2017, high-tech firms in Europe which were prone to over-invest, or managers prone to make aggressive investment decisions, were highly likely to cut back on investment after adopting conservatism accounting. This finding is consistent with the finding in other literature, which shows a negative relation between conservatism and investment under circumstances where firms are prone to over-invest. However, when looking at another angle, this contradiction with other literature could be subject to a different interpretation. Comparing the two findings listed above, and also considering the finding from other literature, it could probably be inferred that the main effect of conservatism on improving investment decisions can occur most often under circumstances where firms are prone to over-invest—conservative accounting is most effective in dealing with excessive investment decisions such as empire building or expropriation for contracting purposes, while the effect of conservative accounting on increasing investment under circumstances where firms are prone to under-invest depends on the specific industry of the firms. However, I suggest caution when interpreting this finding since, again, the inherent limitations of this paper. Next, I also investigate the results on other variables. First of all, the coefficients on Size and leverage have significant and negative influence on investment (Size=-2.604, Leverage=-8.15, respectively), and the coefficient on AccDep is significant and positive (AccDep=4.797). Size is measured as the log of market value of equity and Leverage is calculated as the ratio of short-term plus long-term debt scaled by market value of equity. In terms of economic significance, a one-decile change in Size (Leverage) results in a decrease in investment of 260.4% and 369.9%, respectively, relative to average total investment, among firms that are over-investing. Apparently, it can be inferred from this result that during the period from 1997 to 2017, high-tech companies in Europe which are larger in size and which raise more debt are more cautious in choosing investment projects—a finding which is understandable since large firms with large amounts of debt are subject to more rules and regulations such as attention from regulators and constraints in debt contract. As for AccDep, a one-decile change in AccDep results in an increase in investment of 479.7%, relative to average total investment, among firms that are under-investing. Depreciation method (AccDep) is an

indicator variable that equals one if the firm uses accelerated depreciation, and zero otherwise. This finding is also consistent with other literature. For example, recent work by Jackson et al. (2009) suggests firms that use accelerated depreciation have larger capital investments. Last, for the rest of control variables—StdCFO, StdSALE, Tangibility, the coefficient on StdCFO has significant and positive influence on investment (StdCFO=0.04), and the coefficient on StdSALE (Tangibility) is positive (negative) but not significant(StdSALE =19.5, Tangibility=-598.3, respectively). In terms of economic significance, a one-decile change in StdCFO/ StdSALE/ Tangibility results in an increase (decrease) in investment of 0.04%, 19.5% and 598.3%, respectively, relative to average total investment.

Table 6

Conditional relation between future investment and accounting conservatism

Investmentt+1=αi +βi +δ1CONSt + δ2CONSt*OverInvt+1 + δ3Overinvt+1 + γControls +γYear+ μt+1

Investmentt+1(1) Investmentt+1(2) CONS -0.0358*** -0.0041 (0.0135) (0.0122) CONS*OverInv -0.0552* (0.0357) OverInv 19.7013*** (2.9560) Size -1.6591** -2.6041*** (0.7653) (0.7433) AccDep 4.7970** 0.8153 (1.9542) (1.2243) StdCFO 0.0003 0.0004** (0.0002) (0.0002) StdSALE 4.4123 0.1952 (9.4528) (6.4004) Tangibility 13.2733 -5.9832 (9.3277) (10.485) Leverage -8.1503** -3.6998 (3.6695) (2.8053) Constant 17.6604*** 20.7402*** (5.5970) (4.8862)

Year-fixed effects Yes Yes

Observations 205 205

R-squared 0.122 0.404

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

(1) This row represents the results on the regression of CONS on Investment when underinvestment is more likely (OverInv=0 or close to 0)

(2) This row represents the results on the regression of CONS on Investment under circumstances more prone to overinvestment (OverInv=1 or close to 1)

The sample covers the period 1990-2007. Investment is a measure of total investment scaled by average total assets. CONS is the Kahn and Watts (2009) firm-year measure of accounting conservatism. Higher values of CONS are associated to

higher conservatism. OverInv is a ranked variable based on the annual unexplained aggregate investment for all firms in the economy. It takes values from 0 to 1; values closer to 0 (1) indicate settings in which under-investment (over-investment) is most likely. Investment is a measure of total investment defined as capital expenditures plus research and development plus acquisition expenditures less cash receipts from sales of property plant and equipment, multiplied by 100 and scaled by average total assets. Δ Debt issuance is defined as the future change in new debt issuance scaled by current sales. Debt issuance equals long-term debt issuance, minus the reduction in long-term debt, plus changes in current debt. Size is measured as the log of market value of equity. Depreciation method (AccDep) is an indicator variable that equals one if the firm uses accelerated depreciation, and zero otherwise. Recent work by Jackson et al. (2009) suggests firms that use accelerated depreciation have larger capital investments.Volatility of cash flow from operations (StdCFO) is the firm-specific standard deviation of the cash flow from operations scaled by average total assets, measured in the five-year period ending in the previous fiscal year (t-5 to t-1).Volatility of sales (StdSales) is the firm-specific standard deviation of annual sales deflated by average total assets, for years t-5 to t-1. Tangibility is the ratio of property, plant and equipment to total assets. Leverage is ratio of short-term plus long-term debt scaled by market value of equity. The symbols ***, **, * denote two-sided significance at the 1%, 5%, and 10% levels, respectively.

5.2 Conservatism and debt financing

As a second analysis, I study whether adopting conservatism contribute to debt financing. Table 2 presents the results given the same set of control variables. As predicted, the coefficient on CONS is significant and positive (CONS=0.0145). In terms of economic significance, a one-decile change in conservatism results in an increase in debt issuance of 1.45%, relative to average total debt issuance, among firms that are under-investing, indicating CONS contributing to debt financing in years when underinvestment is more likely. This finding is consistent with other literature. For example, according to Juan.et al (2016), it is discovered that there is a positive association between conservatism and future change in debt issuance in settings prone to underinvestment, and in terms of economic significance, a one-decile change in conservatism results in an increase in future debt issuance of 2.6%, relative to average total debt, among firms that are under-investing. This finding validates that adopting conservatism contributes to the conflicts on asymmetric pay-offs between debt holders and management since adopting conservative accounting means that debt holders are more and better informed about the underlying economic performance and whether the obligations of management are well performed. Again, as predicted, the coefficient on CONS*OverInv is significant and negative (CONS*OverInv=-1.614). In terms of economic significance, a one-decile change in conservatism results in a decrease in debt issuance of 161.4%, relative to average total debt issuance, among firms that are over-investing, This result is also consistent with other literature. The reason behind this finding could be interpreted as conservative accounting contributes to reducing excessive debt financing and self-interest managers engaging in empire building and therefore reduces agency conflicts.

Among control variables, Size and StdCFO both have signicant and positive influence on Δ Debt issuance (Size=31.05, StdCFO=0.277, respectively). In terms of economic significance, a one-decile change in Size (StdCFO) results in an increase in debt issuance of 31.05% and 27.7%, respectively, relative to average total debt issuance, among firms that are under-investing. Size is measured as the log of market value of equity and Volatility of cash flow from operations (StdCFO) is the firm-specific standard deviation of the cash flow from operations scaled by average total assets, measured in the five-year period ending in the previous fiscal year (t-5 to t-1). Clearly, firm size and cash flow are two variables that can potentially positively influence debt financing when the firm in question adopts conservative accounting. One possible explanation could be that for larger firms with volatile cash flows, adopting conservatism adds to their ability to signing debt contracts, which is understandable since this kind of firms are occupying more resources and capacity than smaller ones with relatively fixed cash flow patterns to attract debt holders. However, due to the inherent limitations on the sample size, I suggest interpreting this result with caution. Tangibility has significant and negative influence on Δ Debt issuance (Tangibility=-225.3). In terms of economic significance, a one-decile change in Tangibility results in a decrease in debt issuance of 225.3%, relative to average total debt issuance, among firms that are under-investing. The coefficient on AccDep is positive but not significant (AccDep=15.12). In terms of economic significance, a one-decile change in AccDep results in an increase in debt issuance of 15.12%, relative to average total debt issuance, among firms that are under-investing. The coefficient on StdSALE is positive but not significant (StdSALE =17.39). In terms of economic significance, a one-decile change in StdSALE results in an increase in debt issuance of 17.39%, relative to average total debt issuance, among firms that are under-investing. The coefficient on Leverage is negative but not significant (Leverage=-1.75). In terms of economic significance, a one-decile change in Leverage results in a decrease in debt issuance of 1.75%, relative to average total debt issuance, among firms that are under-investing.

Table 7

Conditional relation between debt issuance and accounting conservatism

ΔDebt issuancet+1= βt + δ1 CONSt + δ2 CONSt *OverInvt + δ3 OverInvt +γ Controlst + γYear+ εt+1

ΔDebt issuance (1) _{ΔDebt issuance} (2)

CONS 0.0145* 1.627*

(0.0179) (0.940)

CONS*OverInv -1.614*