1 DOES DISCLOSURE OF NON -FINANCIAL STATEMENT INFORMATION REDUCE FIRMS PROPENSITY TO UNDER INVEST? By HUNG YUAN LU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 HungYuan Lu
3 To my parents, my wife, and my son
4 ACKNOWLEDGMENTS I am grateful to my dissertation committee Bipin Ajinkya (Chair), Jay Ritter, Surjit Tinaikar, and Jenny Tucker for their support and guidance. I also want to thank Monika Causholli, Michael Donohoe, Verdi Rodrigo, Weining Zhang, and workshop participants at the University of Florida, Singapore M anagement University, Louisiana State University, California State University Fullerton, City University of New YorkQueens College, and Indiana University Northwest for helpful comments and suggestions.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 7 ABSTRACT .......................................................................................................................................... 8 CHAPTER 1 INTRODUCTION ....................................................................................................................... 10 2 BACKGROUND AND LITERATURE REVIEW ................................................................... 17 The Disclosure Literature ........................................................................................................... 17 The Financing Constraints Literature ........................................................................................ 18 Concurrent Research and Incrementa l Contribution ................................................................. 20 3 HYPOTHESES DEVELOPMENT AND EMPIRICAL CONSTRUCTS .............................. 23 Hypothesis Development ............................................................................................................ 23 Operationa lization of Theoretical Constructs ............................................................................ 24 Identifying Under Investment ............................................................................................. 24 Identifying Variations in NFS Disclosure .......................................................................... 26 Identifying Variations in Financing Activities .................................................................. 29 4 SAMPLE SELECTION AND EMPIRICAL SPECIFICATION ............................................. 33 Sample Selection ......................................................................................................................... 33 Empirical Specif ication ............................................................................................................... 35 Disclosure and Investment: Dependent Variable ............................................................... 35 Disclosure and Investment: Independent Variables .......................................................... 35 Disclosure and Investment: Empirical Specifications ....................................................... 37 Disclosure and Financing: Dependent Variables ............................................................... 38 Disclosure and Financing: Empirical Specifications ......................................................... 38 5 EMPIRICAL RESULTS ............................................................................................................. 41 NFS Disclosure and Investment ................................................................................................. 41 Hypotheses ........................................................................................................................... 41 Results and Discussion ........................................................................................................ 41 NFS Disclosure and Financing Choice ...................................................................................... 42 Hypotheses ........................................................................................................................... 42 Results and Discussion ........................................................................................................ 42
6 6 ROBUSTNESS TESTS .............................................................................................................. 47 Using the Ranks of Disclosure Scores ....................................................................................... 47 Controlling for Higher Orders of Tobins q and Lag Investment ............................................ 47 Using a One Year Span to Measure Inves tment ....................................................................... 47 Alternative Weightings of NFS Disclosure Items ..................................................................... 47 7 ENDOGENEITY OF NFS DISCLOSURE ............................................................................... 51 8 THE ROLE OF NFS DISCLOSURE IN NON -RECESSIONARY PERIODS ..................... 59 9 AN ALTERNATIVE MEASURE OF DISCLOSURE: MANAGEMENT EARNINGS GUIDANCE ................................................................................................................................ 64 10 CONCLUSION AND FUTURE RESEARCH ......................................................................... 68 APPENDIX A VARIABLE DEFINITIONS ...................................................................................................... 71 B DISCLOSURE SCORES: INDEX, CALCULATION, AND EXAMPLES ........................... 73 C DISCLOSURE SCORES: DESCRIPTIVE STATISTICS ....................................................... 75 D THE DISCLOSURE SCORES: EXAMPLES .......................................................................... 77 LIST OF REFERENCES ................................................................................................................... 89 BIOGRAPHICAL S KETCH ............................................................................................................. 94
7 LIST OF TABLES Table page 3 1 Descriptive statistics: Intertemporal pattern of aggregate investment ................................ 31 3 2 Descriptive statistics: Investment patterns of manufacturing firms, separated by industries ................................................................................................................................. 32 4 1 Sample Selection .................................................................................................................... 39 4 2 Descriptive Statistics .............................................................................................................. 39 4 3 Intertemporal pattern of industry -wide investment: Electronic Equipment Firms ............ 40 5 1 Disclosure level and investment ............................................................................................ 45 5 2 Disclosure level and financing choice: Odds Ratios ............................................................ 46 5 3 Disclosure level and financing choice: Marginal Effects .................................................... 46 6 1 Robustness tests: rank and M S specifications ..................................................................... 49 6 2 Ro bustness tests: one year specification and alternative weighting schemes .................... 50 7 1 Simultaneous equations ......................................................................................................... 56 7 2 Si multaneous Equations: Continued ..................................................................................... 57 7 3 Simultaneous equations with alternative combination of instruments ............................... 58 8 1 Descriptive statistics: NFS disclosure in a non recessionary period .................................. 62 9 1 Management earnings forecasts and investment: d escriptive statistics .............................. 67 9 2 Management earnings forecasts and under investment: 2SLS regression results .............. 67
8 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DOES DISCLOSURE OF NON -FINANCIAL STATEMENT INFORMATION REDUCE FIRMS PROPENSITY TO UNDER INVEST? By HUNG YUAN LU August 2009 Chair: Bipin Ajinkya Major: Business Administration In the presence of information asymmetry, Myers and Majluf (1984) and Greenwald, Stiglitz, and Weiss (1984) demonstrate that a firm may pass up positive net present value (NPV) projects due to adverse selection in the equity market. The same model can be applied analogously in the debt market (Myers and Majluf, 1984) This study presents an empirical examination of whether disclosure of non-financial statement (NFS) information mitigates the under investment problem, presumably by reducing information asym metry between managers and potential stakeholders. Because NFS disclosures are typically not comparable across industries, this study focuses on one industry: the electronic equipment industry. The sample includes a balanced panel of 222 firms for the per iod from 2001 to 2003. The proxy for NFS disclosures is hand-collected from the sample firms 2001 annual reports and the proxy for under -investment is the cross -sectional variation in firms investment level for years 2002 and 2003. I present three findings. First, I document that managers who provide more NFS information are less likely to under -invest. Second, by classifying NFS disclosures into t hose that are more relevant to equity holders and those more relevant to debt holders, I find that both types of disclosures are negatively associated with the degree of under investment. Last, I provide
9 evidence that both equityand debt related NFS disc losures are positively associated with the level of subsequent equity financing. These results hold after I use a two-stage least square s (2SLS) specification to control for the endogeneity between investment and disclosure and that between equity financi ng and disclosure.
10 CHAPTER 1 INTRODUCTION This study examines the role of voluntary disclosure in mitigating investment inefficiencies. Specifically, the study investigates whether disclosure of non-financial statement (NFS) information reduce s information asymmetry between a firm and its potential stakeholders thereby mitigating the under investment problem. [Myers and Majluf, 1984; Greenwald, Stiglitz and Weiss, 1984] I examine three questions. First, I test whether a firm that provides more NFS disclosures face s fewer financi ng constraint s and therefore is less like ly to under -invest than a firm that withho lds such information. Second, by classifying NFS disclosures into (i) disclosures that are relatively more relevant to potential equity ho lders and (ii) disclosures that are relatively more relevant to potential debt holders, I investigate whether the levels of these two types of disclosures are associated with the degree of under -investment. Third, I examine whether the levels of equity and debt related NFS disclosures are associated with a firms subsequent choice of equity or debt financing. Myers and Majluf (1984) and Greenwald, Stiglitz and Weiss (1984) demonstrate that in the presence of information asymmetry, a firm may pass up positive NPV projects due to costly external financing.1 In both models, a man a ger has perfect information about the value of the firms assets in place and growth options but the information about the assets in place is not available to potential equity holders The uneven distribution of information creates mispricing in the firms shares and can in some cases lead to existing shareholders forgoing positive NPV projects due to their reluctance to issue under -priced shares. In the debt version of the model, (Mye rs and Majluf, 1984) a manager has information about the firm s default risks but the 1 Several assumptions are required. First, managers must act in the best long term interest of pre issue shareholders and second, those pre issue shareholders do not actively rebalance t heir person portfolios.
11 information is not available to potential lenders. The information asymmetry between the manager and potential lenders could lead to under -investment due to the mispricin g of debt Since the under -investment problem is attributed to information asymmetry between managers and potential stakeholders, a natural extension of the literature is to empirically test whether public disclosure of equityand debt related information relaxes a firms financing constraints and mitigates the under investment problem. The specific disclosure examined in this study is the NFS information voluntarily disclosed by a manager. The choice is motivated by the fact that the theories that motivat e this study concern the information asymmetry between a manager and the firms potential stakeholders. Therefore, the existence and specificities of voluntary disclosure are the relevant attributes in examining this research question.2 Since financial st atement information is required and audited, there is not much cross -sectional variation in the existence or specificities of such information after the financial statements are published. On the other hand, NFS disclosure is highly discretionary, varies s ubstantially across firms, and is an important form of voluntary disclosure for managers to communicate info rmation to potential investors (Lang and Lundholm, 2000) The study of effects of such disclosure s should be of interest to companies and to policy makers. The two main constructs in this study are the degree of under -investment and the level of NFS disclosures. Following one of the approaches used by Biddle et al (2008) to identify under investment, I measure under investment using investment levels across sample firms at a time during which the aggregate investment is low relative to the adjacent years.3 The choice is 2 In other studies such as those that examine the relation between voluntary disclosure and the cost of capital, the timeliness attribute of voluntary disclosure may be of first order effect because the theory that motivates th ose studies concerns the information asymmetry between the informed and uninformed investors (Diamond and Verrecchia, 1991) 3 The relatively low aggregate investment occurs in 2002 and 2003, which form my sample years. See Chapter 3 Hypotheses Developments a nd Empirical Constructs for details.
12 motivated by Tirole (2006), who argues that under investment is most likely to occur when the market is down and aggregate investment is low. I measure the level of NFS disclosures by hand collecting the NFS information from sample firms annual reports filed with the SEC. I use the NFS information in the annual reports as a proxy for the level of NFS disclosures.4 The types of information collected in this study include those that are relatively more relevant to equity investors, such as products, customers, competitors, and backlogs, and those that are relatively more relevant to debt investors, such as contractual obligations and a firms existing credit facilities. These classes of information are not generally extractable from a firms financial statements per se but may play an important role in reducing information asymmetry between a firm and its potential stakeholders. Cole and Jones (2005) call for papers that examine the nature and role of specific nonfinancial -statement disclosures in MD&A and elsewhere in the Form 10 -K.5 The American Institute of Certified Public Accountants' (AICPA) report on improving business r eporting (Jenkins' report) also recommends research that improves the understanding of costs and benefits of business reporting. Due to the costs of hand collecting data, relatively few studies have examined the NFS disclosures provided by companies. By manually reading the sample firms annual reports, I provide a descriptive summary of firms de facto disclosure practices and investigate the effects of NFS disclosures on firms financing constraints. I choose the electronic equipment industry for this study The examination of NFS disclosures in the electronic equipment industry should be of particular interest to policy makers and academics because the costs and benefits of disclosures in this industry are particularly high 4 Lang and Lundholm (1993) show that analysts ratings of annual report disclosures are positively correlated with analysts ratings of disclosures provided through other venues. 5 MD&A refers to Management's Discussi on and Analysis of Financial Condition and Results of Operations It is one of the required disclosure items (typically item 7) in the 10K filings.
13 relative to other industrie s. On the cost side, the electronic equipment industry is characterized by fierce competition and high proprietary costs of disclosure. The majority of firms in the electronic equipment industry are involved in patent infringement litigations, either as plaintiffs or defendants. Disclosure of proprietary information, such as products in development or names of major customers, could have undesired consequences. On the benefit side, NFS information is one of the major inputs by which potential investors can value the electronic equipment companies. The industry is one of the four high-technology industries as classified by Francis and Schipper (1999) and is characterized by rapid technological change. The short product cycle in this industry can render a firm s investment in machine, equipment, and human capital obsolete in a short time. As a result, a great amount of relevant information cannot be gleaned from the financial statements per se; disclosures of NFS information can help investors better understand a firms business risks and assess a firms value. Moreover, due to the high uncertainty regarding a firms asset values, it can be difficult for an electronic equipment firm to borrow against existing assets. Consequently, disclosure of NFS information c an play a major role in facilitating the flow of funds from investors to the firm. My first research question examine s the association between the level of NFS disclosures and under -investment. The results indicate that a firm s NFS disclosure level is ne gatively associated with the degree of under investment, suggesting that the firm can relax its financing constraints and refrain from forgoing positive NPV projects by providing more NFS information. On average, a one -standard deviation increase in disclo sure level would increase investment by 2.92% of total assets.6 My second research question tests whether the levels of equity and debt related NFS disclosures are both associated with the degree of under investment. As expected, 6 The mean and median investments are 15.64% and 13.22% of total assets for the sample firms.
14 the results show that both disclosure levels are negatively associated with the degree of under investment. On average, a one -standard -deviation increase in equity and debt related disclosure levels increase investment by 2.32% and 1.50% of total assets, respectively. My last research question investigate s the association between the two disclosure levels and a firms subsequent financing decisions. If equity and debt related NFS disclosures reduce information asymmetry between a manager and potential equity and debt holders, one should expect the disclosure levels to be positively associated with subsequent equity and debt financing, respectively. For equity related disclosures, the result is consistent with the hypothesis; a firm with higher levels of equityrelated disclosures is more likely to issue equity in the subsequent period. This suggests that equity related NFS disclosures reduce information asymmetry between a manager and equity investors. However, for debt related disclosures, the result is not as expected. I do not find that debt related NFS disclosure level is positively associated with subsequent debt financing. Instead, the data show that the higher the debt -related NFS disclosure level, the more likely it is that the firms will issue equity in the subsequent p eriod. The results are consistent with the proposition that public disclosures of debt related information are informative to potential equity investors, but not to potential lenders. I conduct several robustness tests First, I rerun my empirical tests us ing ranks of disclosure score to minimize the effects of outliers ( Lang and Lundholm, 1993; Botosan, 1997; Hail, 2002). Second, I follow McNichols and Stubben (2008) and contr ol for log of growth of assets cash flow, lagged investment, and the interactions of Tobins q and its quintiles in my empirical specifications Third, I use a on e -year span to measure the sample firms investment and financing activities.7 Last, I use alternative weightings of disclosure items developed by 7 A two ye ar span was used in the primary analysis.
15 Stanga (1976) to test whether my results are robust to the choice of disclosure weightings. The results from my four robustness tests are qualitatively similar to those of my primary tests. Motivated by Tirole ( 2006, Chap 6, pg. 244), I measure under investment by comparing inv estment levels across firms at a time in which the market is down and the aggregate investment is low. To test if Tiroles (2006) theory is descriptiv e I investigate whether the effects of NFS disclosure on under investment vary across different macro con ditions. Consistent with Tirole s (2006) prediction, the results indicate that the effects of NFS disclosure on under investment are weaker in non recessionary periods, presumably due to the fact that under investment is less prevalent in such periods. If a firms voluntary disclosure behavior in the current period is linked to the firms plan to invest or issue equity in the subsequent period, my tests are subject to the endogeneity between disclosure and investment and that between disclosure and equity financing. To control for the reverse causality bias, I employ a simultaneous equations approach and re -estimate the coefficients using a two -stage least squares ( 2SLS) method. With this control I still find that a firm with a higher level of disclosure is less likely to under invest and more likely to issue equity in the subsequent period. The study contributes to the literature in two ways. First, the study extends the prior literature and investigates the role of corporate disclosure in mitigating the u nder -investment problem Specifically I provide empirical evidence on how equity and debt related disclosures are associated with under investment. Biddle and Hilary (2006) and Biddle et al (2008) examine the association between accrual quality and investment inefficiencies while this study focuses on the impact of corporate disclosure on investment inefficiencies Prior studies have examined the monitoring role of disclosure and investigated how disclosure mitigates the over -investment
16 problem. ( Bens & Monahan, 2004; Bushman, Piotroski, and Smith 2006; Hope and Thomas 2008) There has been limited evidence on how corporate disclosure mitigates the under investment problem. This study fills the void by providing evidence on the association between the level of NFS disclosure and degree of under investment. Second, this study bridges the gap between two major streams of the literature: the disclosure literature in accounting and the financing constraints literature in finance and economics Motivated by models of asymmetric information, a well developed stream of finance and economic s literature studies the existence of financing constraints.8 However, the literature has not examined the question of whether disclosure relaxes financing constraints by reducing the information asymmetry between managers and potential investors. The accounting literature examining the consequences of disclosure has largely focused on the negative relation between the disclosure level and the cost of capital. There has been limited evidence on how a firms disclosure level affects the managements real decisions. This study links the two streams of literature by examining the role of disclosure on the under investment problem and exploring how a firms disclosure practices affect the managers investment and subsequent financing decisions. 8 Financing constraint refers to situations in which a manager may pass up positive NPV projects due to problems in external financing. Therefore, the term can be used interchangeably with under investment.
17 CHAPTER 2 BACKGROUND AND LITER ATURE REVIEW The Disclosure Literature Healy and Palepus (2001) review clas sifies the empirical research on disclosure into four categories: disclosure regulation, information intermediaries, the determinants of disclosure, and the economic consequences of disclosure. The first category, disclosure regulation, concerns the effectiveness of disclosure regulation and the economic rationale that justifies regulating corporate disclosure. The second category, information intermediaries, centers on the role of information intermediaries, such as auditor s financial analysts, institutional investors, and banks. The th ird category, the determinants of disclosure, studies managements reporting choices. The last category, which is the focus of this study, investigates the economic consequences of disclosure. A line of empirical literature examining the economic conseque nces of disclosure focuses on the relation between disclosure level and the cost of capital. (Botosan, 1997; Sengupta, 1998; Botosan and Plumlee, 2000) This empirical inquiry is motivated by two streams of theoretical models. Diamond and Verrecchia (1991) demonstrate that a firm committed to a higher level of disclosures enjoys a lower cost of capital by limiting the adverse selection costs associated with information asymmetry.1 Barry and Brown (1985) establish a link between disclosure and the cost of c apital by focusing on risks associated with estimating firm -specific parameters. Other empirical studies that examine the consequences of disclosure include Welker (1995), who finds a positive relation between disclosure rankings and market liquidity, and Leone et al (2007), who show that the degree of IPO underpricing is associated with the specificity of disclosure on the use of proceeds in the prospectus. 1 Other studies related to this line of work include Kim and Verrecchia (1994) and McNichols and Trueman (1994)
18 This study explores one potential consequence of disclosure: the role of disclosure in mitigating investment inefficiencies. Specifically, the study investigates whether a firms disclosure practices reduce the wedge between the internal and external cost of capital, i.e., the level of mispricing, and prevent managers from forgoing positive net present value (NPV) projects. The theories that motivate this line of empirical work are developed in the financing constraints literature, which will be discussed in the next section. One important feature of this study is that it adds to the understanding of d isclosures role in a production economy. Dye (2001) contends that a descriptive model of efficiency-based disclosure must include production as an element .2 He argues that in a pure exchange economy the only function of disclosure is to redistribute weal th. In such a setting, disclosure would be at best irrelevant and at worst harmful (when agents are risk averse), and one would not observe any disclosure in equilibrium. In support of Dyes (2001) argument, Kanodia (2007) advocates a new approach to the s tudy of accounting and disclosure, one that focuses on the real effects of accounting This line of research addresses how accounting or disclosure affects a firm s real decisions and more generally resource allocation in the economy. The Financing Co nstraints Literature The benchmark model for the financing constraints3 literature is the frictionless market envisioned by Modigliani and Miller (1958) In such a perfect market, the costs of capital are the same for all projects in the same risk portfo lio, irrespective of how the projects are financed. Whether the projects are financed by equity, debt, or internal funds is irrelevant. 2 Efficiency based disclosure examines the efficiency gains or losses from firms ex ante commitment to disclosur e (Verrecchia 2001) 3 Financing constraint refers to situations in which a manager passes up positive NPV projects due to problems in external financing. The term is used interchangeably with under investment throughout the study.
19 Dev i ating from Modigliani and Millers (1958) perfect capital market framework, Myers and Majluf (1984) and Greenwald, Stiglitz and Weiss (1984) demonstrate that, in the presence of information asymmetry, a firm may pass up positive NPV projects due to costly external financing .4 In both studies, the manager is assumed to have perfect information about the value of the firms assets in place but the information is not available to potential shareholders. Since potential investors cannot distinguish a firm with a high value of assets from a firm with a low value of assets, they price all firms at the population average. A s a r esult, the high value firms shares are under p riced and the low value firms shares are over -priced. If the high value firms managers act on behalf of existing shareholders, they are reluctant to issue under -priced equity to fund investment pr ojects because the issuance will dilute existing shareholders claims. In some cases the dilution effects may outweigh the benefits of the investment projects. Hence it is to the current shareholders best interest to forgo those investment projects even t hough the projects NPV is positive. In equilibrium, the low value firms issue shares to fund their investment projects, but the high value firms forgo their positive NPV projects and under invest. The same model can be applied analogously in the debt market. (Myers and Majluf, 1984) Assuming firms are privately informed about their own default risks, those with lower risks are reluctant to borrow, leading to under investment. M otivated by models of a s ymmetric information, Fazzari et al (1988) empirical ly test the presence of financi ng constraints in the capital market by using investment -cash flow sensitivity as a measure of financing constraints.5 6 Fazzari et al (1988) show that investment -cash flow 4 Both studies are applications of the lemons proble m first considered by Akerlof (1970) 5 The argument is that if a firm is not financially constrained, its investment policy should not be correlated with changes in internal resources. 6 This assumption is questioned by a series of studies (e.g., Alti 2003; Cleary 1999; Erickson and Whited 2000; Kaplan and Zingales 1997).
20 sensitivities are larger for low -dividend firms .7 To the extent that high retention practices are proxies for financing constraints, the study provides the first empirical evidence of the existence of financing constraints in the capital market. Following Fazzari et al (1988), several studies have tested th e presence of financing constraints using different a priori class i fications, such as keiretsu (corporate group) affiliation (Hoshi et al., 1991) the presence of bond ratings (Whited, 1992), age, the status of exchange listing, the pattern of insider tra ding, or distribution of equity ownership (Oliner and Rudebusch, 1992). The p rior literatures classifications of constrained or unconstrained firms are based on the researchers arguments about which groups of firms should be subject to higher or lower l evels of information asymmetry, but the sorting criteria can proxy for theoretical constructs other than information costs. If a firms financing constraints are the consequences of asymmetric information, one should expect a negative relation between the firms disclosure practices and its levels of financing constraints. This argument motivates this study, which provides empirical evidence on whether a firms disclosure practices relax its financing constraints and mitigate the under investment problem. Concurrent Research and Incremental Contribution Economic theories suggest two types of investment inefficiencies: under -investment and over -investment The under -investment problem refers to situations in which a firm passes up positive NPV projects due t o information asymmetry between the manager and the potential stakeholders. (Myers and Majluf, 1984; Greenwald et al 1984) The over -investment problem pertains to the idea that a firms investments may be higher than the optimal level because the 7 Fazz ari et al (1988) assume that low dividend paying firms are financially constrained.
21 managem ent can extract private benefits by over investing (Jensen, 1986) .8 Alternatively, Heaton (2002) posits that managerial optimism could potentially explain both under and over investment because an optimistic manager is reluctant to issue equity when he/sh e mistakenly thinks the firms shares are under -valued or invests too much when he/she over estimates the future cash flows. Several accounting stud ie s have examined whether information quality is associated with investment efficiency. Biddle and Hilary ( 2006) study whether accruals quality is positively associated with investment efficiency. They use investment -cash flow sensitivity to measure investment efficiency By design, the measure cannot distinguish the effect of accruals quality on over -investment from its effect on under investment. Other researchers have stud ied how information quality mitigates the over -investment problem, which focuses on the monitoring role of disclosure. Bens and Monahan (2004) find that a firm s disclosure qua lity, measured by AIMR rankings, mitigates the over investment problem and is positively associated with the excess value of diversification. Hope and Thomas (2008) show that disclosure of geographic earnings can serve as a monitoring9 mechanism and limit managers over investment behavior. Bushman, Piotroski, and Smith (2006) examine whether timely loss recognition serves as a governance mechanism and curtail s managements over investment behavior when investment opportunities decline. My study differs fro m th e se stud ie s by focusing on the role of disclosure in mitigating the under -investment problem. 8 Potential reasons that management can extract private benefits from over investing include ( 1) payouts to shareholders reduce the resources under managements control, (2) managements compensation is positively related to growth in sales (Murphy, 1985) and (3) larger corporations offer more promotion opportunities as rewards to management 9 In a principal agent framework, monitoring refers to the principal s ability to observe or infer the agent s actions.
22 I am aware of only one recent study that addresses the role of information quality in mitigating the under -investment problem. Biddle, Hilary, and Verdi (2008) examine the effects of accrual quality on under investment and on over investment. My study makes an incremental contribution to their s in three ways. First, I focus on a firms disclosure practices, whereas their study focuses on the characteristics of accounting numbers (accruals quality). Second, I investigate the effects of both equityrelated and debt related disclosures on under investment Third, I examine the effects of equity and debt related disclosures on a firms subsequent financing decisi ons.
23 CHAPTER 3 HYPOTHESES DEVELOPME NT AND EMPIRICAL CONSTRUCTS Hypothesis Development My hypotheses about the effects of corporate disclosure on under investment are based on models developed by Myers and Majluf (1984) and Greenwald, Stiglitz and Weiss (1984). Both studies show that a firm may under -invest due to asymmetric information between its manager and potential stakeholders. Specifically, Myers and Majluf (1984) demonstrate that investors ex ante expectations on the standard deviation of a firms assets in place is positively associated with the loss in firm value and negatively associated with the probability that the firm will issue and invest (Myers and Majluf, 1984, p. 206) In other words, a firm is more likely to under -invest if the value of the firms assets in place has a high standard deviation as perceived by potential investors. This theoretical prediction forms the basis of my empirical tests. I investigate whether a firms level of NFS disclosure mitigates the under -investment problem by reducing potential investors perceived uncertainty in the value of the firms assets in place. Formally, I hypothesize the following: H1a: A firm that provides more NFS information is less likely to under invest than a firm that withholds su ch information. In Myers and Majluf (1984) and Greenwald, Stiglitz and Weiss (1984), the information that a manager has but that potential equity investors do not have is the information about the firms assets in place. In the debt version of the models, the information that a borrower has but that lenders do not have is the information about the firms default risks. I test whether equityand debt related NFS disclosures mitigate the under -investment problem by reducing information asymmetry between a fi rm and its potential equity and debt investors, respectively H1b: A firm that provides more equity related NFS information is less likely to under -invest than a firm that withholds such information.
24 H1c: A firm that provides more debt related NFS infor mation is less likely to under invest than a firm that withholds such information. The intuition behind the under -investment model is that whe n a firm's stock is under valued due to information asymmetry, the firm may refrain from issuing equity to finance positive NPV projects. Therefore, compared with a group of firm s that has low information asymmetry, i.e., more correctly priced, a group of firm s with high information asymmetry would have a higher pro portion of firms that are undervalued and thus under -invest. If one compares the average investment and financing activities of those two groups, one should expect the low information asymmetry group to have higher average investment and financing levels t han the high information asymmetry group. If disclosure of equityand debt related NFS information reduces the information asymmetry between a firm and its potential stakeholders one should expect the level of equityand debt related NFS disclosures to be positively associated with the level of subsequent equity and debt financing, respectively. H2a: A firm that provides more equity related NFS information is more likely to issue equity in subsequent periods. H2b: A firm that provides more debt related NFS information is more likely to issue debt in subsequent periods. Operationalization of Theoretical Constructs Identifying Under -Investment To identify under -investment, three alternative research designs are possible. One alternative is to follow Fazz ari et al (1988) and classify firms into low medi um and highdisclosure groups. Under this approach, the investment -cash flow sensitivity of each group is used as a proxy for under investment1 and the null hypotheses are rejected if the group with highe r disclosure activities exhibits lower investment-cash flow sensitivity than the group with 1 The investment cash flow sensitivity for each group is obtained by regress ing investment on cash flow and growth opportunities for firms in each group.
25 lower disclosure activities. The limitations of this research design are first, the use of investment -cash flow sensitivity cannot distinguish the effects of discl osure on under investment from that on over -investment because both under and over -investment predict a positive relation between investment and cash flow. Second, cash flow may capture measurement errors in Tobins q and, more importantly, the degree to which they are captured may vary across different groups of firms, creating an omitted variable problem (Kaplan and Zingales, 1997) The second option is to follow Richardson (2006) and McNichols and Stubben (2008) and specify an econometric model of expected investment. Under this approach, the negative residuals extracted from the empirical model are used as proxies for under -investment. The potential limitation of this approach is that it assumes that on average firms invest optimally because by c onstruction the mean of the residuals from a regression model is zero. (Bergstresser, 2006) The assumption is unwarranted. I adopt the third option, which is to follow Biddle et al (2008) and identify periods in which firms are more prone to under invest. Tirole (2006) argues that when the average probability of a projects success within an economy increases, i.e. when the economy is booming, it is more likely to have cross -subsidization between issuers (that is, no under investment) than to have market b reakdowns, which result in under investment.2 Based on this theoretical argument, I measure under -investment by comparing investment levels across firms at a time in which the market is down and the aggregate investment is low. The limitation of this approach is that even in periods in which firms are on average more prone to under invest, some firms may still over invest. This creates measurement errors in my empirical proxies. 2 In the cross subsidization scenario, under investment is absent even though the information problem creates a wealth transfer from the low risk borrowers to the highrisk borrowers. In the market breakdown scenario, under investment is p resent because the information problem prevents the low risk borrowers from getting the finance at a price that justifies the investment.
26 As Table 3 1 shows, the aggregate, mean, and median investment s for firms in th e COMPUSTAT universe drop s ignificantly in 2002 and reach their lowest level s of the 19952006 period in 2003. Accordingly, I measure my dependent variable using the sum of investment s in 2002 and 2003.3 The other variables are either measured at the end o f 2001 (if it is a stock variable) as the average of 2002 and 2003 (if it is a flow variable) or as the difference between 2001 and the average of 2002 and 2003 (if it is a change of a flow variable such as the change in cash flow ). Identifying Variatio ns in NFS Disclosure The specific disclosure in this study is the NFS information voluntarily disclosed by managers and the disclosure attributes associated with this construct are the existence and specificities of disclosure. I focus on NFS disclosures b ecause financial statement information is required and audited; there is not much cross -sectional variation in the existence or specificities of such information. On the other hand, NFS disclosure is highly discretionary, varies substantially across firms, and is an important form of voluntary disclosure for managers to communicate inf ormation to potential investors (Lang and Lundholm, 2000) The study of effects of such disclosures should be of interest to companies and to policy makers. The disclosure in dex in this study is a self -constructed, empirically -based measure. I first conduct a pilot study and read the annual reports of 20 randomly selected sample firms to identify key NFS disclosures provided in their reports. I summarize the firms de facto di sclosure practices and include in the index only the NFS information that is verifiable or hard so that 3 I use a two year interval to measure the cross sectional variation in firms investment level. The investment litera ture does not have a consensus on the functional form of the adjustment cost of investment. The standard neoclassical model assumes that the adjustment cost is a convex function of investment, but the empirical studies find otherwise. (Doms and Dunne, 1998; Cooper et al. 1999) To avoid potential problems caused by making the wrong assumptions on the adjustment cost, I use a period of middle length (two years) to measure the cross sectional variation in firms investment. I also conduct a robustness test usi ng a one year span.
27 objective coding is feasible. Soft information, such as disclosures about a companys strategies, is not included in the index because it is very di fficult to assign scores to such information. To the extent that important information is communicated through this soft disclosure, this omission could reduce the power of my empirical tests. The variation in a firms disclosure levels examined in this study is attributed to the firms voluntary disclosure practices. Regulation S -K dictates the disclosure requirements for a public ly traded firms 10 -K filings, but the firm has much discretion in determining how much NFS information to provide above the minimum legal threshold. Similar to Botosan (1997) and Hail (2002), this study examines the variation in disclosure level across firms at a point in time. By design, the observed variability in disclosure level is attributed to a firms voluntary disclosur e behavior since the legal requirements for annual report disclosures are constant across reporting firms at a point in time. Following Botosan (1997) and Hail (2002), I use the level of NFS disclosures in the annual reports ( in form 10 -Ks) as a proxy for a firms overall level of NFS disclosures.4 I construct three disclosure scores: DISC_ALL, which measures the overall level of NFS disclosure; DISC_E, which measures the level of NFS disclosures that are relatively more relevant to equity holders; and DISC_D, which measures the level of NFS disclosures that are relatively more relevant to debt holders. DISC_ALL includes key disclosure items in the 10-Ks that are (1) not extractable from financial statements per se, but are (2) verifiable or hard so that the coding of the information is possible. T he disclosure index consists of 15 main categories, including 12 categories under which there are two to six sub -categories. The sub -category measures the 4 Lang and Lundholm (1993) find that analysts ratings of annual report disclosures are positively correlated with analysts ratings of disclosures provided through other venues. However, it is possible that my measures do not adequately capture the NFS disclosures revealed through other venues, such as earnings press release and conference calls. Thus, the noise in the empirical disclosure index constructed in this paper may potentially induce measurement error.
28 specificities, or qualities, of the main category disclosures. Three main categories company backgrou nd, industry background, and contractual obligations do not have sub -categories. I split the disclosure items in DISC_ALL into two groups, DISC_E and DISC_D, to measure the levels of NFS disclosures that are relatively more relevant to equity and debt inve stors, respectively. Theoretically, an e quity investor is interested in information th at affects the mean of a firm s value distribution ,5 whereas a debt investor is interested in information that affects the left tail of the value distribution, i.e., the probability of default Therefore, I include in DISC_D the disclosure items that are relatively more informative about a firms liquidity positions and default risks. The rest of the disclosure items are included in DISC_E.6 The index of my disclosure scores is provided in Appendix B and the examples of each disclosure item are provided in Appendix D DISC_E includes nine of the 15 main categories and DISC_D includes the remaining six. Most of the disclosure items are extract ed from Item 1, Business, and Item 7, Managements Discussion and Analysis of Financial Condition and Results of Operation (MD&A), of the annual reports on Form 10 -K. The information on pending lawsuits is from Item 3, Legal Proceedings. Firms typically p rovide liquidity and default risk related information (DISC_D) under the Liquidity and Capital Resources Section of MD&A. Regulation S -K requires all public ly traded firm s to disclose information on liquidity [Item 303(a) (1)] and capital resources [Item 3 03(a) (2)], but s par imilar to other NFS disclosures, 5 Assuming he or she is risk neutral. 6 Arguably, all disclosure items are relevant both to equity and debt investors so that it is not possible to label any disclosure items as only relevant to equity investors or only relevant to debt investors. It is unlikely that a piece of information only changes an individuals belief on the left tail of a distribution without altering his or her perceptions on the mean, or vice versa. Therefore, the classification is based on a subjective assessment of the relevance of each disclosure item to equity inve stors compared to that of debt investors. I acknowledge that the noise in the proxies for DISC_D and DISC_E could induce measurement error and reduce the power of my test
29 the firm has much discretion in determining how much and what to provide about its default risks and liquidity positions. To minimize human bias, the scoring system is based on counting the number of ite ms disclosed. Each disclosure sub-category counts as one point, with the exception of company background, industry background, and contractual obligations. Those three main disclosure categories do not have sub categories and thus receive higher weights. T hey count as two points.7 The final scores for DISC_ALL, DISC_E and DISC_D are calculated by dividing the raw counts by the maximum possible counts each company could have obtained. The maximum possible raw count for the first nine main categories under D ISC_E is 24, whereas the maximum possible raw count for the last six main categories under DISC_D varies from company to company. If a firm does not have Term Loans, Revolving Credits, Forward Contracts/Swaps, or Public Debt the scores under each main cat egory are excluded when calculating the maximum possible raw counts for the firm. By construction, all three disclosure scores are numbers between 0 and 1. The descriptive statistics for the disclosure scores are provided in Appendix C.8 Identifying Variat ions in Financing Activities To identify variations in a firm s financing activities, I divide my sample firms into four groups: high equity and high debt financing, high equity and low debt financing, low equity and high debt financing, and low equity and low debt financing. A firms equity or debt financing activity is considered high if the level of the firms equity or debt financing is higher than the industry median. I use a multinomial logit model to test the relation between a firms disclosure practices and its financing activity. The choice of a multinomial logit model is motivated by the 7 The inferences do not change if each of these three categories is counted as one p oint. 8 Later in Chapter 6, a robustness test is presented using an alternative weighting of disclosure items.
30 fact that my response variable, the choice of equity or debt financing, is a set of categories that cannot be ordered in meaningful ways.
31 Table 3 1. Descriptive statistics: Intertemporal pattern of aggregate investment Fis cal Mean Median Aggregate Year No. of Firms Investment Investment Investment 1995 5714 16.15% 9.69% 7.57% 1996 5852 15.17% 10.18% 7.65% 1997 5713 15.87% 10.87% 7.32% 1998 5966 19.25% 12.27% 6.97% 1999 5751 15.72% 10.08% 6.27% 2000 5385 13.46% 9.49% 6.44% 2001 5181 13.47% 8.82% 5.46% 2002 5371 13.23% 7.59% 4.66% 2003 5479 12.03% 7.09% 4.07% 2004 5521 14.92% 8.06% 4.27% 2005 5392 13.46% 8.31% 4.57% 2006 4433 14.00% 8.73% 4.85% Investment = capex (data128) + R&D (data46) + Acquisition (data129) sale of PPE (data107) Mean Investment = mean of (investmenti / asseti) ,i = 1 to the number of firms Median Investment = median of (investmenti / asseti) ,i = 1 to the number of firms Aggregate Investment = sum of investme nti / sum of asseti, i= 1 to the number of firms
32 Table 3 2. Descriptive statistics: Investment patterns of manufacturing firms, separated by industries No. of aggregate median aggregate median aggregate median Fama French Industry Firms Capex Capex R&D R&D PPE PPE Pharmaceutical Products 280 4.56% 2.57% 10.65% 20.42% 21.25% 10.23% Electronic Equipment 266 5.15% 2.71% 7.54% 8.85% 23.22% 13.78% Computers 166 3.77% 2.51% 7.13% 9.62% 12.97% 8.31% Medical Equipment 149 4.34% 2.66% 6.46% 6.58% 19.34% 13.26% Machinery 130 3.80% 2.72% 3.13% 2.44% 19.69% 17.36% Measuring and Control Equip 103 3.82% 2.40% 8.55% 9.40% 15.18% 12.16% Chemicals 70 3.91% 3.45% 2.54% 1.55% 35.54% 30.90% Construction Materials 66 3.56% 3.31% 0.62% 0.04% 37.56% 32.13% Electrical Equipment 65 2.79% 2.64% 2.79% 3.26% 20.34% 21.42% Apparel 61 3.11% 2.44% 0.15% 0.00% 17.15% 13.36% Consumer Goods 56 4.60% 3.25% 3.29% 1.12% 29.08% 19.70% Food Products 54 3.09% 4.22% 0.42% 0.00% 24.76% 34.65% Automobiles and Trucks 50 4.70% 3.99% 2.15% 2.05% 18.42% 23.99% Steel Works, Etc. 47 3.61% 3.01% 0.51% 0.00% 44.48% 39.01% Business Supplies 44 3.48% 3.47% 1.32% 0.75% 40.18% 36.73% Printing and Publishing 32 2.35% 2.47% 0.07% 0.00% 18.05% 18.12% Rubber and Plastic Products 26 3.97% 3.82% 1.31% 1.16% 29.14% 34.51% Aircraft 20 2.29% 2.90% 3.24% 1.11% 20.16% 21.86% Median (Capex R&D, or PPE) = median of (Capex, R&D, or PPE / total asset) Aggregate Investment = sum of Capex, R&D or PPE / sum of total asset, The figures reported are the average of 20012003
33 CHAPTER 4 SAMPLE SELECTION AND EMPIRICAL SPECIFICATION Sample Selection Following the prior literature on investment and financi ng constraints (Fazarri et al, 1988; Almeida et al, 2004) I use manufacturing firms (SICs 2000 to 3999) as my initial sample. There are 1 ,754 U.S. manufacturing firms with available data on COMPUSTAT for 2001 200 3 My variable of interest is the level of NFS disclosure s in firms annual reports Because of the h igh labor costs of hand collecting the disclosure dat a I considered two alternatives in selecting a subsample. The first option is to randomly select a small subsample from each industry. The second option is to restrict my sample to one industry. I adopt the second option. P rior studies have focus ed on particular industries because disclosure practices are typically industry specific and may not be comparable across industries.1 Since my variable of interest is the level of NFS disclosur es that are relevant to potential equity and debt investors, I focus on a single industry so that the effects of disclosure can be compared across firms. I partition the available manufacturing firms based on their Fama -French (1997) industry classificatio n. Table 3 2 shows the descriptive statistics of manufacturing firms investment activities during the 2001 to 2003 period, grouped by industry.2 T he two industries with the largest number of firms are pharmaceutical products (2 80 firms) and electronic eq uipment (266 firms). I conducted a pilot study to examine what the major investment activities are for these two industries. I restrict my final sample to firms in the 1 For example, Amir and Lev (1996) examine the NFS disclosure s in the wireless communication industry. Francis, Schipper, and Vincent (2003) stu dy the NFS disclosures in the airline, homebuilding, and restaurant industry. 2 T his two year period is my specification of aggregate under investment; see description of the de pendent variable in Chapter 3.
34 electronic equipment industry because the industrys investment activities include both fixed capital and human capital investment. 3 There are 284 U.S. electronic equipment firms traded on NYSE/AMEX/NASDAQ in the 2001 COMPUSTAT Annual Industrial File. I exclude 18 firms that are missing in the 2002 or 2003 file.4 Out of the balanced panel of 266 firms, 13 are excluded due to missing financial variables. To obtain the disclosure data, I exclude five firms that are small business issuers,5 one firm that is a foreign issuer,6 and 25 firms with missing MD&As.7 Thus, my final sample includes 222 firms in the electronic equipment industry. Table 4 1 summarizes the sample selection procedure and Table 4 2 presents the descriptive statistics for my final 222 firms. The intertemporal trend of industry-wide investment for the electronic equipment indus try is provided in Table 4 3.8 3 The investment activity for pharmaceutical f irms is predominantly research and development expense (human capital investment), mainly composed of salaries paid to scientists and researchers. Hall (2002) argues that research and development spending (investment in human capital) has very high adjust ment costs so that its response to change in economic conditions (including the mispricing of firms shares) may be sluggish. As a result, it would be difficult to detect under investment in an industry with extremely high R&D spending. My choice of elect ronic equipment firms as my final sample would have more power to detect under investment. However, I acknowledge that the results may not be generalizable to industries with extremely high R&D spending. 4 Out of the 18 firms that are missing in the 2002 or 2003 file, 17 were acquired or merged with other companies and one had a corporate restructuring. 5 Small business issuers file form 10KSB. 6 Foreign issuers file form 6 K 7 Regulation S K Item 303 requires that firms provide MD&As in their annual repor ts on 10 K but some firms incorporate their MD&As by reference to exhibits (typically Exhibit 13) or to their annual reports to shareholders.7 If the MD& As are incorporated by reference to exhibits and the information is available through the SECs website, the observation is kept in the sample. Twenty five firms are dropped because first, they incorporated their MD&As by reference to their annual reports to shareholders and second, they do not have their 2001 annual reports to shareholders available on the company websites in 2009 when the paper is written. 8 I choose year 2002 and 2003 to be my sample years because the US aggregate investment is lowest in those two years. Since the industry wide investment may exhibit a different intertemporal pattern than the economy wide investment, I present the intertemporal trend of industry wide investment for the US electronic equipment firms. As shown in Table 43, the aggregate investment of the US electronic equipment firms reached its lowest level in 2002 and 2003, which validates my selection of those two years to measure under investment.
35 Empirical Specification Disclosure and Investment: Dependent Variable To measure the association between firms NFS disclosure levels and under -investment, Ordinary Lease Square s (OLS) regression is used. Following Richardson (2006), I measure the dependent variable, INVEST, as the sum of capital expenditure, R&D expenditure, and acquisitions, minus sale of property, plant, and equipment, deflated by total assets multiplied by 100 so that INVEST is expressed as a percentage of total assets:9 INVEST = [Capital Expenditure + R&D expenditure + Acquisitions Sale of PPE ], d eflated by t otal a sset s and expressed as percentage points. Disclosure and Investment: Independent Variables My variables of interest are the three disclosure scores: DISC_ALL, DISC_E, and DISC_D. The first disclosure score, DISC_ALL, measures a firms overall disclosure level of NFS information in the annual reports and is used to test Hypothesis 1a, whether NFS d isclosures mitigate under investment. My second and third disclosure scores, DISC_E and DISC_D, measure the level of NFS disclosures that are relatively more relevant to equity and debt holders, respectively. Those two scores are used to test Hypotheses 1b and 1c, whether equity and debt -related disclosures mitigate under -investment. My control variables include the following: Tobins Q (TOBIN_Q predicted sign: positive), Size (SIZE predicted sign: positive), Leverage (LEVERAGE, predicted sign: negative ), Slack (SLACK predicted sign: positive), and Bankruptcy (BANKRUPT predicted sign: unknown). 9 An alternative proxy used by prior literature to measure investment is capital expenditure deflated by property, plant and equipment. I adopt Richardsons (2006) measure because, as shown in Table 1, R&D constitutes a major share of the total investment expenditures for electronic equipment firms. Aggregate R&D as a percentage of assets is 7.54%, whereas aggregate capital expenditures as a percentage of assets is 5.15% for the electronic equipment firms.
36 Tobins (1969) q theory predicts that the rate of investment (the speed at which a firm invests in its fixed capital) should be positively associated with marginal q, the ratio of the market value of new capital to its replacement cost. Following Rauh (2006), I measure Tobins Q using the market to book ratio of firm assets. The numerator is the sum of the market value of equity and book assets minus the sum of t he book value of common equity and deferred taxes. The denominator is book assets. Kiyotaki and Moore (1997) argue that a firms asset base exerts significant influence on the firms ability to fund its investments. An initial negative shock to asset prices reduces the collateral value of firms with smaller asset bases, thereby leading to less investment since the firms ability to borrow deteriorates with the drop in collateral value. Lower investment generates lower profit and further depresses asset pr ices. The effect is amplified through the cycle. I control for the effect of asset base on under investment by including log sales in my empirical model. Myers (1977) suggests that the conflict of interest between bondholders and shareholders might cause managers to forgo some positive NPV projects. He argues that when default risk is high, part of the value created by the investment projects is appropriated by the bondholder. When the wealth transfer from stockholder to bondholder is larger than the NPV of the investment projects, it is in the best interest of current stockholders to forgo those projects even though the NPV of the investment is positive. I control for such possibility by including leverage in my empirical model. As argued in Myers and Ma jluf (1984), if a firms internal resources (slack) are sufficient to fund all positive NPV projects, there will be no under investment problem. I
37 control for the effect of slack on under investment by including cash balances deflated by total assets in my empirical model. The slack variable measures the available internal resources at the beginning of the period, but a contemporaneous cash flow shock may also affect firms investment behavior as argued in Fazarri et al (1988). I control for unexpected cas h flow shocks by including the change in operating cash flow deflated by total assets in my empirical model.10 Fazarri et al (1988) find that financially constrained firms have higher cash flow investment sensitivities. Kaplan and Zingales (1997) show contr adicting results by imposing different criteria to identify financially constrained firms. Hubbard (1998) suggests that the inconsistencies between the two studies may be because the behavior of financially constrained firms is different from that of finan cially distressed firms. I control for the effect of financial distress on under investment by including Altmans Z -score (Altman, 1968) multiplied by negative one as a measure of bankruptcy risk, so that higher BANKRUPT corresponds to higher bankruptcy ri sk. Disclosure and Investment: Empirical Specifications The following are m y empirical specifications for H ypotheses 1a, 1b, and 1c: H1a : NFS Disclosure and Under Investment INVESTt 0 1DISC_ALLt 1 2TOBIN_Qt 1 3SIZEt 1 4LEVERAGEt 1 + 5SLACKt 1 6CF_SHOCKt + 7BANKRUPTt 1 (3 1) H1b and H1c: Equity and Debt Related Disclosures and Under Investment INVESTt 0 1DISC_Et 1 1DISC_Dt 1 3TOBIN_Qt 1 4SIZEt 1 5LEVERAGEt 1 6SLACKt 1 7CF_SHOCKt + 8BANKRUPTt 1 (3 2) 10 I also estimate the empirical model using cash flow instead of change in cash flow; the inferences do not change.
38 t = the average of 2002 and 2003 t 1 =2001 Disclosure and Financing: Dependent Variables To measure the association between firms disclosure level and managers subsequent financing decisions, a multi nomial logit model is used. The dependent variable, FIN_CHOICE, is defined as follows: FIN_CHOICE = 1, if E_ISSUE = HI and D_ISSUE = HI FIN_CHOICE = 2, if E_ISSUE = HI and D_ISSUE = LO FIN_CHOICE = 3, if E_ISSUE = LO and D_ISSUE = HI FIN_CHOICE = 4, if E_I SSUE = LO and D_ISSUE = LO E_ISSUE is HI if firms level of net equity issuance is higher than the industry median. E_ISSUE is LO otherwise. D_ISSUE = HI if firms change in longterm debt is higher than the industry median. D_ISSUE is LO otherwise. Disc losure and Financing: Empirical Specifications My dependent variable for H ypotheses 2a and 2b is FIN_CHOICE and my independent variables are DISC_E and DISC_D, plus controls. I use the same control variables as in equations ( 3 1) and ( 3 2) as the multinom ial logit test is an extension of models (3 1) and ( 3 2). For the multinomial test, I use FIN_CHOICE = 4, the group with LO E_ISSUE and LO D_ISSUE, as my base because the studys interest lies in understanding which firms under -invest and why. My empirica l model for H ypotheses H2a and H2b is as follows: H2a and H2b: NFS Disclosures and Subsequent Equity and Debt Financing P (FIN_CHOICE = j)t = F (DISC_Et 1, DISC_Dt 1, TOBIN_Qt 1, SIZEt 1, LEVERAGEt 1, SLACKt 1,CF_SHOCKt, BANKRUPTt 1), j = 1 4 (3 3)
39 t = the average of 2002 and 2003 t 1 =2001 Table 4 1. Sample Selection U.S. Electronic Equipment Firms traded on NYSE, AMEX and NASDAQ 284 from the 2001 COMPUSTAT Annual Industrial File Less Firms that are not in the 2002 and 2003 COMPUSTAT Annual Industrial File (18) A Balanced Panel of U.S. Electronic Equipment Firms traded on NYSE, AMEX 266 and NASDAQ from the 2001 to 2003 COMPUSTAT Annual Industrial File Financial Data Less Firms with missing financial variables (13) Firms with available financial data 253 Disclosure Data Less Firms that are small business issuers (file 10 KSB) (5) Less Firms that are foreign issuers (file 6 -K) (1) Less Firms missing MD&As (ITEM 7) (25) Number of Firms in the Sample 222 Table 4 2. Descriptive Statistics Variables No. of Obs. Mean Std. Dev. Median 25% 75% INVEST 222 15.64 11.28 13.22 8.04 20.02 DISC_ALL 222 0.484 0.110 0.483 0.421 0.559 DISC_E 222 0.565 0.141 0.583 0.458 0.667 DISC_D 222 0.495 0.227 0.500 0.364 0.636 TOBIN_Q 222 2.120 1.665 1.675 1.084 2.412 SIZE 222 4.778 1.754 4.614 3.714 5.892 LEVEAGE 222 0.127 0.173 0.037 0.000 0.212 SLACK 222 0.319 0.236 0.271 0.109 0.513 CF_SHOCK 222 0.010 0.135 0.003 0.061 0.078 BANKRUPT 222 7.644 11.445 4.568 9.132 1.818 All variables are defined in Appendix A.
40 Table 4 3 Intertemporal pattern of industry -wide investment : Electronic Equipment Firms Fiscal Mean Median Aggregate Year No. of Firms Investment Investment Investment 1995 300 19.19% 16.94% 22.97% 1996 303 20.72% 18.14% 22.68% 1997 306 20.02% 17.08% 19.96% 1998 339 28.92% 20.48% 18.29% 1999 321 19.92% 16.24% 16.55% 2000 300 17.23% 14.54% 18.16% 2001 284 18.37% 16.40% 18.37% 2002 302 19.60% 14.56% 13.13% 2003 301 17.10% 12.82% 12.14% 2004 301 16.70% 12.96% 13.47% 2005 300 17.30% 14.51% 15.36% 2006 236 17.78% 14.03% 15.45% Investment = capex + R&D + Acquisition sale of PPE Mean Investment = mean of (investmenti / asseti) ,i = 1 to the number of firms Median Investment = median of (investmenti / asseti) ,i = 1 to the number of firms Aggregate Investment = sum of investmenti / sum of asseti, i= 1 to the number of firms
41 CHAPTER 5 EMPIRICAL RESULTS NFS Disclosure and Investment Hypotheses In Chapter 2 I provide the theoretical motivation linking NFS disclosure and Investment. Accordingly, I hypothesize the following: H1a A firm that provides more NFS information is less likely to under invest than a firm that withh olds such information. H1b A firm that provides more equity -related NFS information is less likely to under -invest than a firm that withholds such information. H1c A firm that provides more debt related NFS information is less likely to under -invest than a firm that withholds such information. Results and Discussion Table 5 1 presents the results of an OLS regression giving the association between the level of NFS disclosures and under -investment. Column 1 presents the results of the benchmark model, the re gression of investment on variables identified from the prior literature. Columns 2 and 3 present the results of empirical tests for H ypotheses 1a, 1b, and 1c. Consistent with H ypothesis 1a, Column 1 of Table 5 1 shows that the coefficient on DISC_ALL is p ositive and significant, suggesting that a firm providing more NFS information is less likely to under invest. On average, a one -standard deviation increase in DISC_SCORE increases investment by 2.92%of total assets.1 For H ypotheses 1b and 1c, Column 2 of Table 5 1 shows that both DISC_E and DISC_D are positive and significant, suggesting that both equity and debt related NFS disclosures reduce information asymmetry between managers and potential stakeholders and mitigate the under investment pr oblem On average, a one -standard -deviation in DISC_E and DISC_D increase s investment by 2.32% and 1.50% of total assets, respectively. For control 1 Th e mean and median investments as a percentage of total assets are 15.6% and 13.2%, respectively.
42 variables, the coefficients on TOBIN_Q and SLACK are positive and significant, consistent with the conclusi ons that firms with higher unrealized growth opportunities and higher cash positions invest more. Surprisingly, the coefficients on SIZE and CF_SHOCK are negative and significant, suggesting that firms invest less if they are larger or experience a positi ve cash flow shock NFS Disclosure and Financing Choice Hypotheses In Chapter 2 I provide the theoretical motivation linking NFS disclosure and financing activities. Accordingly, I hypothesize the following: H2a: A firm that provides more equity related NFS information is more likely to issue equity in subsequent periods. H2b: A firm that provides more debt related NFS information is more likely to issue debt in subsequent periods. Results and Discussion Table 5 2 presents the results of the association between the levels of equityand debt related NFS disclosures and subsequent financing choices. I use a multinomial logit model to carry out the analysis. The base group is (FIN_CHOICE = 4), which is composed of firms that issue less debt and less equity relative to their industry peers. (LO E_ISSUE, LO D_ISSUE) The odds ratios are reported for ease of interpretation.2 The odds ratios represent the estimated proportional change in Prob (FIN_CHOICE = i) / Prob (FIN_CHOICE= 4) in response to a one unit cha nge in the independent variables. Consistent with Hypothesis 2a, I find that the estimated change in the odds ratios of being P1 (HI E and HI D) vs. P4 (LO -E and LO D) and of being P2 (HI -E and LO D) vs. P4 (LO E 2 An alternative would be to report the estimated coefficients, which can be obtained by taking the natural logarithm of the odds ratios.
43 and LO D) in response to a change in DISC _E is greater than one (the odds ratios are 21.926, 21.475, respectively) and statistically significant, suggesting that a firm providing more equityrelated NFS information is more likely to issue equity to fund its investment in the subsequent period. For Hypothesis 2b, the results are not as expected. The odds ratio of P1 (HI E and HI D) vs. P4 (LO -E and LO D) group for DISC_D is greater than one but only marginally significant (z -value 1.28).The odds ratio of P3 (LO -E and HI D) vs. P4 (LO E and LO -D) gr oup for DISC_D is less than one, implying that the sign of the coefficient is of the opposite direction. Based on the test results, I fail to reject Hypothesis 2b and cannot conclude that a firm that discloses more information about its default risks is mo re likely to issue debt to finance its investment in the subsequent period. Table 5 3 presents the marginal effects of the multinomial model. The odds ratios reported in Table 5 2 provide information on the pairwise comparisons of P1, P2, and P3 vs. P4 w hereas the marginal effects in Table 5 3 show the estimated change in probabilities of being in each group in response to a one unit change in each independent variable, holding other variables constant. The partial derivative, ean of each independent variable. Column 2 of Table 5 3 reports the marginal effects for the change in each independent variable for each group. Column 3 of Table 5 3 reports the inter -quartile range of each independent variable. To assess the economic sig nificance of the change in subsequent financing choices in response to a change in each independent variable, Column 4 of Table 5 3 shows how the probability of being in each group would have changed if each independent variable increases from its 25th per centile to its 75th percentile while holding all other variables at their means.
44 Consistent with Hypothesis 2a, when DISC_E increases from its 25th percentile to its 75th percentile, the estimated probabilities of being in P1 (HI -E and HI D) and P2 (HI -E and LO D) increase by 6.19% and 5.47%, respectively, suggesting that a firm with higher level of equity -related NFS disclosures is more likely to issue equity in the subsequent period. The results for DISC_D are inconsistent with hypothesis 2b. When DISC_D increases by its inter quartile range, the estimated probabilities of being in P1 (HI -E and HI D) and P2 (HI E and LO D) increase by 6.53% and 2.65%, respectively, but the estimated probability of being in P3 (LO E and HI D) decreases by 6.71%, suggesting that a firm with higher level of debt related NFS disclosures is more likely to issue equity, but less likely to issue debt. Combining this result with that from the investment model (higher DISC_D is associated with higher investment), the data appears to be consistent with that NFS disclosures on a firms default risks reduce information asymmetry between its manager and potential equity investors, but not between its manager and potential debt investors, as originally hypothesized. For the control var iables, the results show that a firm with higher market -to -book ratios is more likely to issue equity, consistent with the market timing hypothesis (Baker and Wurgler, 2002). I also find that larger firms (measured by log sales) are more likely to issue debt to fund their investment projects, consistent with Gertler and Gilchrist (1994). Moreover, the results are also consistent with the trade -off theory; a firm with higher debt ratios and bankruptcy risks is more likely to issue equity.
45 Table 5 1. Disc losure level and investment Expected Sign Benchmark Hypothesis 1a Intercept 17.634 6.629 6.912 (6.86) (1.95) (1.94) DISC_ALL + H1a 26.553 *** (4.89) DISC_E + H1b 16.45 *** (3.30) DISC_D + H1c 6.607 ** (2.14) TOBIN_Q + 1.314 ** 1.263 *** 1.274 *** (2.29) (2.56) (2.57) SIZE + 1.178 *** 1.383 *** 1.39 *** ( 2.75) ( 3.43) ( 3.38) LEVERAGE 6.889 7.366 ** 7.091 ( 1.57) ( 1.70) ( 1.66) SLACK + 13.311 *** 10.429 *** 10.317 *** (3.38) (2.75) (2.66) CF_SHOCK + 11.585 13.217 ** 13.336 ** ( 1.74) ( 2.05) ( 2.05) BANKRUPT ? 0.315 *** 0.285 *** 0.282 *** (3.73) (3.79) (3.67) R square d 20.34% 26.71% 26.65% No. of obs. 222 222 222 Whites  heteroskedasticity adjusted t -values are provided in parentheses below each coefficient. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one tailed test if the sign is pre -determined) A ll variables are defined in Appendix A
46 Table 5 2. Disclosure level and financing choice: Odds Ratios Expected Odds Ratio Expected Odds Ratio Expected Odds Ratio Sign P1 vs.P4 Sign P2 vs.P4 Sign P3 vs.P4 DISC_E + (H2a) 21.926 ** + (H2a) 21.475 ** 5.533 (1.91) (1.89) (1.1) DISC_D + (H2b) 3.696 2.155 + (H2b) 0.526 (1.28) (0.73) ( 0.67) TOBIN_Q 0.936 1.871 *** 0.762 ( 0.30) (2.62) ( 1.16) SIZE 0.857 1.007 1.218 ( 1.09) (0.06) (1.47) LEVERAGE 0.001 *** 0.905 0.002 *** ( 3.26) ( 0.07) ( 3.39) SLACK 1.179 0.387 0.305 (0.14) ( 0.80) ( 1.00) CF_SHOCK 21.846 23.728 15.39 (1.84) (1.83) (1.64) BANKRUPT 0.968 1.1 ** 0.948 ( 1.04) (2.26) ( 1.62) Log likelihood 255.49 Psedo R sq. 16.98% No. of obs. 222 z -values are provided in parentheses below each coefficient. **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one -tailed test s for DISC_E and DISC_D and two tailed tests for the other variables) All variable s are defined in Appendix A Table 5 3. Disclosure level and financing choice: Marginal Effects P1 P2 P3 P4 (HIE, HI D) (HIE, LO D) (LO E, HI D) (LO E, LO D) DISC_E 6.19% 5.47% 0.73% 10.93% DISC_D 6.53% 2.65% 6.71% 2.46% TOBIN_Q 2.70% 14.21% 10.89% 0.62% SIZE 8.85% 0.39% 10.20% 0.96% LEVERAGE 17.08% 13.72% 15.07% 18.42% SLACK 6.19% 4.17% 7.42% 5.40% CF_SHOCK 3.17% 3.10% 2.13% 8.40% BANKRUPT 6.26% 17.22% 9.76% 1.20% The table presents the estimated change in probabilities of being in each group in response to a one unit change in each independent variable, holding other variables constant. The marginal effects, mean of each independent variable. All variables are defined in Appendix A.
47 CHAPTER 6 ROBUSTNESS TESTS Using the Ranks of Disclosure Scores Following Lang and Lundholm (1993), Botosan (1997), and Hail (2002), I rerun my investment models using t he ranks of DISC_ALL, DISC_E, and DISC_D to minimize the effect of outliers. The results are presented in Columns 1 and 2 of Table 6 1. The coefficients on RANK_ALL, RANK_E, and RANK_D are all positive and significant and thus my inferences from the primar y tests are robust to this specification. Controlling for Higher Orders of Tobins q and Lag Investment Following McNichols and Stubben (2008), I control for log of asset growth, cash flow, lagged investment, and the interactions of Tobins q and its qui ntiles in this specification. The results are presented in Columns 3 and 4 of Table 6 1. The coefficients on DISC_ALL and DISC_E are positive and significant at the 5% level. The coefficient on DISC_D is positive but insignificant at the conventional leve l. Overall, my inferences for DISC_ALL and DISC_E from my primary tests are robust to this specification but my inference for DISC_D is not. Using a One -Year Span to Measure Investment I also conduct a robustness test using a one -year span to measure the cross -sectional variation in firms investment levels. The results are presented in columns 1 and 2 of the Table 6 2. The coefficients on DISC_ALL, DISC_E, and DISC_D are positive and significant and thus the inferences from my primary tests are robust to this specification. Alternative Weightings of NFS Disclosure Items Since the weightings on the disclosure item are subjectively determined, I test whether the results from the main analysis are robust to alternative weighting schemes. Using responses fro m questionnaires sent to a group of Chartered Financial Analysts, Stanga (1976) develops a
48 weighting of annual report disclosure items based on the analysts assessment of the relative importance of each disclosure item. Using Stangas (1976) weighting to construct the disclosure scores, I rerun the regression for my investment model. The results are provided in Column 3 and 4 of Table 6 2. The coefficients on DISC_ALL_S, DISC_E_S, and DISC_D_S1 are all positive and significant. The results from my primary tests are robust to alternative weightings of disclosure items. 1 DISC_ALL_S, DISC_ E_S, and DISC_D_S are the disclosure scores constructed using Stangas (1976) weightings.
49 Table 6 1. Robustness tests: rank and M -S specifications Dep INVEST INVEST INVEST INVEST Exp. Ranks Ranks M S M S Sign Spec. Spec. Spec. Spec. Intercept 14.504 12.81 3.199 2.86 (5.36) (4.51) (1.04) (0.91) DISC_ALL + 12.348 ** (2.27) DISC_E + 9.933 (2.07) ** DISC_D + 1.271 (0.49) RANK_ALL 0.043 *** (4.34) RANK_E + 0.037 *** (3.20) RANK_D + 0.022 ** (1.91) TOBIN_Q + 1.228 ** 1.243 ** 0.824 0.778 (2.43) (2.45) ( 0.26) ( 0.25) SIZE + 1.335 *** 1.348 *** ( 3.27) ( 3.28) LEVERAGE 7.59 7.205 ( 1.74) ( 1.69) SLACK + 10.532 *** 10.221 *** (2.79) (2.67) CF_SHOCK + 12.754 ** 6.423 ( 1.99) ( 2.05) BANKRUPT ? 0.28 *** 0.281 *** (3.67) (3.63) Q_QRT2 1.929 1.874 ( 1.01) ( 0.99) Q_QRT3 0.157 0.107 ( 0.07) ( 0.05) Q_QRT4 0.831 0.779 (0.30) (0.28) CF 18.334 *** 17.726 *** ( 3.36) ( 3.28) GROWTH 0.06 0.104 (0.04) ( 0.07) LAG_INVEST 47.025 *** 46.85 *** (5.42) (5.34) R square d 26.03% 26.55% 45.01% 45.19% No. of obs. 222 222 222 222 W hites  heteroskedasticityadjusted t values are provided in parentheses below each coefficient. **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one -tailed test if the sign is pre -determined) M S specification refers to empirical models in McNichols and Stubben (2008) All variables are defined in Appendix A.
50 Table 6 2. Robustness tests: one year specific ation and alternative weighting schemes Dep INVEST_02 INVEST_02 INVEST INVEST Exp. 1 Year 1 Year Stanga Stanga Sign Spec. Spec. Spec. Spec. Intercept 8.722 9.033 8.63 ** 9.895 *** (2.60) (2.64) (2.56) (2.73) DISC_ALL + 29.274 *** (4.17) DISC_E + 17.801 *** (3.14) DISC_D + 7.787 *** (2.39) DISC_ALL_S 16.989 *** (3.99) DISC_E_S + 9.183 ** (2.05) DISC_D_S + 6.775 ** (2.22) TOBIN_Q + 1.938 *** 1.943 *** 1.397 *** 1.344 *** (3.33) (3.32) (2.78) (2.63) SIZE + 1.799 *** 1.82 *** 1.329 *** 1.392 *** ( 3.57) ( 3.60) ( 3.27) ( 3.35) LEVERAGE 10.894 ** 10.693 ** 6.959 7.483 ( 2.43) ( 2.40) ( 1.6) ( 1.75) SLACK + 7.332 7.286 10.899 *** 11.375 *** (1.90) (1.80) (2.85) (2.95) CF_SHOCK + 26.976 27.071 12.651 12.594 ( 1.96) ( 1.95) ( 1.93) ( 1.9) BANKRUPT ? 0.315 *** 0.311 *** 0.304 *** 0.295 *** (3.24) (3.16) (3.91) (3.63) R square d 34.14% 34.21% 24.65% 24.47% No. of obs. 222 222 222 222 Whites  heteroskedasticityadjusted t -values are provided in parentheses below each coefficient. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one -tailed test if the sign is pre -determined) All variables are defined in Appendix A
51 CHAPTER 7 ENDOGENEITY OF NFS DISCLOSURE One concern when using OLS to estimate the effects of NFS disclosure on under investment is the possibility of endogeneity between disclosure and investment and that between disclosure and financing. Lang and Lundholm (1993, 2000) and Frankel, McNichols, a nd Wilson (1995) find that firms tend to disclose more when they need to access the capital market. Although the use of a lagged disclosure variable in the main tests might mitigate against the issue of reverse causality flowing from investment and financi ng to disclosure, it is still possible that a firm committed to a higher level of future investment and financing chooses to disclose more in the current period. If this is the case, ordinary least square s estimation will yield biased and inconsistent coef ficient estimates. I address this issue by employing two sets of simultaneous equation systems and estimating both sets of equations using the method of two-stage least squares (2SLS). The structural equations are as follows: Disclosure and Investment: DIS C_ALLt 1 01INVEST t 2SIZEt 1 3AB_RETt 1 4N_ANALt 1 5CORRt 1 6SURPt 17LOSSt 1t 1t 1 10INST_OWNt 1 11OUT_DIRECt 1 (7 1) INVEST t 0 1DISC_ALLt 12TOBIN_Qt 13SIZEt 14LEVERAGEt 1 5SLACKt 16CF_SHOCK t 7BANKRUPTt 1 (7 2 ) Disclosure and Equity Financing: DISC_ALLt 1 01INVEST t 2SIZEt 1 3AB_RETt 1 4N_ANALt 1 5CORRt 1 6SURPt 17LOSSt 1t 1t 1 10INST_OWNt 1 11OUT_DIRECt 1 (7 3)
52 E_FINt 1 0 1DISC_ALLt 1 2TOBIN_Qt 1 3SIZEt 1 4LEVERAGEt 1 + 5SLACKt 1 6CF_SHOCKt 7BANKRUPTt 1 (7 4) Equations 7 1 and 7 2 represent the set of equations for disclosure and investment. Equations 7 3 and 7 4 represent the set of equations for disclosure and equity financing. The identical independent variables in Equations 7 1 and 7 3 are drawn from prior literature (e.g. Lang and Lundholm, 1993; Lang, Lins and Miller, 2003; Ajinkya et al 2005).1 The endogenous variables in the two sets of simultaneous equations ar e DISC_ALL and INVEST and DISC_ALL and E_FIN.2 t 1 against all exogenous variables and then use the predicted DISC_ALL to estimate Equations 7 2 and 7 4. I use four different specification s to construct my instrumental variables because I face a tradeoff between missing observations and the strength of the instruments. If I use all variables in Equations 7 1 and 7 3 to construct my instrumental variables, my sample size would drop signific antly to 135 and my estimation of Equations 7 2 and 7 4 would suffer from low power and selection bias.3 If I use only variables with relatively fewer missing observations to construct my instrumental variable, I would have a larger sample size with 206 observations but possibly biased or inconsistent in such a small sample. For completeness, I report the results for all four specifications. The variables that suffer the most from missing observations are the 1 AB_RET is the market adjusted return for 2001. N_ANAL is the number of analysts following the firm. CORR is the earnings returns correlation for the eight quarters ending in 2001. SURP is the absolute value of the difference between 2001 EPS and 2000 EPS, deflated by price. LOSS equals 1 if 2001 earnings are negative, and 0 otherwise BIG_FIVE equals 1 if a firm hires a big five auditor, and 0 otherwise RET_STD is the standard deviation of monthly returns for year 2000 and 2001. Firms must have at least 20 observations to be included in the calculation. INST_OWN is the percentage of shares owned by institutions. OUT_DIREC is the percentage of outside directors. The earnings data are from COMPUSTAT. The returns data are from CRSP. The analysts data are from FIRST CALL. The governance data (INST_OWN and OUT_DIREC) are from Compact Disclosure. 2 E_FIN measures a firms level of equity financing in 2002 and 2003, as a percentage of total assets. 3 Most likely I will lose firms that are small and young.
53 standard deviation of returns (RET_STD) and the governance variables (INST_OWN and OUT_DIREC). Accordingly, my four specifications are (1) excluding both RET_STD and governance variables, (2) excluding governance variables, (3) excluding RET_STD, and (4) including all variables. As shown in Table 7 1 and 7 2, the coefficients of DISC_ALL for Equation 5, the investment model, are all positive and statistically significant at the 1% level for all four specifi cations. The lower bound for the coefficients of DISC_ALL from the four specifications is 49.23, which is larger than 26.55, the coefficient of DISC_ALL from the main test. The coefficients of DISC_ALL for Equation 74, the equity financing model, are posi tive in all specifications and statistically significant in three out of the four specifications.4 Overall, the results suggest that the inferences from my main tests do not change under the four 2SLS specifications. Caution must be exercised when interpre ting the results of the 2SLS estimation. As widely discussed in the literature, the fundamental problem with the instrumental variable (IV) technique is the difficulty of finding proper instruments variables that are correlated with the endogenous regresso r but uncorrelated with errors in the structural equation. In Table s 7 1 and 7 2 I report the results of four alternative 2SLS specifications with varying degree s of sample siz e. In the following paragraphs I report and discuss alternative 2SLS specificat ions with varying degrees of strength on the instruments. The first potential problem of the 2SLS approach stems from a weak correlation between the endogenous regressor and the instruments. Bound et al (1995) demonstrates that when the correlation betw een the instruments and the endogenous regressor is weak, a small correlation between the instruments and the errors will lead to large inconsistencies in the IV estimates. 4 The coefficients on DISC_ALL for the equity financing model vary significantly across the four specifications, suggesting that the estimations are sensitive to the loss of sample firms.
54 Moreover, the problem is exacerbated if the sample size is small. Table 7 3 repor ts the partial R square d partial F -statistic and the coefficients of the first stage estimation under alternative combinations of instruments.5 The estimated coefficient and t -statistic on DISC_ALL from the structural equation for each specification is a lso reported. In the first specification, I include all nine instruments identified from the prior disclosure literature. The results are shown in Column 1 of Table 7 3. The partial R -square d and partial F -statistic are 27.64% and 5.05 for this specificati on. The partial R -square d is reasonable, but the partial F -statistic is too low to warrant predictive power, e.g., p-value larger than 10%. The results are reported i n Column 2 of Table 7 3. The partial R -square d and partial F -statistic are 25.27% and 10.49. The partial R -square d is similar to that of the first specification but the partial F -statistic improve s from 5.05 to 10.49. statistically significant. The second potential problem of the 2SLS approach is the possibility that the instruments are correlated with errors i n the structural equation. Among the five instruments from my second specification, institutional ownership is most likely to be endogenous. A firm with high institutional ownership may be less likely to under invest because institutional investors provide financing through private placement upon receiving private information from the management. I address this concern by treating institutional ownership as an endogenous variable. The results are reported in Column 3 of Table 7 3. The partial R -square d and partial F -statistic are 19.08% 5 In a 2SLS estimation, the endogenous variable is first regressed on the instruments plus the control variables from the structural equation. T he R squared and F statistic obtained from the first stage estimation measure the explanatory power and its statistical significance for all variables included in the first stage model. To measure how well the instruments explain the endogenous variable, partial R squared and partial F statistic are the proper statistics. They measure the explanato ry power and its statistic al significance for the instruments only.
55 and 9.75; the coefficient on DISC_ALL is 61.40. My inference does not change when I treat institutional ownership as an endogenous variable. To sum up, Table 7 3 shows that the coefficients on DISC_ALL are all positive and statistically significant under three alternative 2SLS specifications. But I cannot conclude that I have addressed all the concerns on the endogeneity problem due to the weak instruments and small sample size.
56 Table 7 1. Simultaneous equations Excluding RET_STD and Governance Excluding Governance First INVEST E_FIN First INVEST E_FIN Stage Stage DISC_ALL 104.397 *** 52.769 ** 67.632 *** 14.295 (2.91) (2.33) (3.59) (1.31) SIZE 0.004 2.017 *** 1.603 *** 0.016 1.878 *** 1.263 *** (0.5) ( 3.25) ( 4.08) (1.97) ( 3.67) ( 4.25) TOBIN_Q 0.003 0.519 0.963 0.002 0.646 1.173 *** (0.45) (0.6) (1.75) (0.17) (0.89) (2.8) LEVERAGE 0.007 5.463 4.142 0.009 4.901 3.743 (0.13) ( 0.85) (1.01) (0.17) ( 0.91) (1.19) SLACK 0.058 4.615 0.505 0.061 8.551 2.644 (1.27) (0.77) ( 0.13) (1.27) (1.8) (0.96) CF_SHOCK 0.05 15.425 ** 3.879 0.084 12.76 0.42 (0.79) ( 1.97) ( 0.79) (1.28) ( 1.88) ( 0.11) BANKRUPT 0.001 0.16 0.109 0.001 0.185 0.144 ** (0.47) (1.23) (1.33) (0.47) (1.73) (2.32) AB_RET 0.004 0.004 (0.45) (0.45) N_ANAL 0.001 (0.001 (0.37) ( 0.46) CORR (0.003 (0.001 ( 0.14) ( 0.03) SURP 0.015 0.002 (0.92) ( 0.14) LOSS 0.045 *** 0.037 ** (2.46) (2.01) BIG_FIVE 0.039 0.055 (1.27) (1.8) RET_STD 0.314 *** (4.39) INTERCEPT 0.373 *** 25.76 16.42 0.226 *** 9.865 0.616 (8.61) ( 1.64) ( 1.65) (4.24) ( 1.16) ( 0.12) No. Obs. 206 191 Adj. R Sq. 0.0408 0.1276 t -values are provided in parentheses below each coefficient. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one -tailed test if the sign is pre -determined) All variables are defined in Appendix A.
57 Table 7 2. Si multaneous Equations: Continued Excluding RET_STD and Governance Including All Variables First INVEST E_FIN First INVEST E_FIN Stage Stage DISC_ALL 49.227 *** 25.758 ** 50.93 *** 6.119 (2.9) (1.99) (3.48) (0.63) SIZE 0.008 1.995 *** 1.734 *** 0.008 2.05 *** 1.41 *** ( 0.81) ( 3.75) ( 4.29) (0.84) ( 3.75) ( 3.87) TOBIN_Q 0.001 1.444 ** 1.322 ** 0.003 1.444 ** 1.374 *** (0.13) (2.11) (2.54) ( 039) (2.04) (2.91) LEVERAGE 0.031 4.633 7.987 0.041 3.75 7.364 ** (0.5) ( 0.84) (1.9) (0.68) ( 0.67) (1.98) SLACK 0.118 ** 3.23 3.665 0.137 ** 1.48 6.604 (2.24) (0.6) (0.9) (2.48) (0.27) (1.78) CF_SHOCK 0.103 14.982 ** 1.693 0.158 16.11 ** 1.785 (1.37) ( 2.09) ( 0.31) (1.95) ( 2.02) (0.34) BANKRUPT 0.002 0.26 ** 0.14 0.001 0.232 ** 0.153 ** (1.5) (2.58) (1.83) (1.28) (2.24) (2.22) AB_RET 0.001 0.002 ( 0.09) (0.21) N_ANAL 0.001 0.002 ( 0.42) ( 1.37) CORR 0.019 0.025 ( 0.77) ( 0.99) SURP 0.035 0.002 (1.27) (0.07) LOSS 0.069 *** 0.06 *** (3.4) (2.9) BIG_FIVE 0.039 0.052 (1.17) (1.58) RET_STD 0.335 *** (4.00) INST_OWN 0.002 *** 0.001 *** (3.41) (3.25) OUT_DIREC 0.000 0.000 ( 0.11) ( 0.15) INTERCEPT 0.346 *** 0.544 4.8 0.178 ** 0.06 2.003 (5.74) (0.07) ( 0.84) (2.47) (0.01) (0.46) No. Obs. 148 135 Adj. R Sq. 0.1965 0.2802 t -values are provided in parentheses below each coefficient. ** and *** indicate significance at the 10%, 5%, and 1% levels, respectively. All variables are defined in Appendix A.
58 Table 7 3. Simultaneous equations with alternative combinatio n of instruments Include All Instruments Exclude Weak Instruments Inst_Own is Endogenous DISC_ALL 47.96 *** 41.21 *** 61.4 *** (Main Regression) (3.38) (2.9) (3.48) SIZE 0.00892 0.01301 0.01301 (0.9 0 ) (1.39) (1.39) TOBIN_Q 0.00324 0.00104 0.00104 ( 0.40) ( 0.14) ( 0.14) LEVERAGE 0.03891 0.02489 0.02489 (0.65) (0.43) (0.43) SLACK 0.13646 ** 0.16003 *** 0.16003 *** (2.47) (3.09) (3.09) CF_SHOCK 0.17318 ** 0.16539 ** 0.16539 ** (2.14) (2.09) (2.09) BANKRUPT 0.00134 0.00162 0.00162 (1.23) (1.57) (1.57) AB_RET 0.00146 (0.17 ) N_ANAL 0.0033 ** 0.00377 ** 0.00377 ** ( 2.03) ( 2.44) ( 2.44) CORR 0.0303 ( 1.21) SURP 0.00421 (0.15) LOSS 0.06173 *** 0.0612 *** 0.0612 *** (2.99) (3.19) (3.19) BIG_FIVE 0.05091 (1.54) RET_STD 0.32445 *** 0.29598 *** 0.29598 *** (3.88) (3.79) (3.79) INST_OWN 0.00164 *** 0.0017 *** 0.0017 *** (3.59) (4.12) (4.12) OUT_DIREC 0.0002 ( 0.38) INTERCEPT 0.17816 ** 0.19901 0.19901 (2.6) (3.42) (3.42) Adj. R Square d 27.75% 28.40% 28.40% Partial R Sq. 27.64% 25.27% 19.08% Partial F 5.05 10.49 9.75 t -values are provided in parentheses below each coefficient. All variables are defined in Appendix A. P artial R -squared and partial F -statistic measure the explanatory power and the statistic al significance for the instruments (excluding the control variables) in the first stage estimation
59 CHAPTER 8 THE ROLE OF NF S DISCLOSURE IN NON -RECESSIONARY PERIODS In this chapter I examine the association between a firms level NFS disclosure and its investment and financing activities during a non-recessionary period. The analysis is to contrast the findings in Chapter 5, w here I present evidence of positive associations between a firms level of NFS disclosure and its investment and financing activities when the US aggregate investment is low. The findings in Chapter 5 are used to support my hypothesis that NFS disclosure reduces under investment. To maximize the power of my empirical tests, I chose a period of low aggregate investment as my sample period because I expect under investment to be more prevalent in such period. The theoretical argument for this claim is provid ed by Tirole (2006), who demonstrate that a firm is less likely to under -invest when the average probability of a projects success within an economy increases.6 Assuming Tiroles (2006) argument is descriptive of the real world, one should expect the effe cts of NFS disclosure on investment and financing to be weaker in non recessionary periods. In this chapter, I empirically test this prediction. To investigate the role of NFS disclosure during a nonrecessionary period, I use data from fiscal year 2005 to 2007. As shown in Table 3 1 the US aggregate investment peaked in 1998, dropped to its lowest level in 2003, and slowly increased after 2003.7 I use 20052007 instead of 19961998 as my sample period because Regulation Fair Disclosure (FD) was not enact ed until 6 The intuition behind Tiroles (2006) claim is that when a manager makes decisions on whether to forgo some positive NPV projects, she trades off the costs of transferring wealth to the new shareholders with the costs of forgoing the projects. If the project being forgone becomes highly profitable, the manager would be more prone to undertake the project at the expense of transferring wealth to the prospective shareholders. 7 Fiscal year 2007 is the most current data available at the time this study is conducted.
60 October 2000 and the private communication between management and institutional investors may contaminate my results. Similar to my 20012003 sample, my 20052007 sample focuses on one industry and includes only electronic equipment firms. The NF S disclosure data are collected from the sample firms 2005 annual reports and the investment and financing data collected from COMPUSTAT Annual Industrial File for 2006 and 2007. The descriptive statistics for both samples are provided in Table 8 1. As ex pected, firms in the 20052007 sample invest more and have higher market -to -book ratio, higher sales, lower leverage, more cash, and lower bankruptcy risks. The scores for the equity related disclosure are approximately the same for both samples but the sc ores for the debt related disclosure are higher for the 20052007 sample. Part of this increase is attributed to the mandatory disclosure requirements of contractual obligations adopted by the SEC under t he Sarbanes Oxley Act Table 8 2 compares the relative effects of NFS disclosure on under investment between the 20012003 sample and the 20052007 sample. Column 1 of Table 8 2 presents the results for the benchmark model. Column 2 of Table 8 2 reports the results of the regres sion of investment on NFS disclosure plus controls. Y2001 and Y2001*DISC_ALL capture the intercept and the incremental slope effects for the 2001 2003 sample. The coefficient on Y2001 is 8.25, suggesting that on average the 20012003 sample firms invest 8.25% of total assets less than the 20052007 sample firms. The coefficient on Y2001*DISC_ALL is positive and statistically significant, suggesting that the effects of NFS disclosure on under -investment is stronger in the 20012003 period, consistent with T iroles (2006) prediction. Column 3 of Table 8 2, reports the regression of equity financing on NFS disclosure. The coefficient on Y2001 is 4.06, suggesting that on average the 20012003 sample firms issue 4.06% of total assets less than the 20052007
61 sam ple firms. The coefficient on Y2001*DISC_ALL is positive and marginally significant, suggesting that the effects of NFS disclosure on equity financing is stronger in the 20012003 sample. These results are consistent with the prediction that the effects o f NFS disclosure on under investment are weaker in non -recessionary periods, presumably due to the fact that under investment is less prevalent in such periods.
62 Table 8 1. Descriptive statistics: NFS disclosure in a non recessionary period 2001 2003 2 005 2007 Difference No. of Obs. 222 227 Variables Mean (Std) Mean (std) INVEST 15.64 17.47 1.827 (11.28) (15.427) DISC_ALL 0.484 0.48 .0008 (0.11) (0.179) DISC_E 0.563 0.54 0.028 (0.141) (0.214) DISC_D 0.495 0.59 0.095 (0.227) (0.240) TOBIN_Q 2.12 2.36 0.237 (1.665) (1.642) SIZE 4.778 5.11 0.335 (1.754) (1.964) LEVEAGE 0.127 0.084 0.043 *** (0.173) (0.126) SLACK 0.319 0.36 0.042 (0.236) (0.236) CF_SHOCK 0.010 0.016 0.026 (0.135) (0.174) BANKRUPT 7.64 7.83 0.182 (11.445) (11.427) *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (two -sample, two tailed t test) Standard deviation of each variable is provided in the parentheses. All variables are defined in Appendix A.
63 Table 8 2. Disclosure and under investment: nonrecessionary period Predicted Sign Benchmark Investing Equity Financing Intercept 15.54 *** 12.34 *** 13.63 *** (7.29) (4.40) ( 3.10) Y2001 8.25 ** 4.06 ( 2.33) ( 0.91) DISC_ALL + 11.56 *** 2.24 (2.44) (0.33) Y2001*DISC_ALL + 16.21 ** 12.00 (2.26) (1.41) TOBIN_Q + 1.52 *** 1.55 *** 2.54 *** (2.87) (3.02) (4.80) SIZE + 0.96 ** 1.25 *** 3.46 *** ( 2.55) ( 3.24) ( 5.81) LEVERAGE 2.49 3.04 9.72 ** ( 0.61) ( 0.72) (2.41) SLACK + 14.66 *** 12.44 *** 2.14 (5.18) (4.49) ( 0.51) CF_SHOCK + 31.99 ** 33.00 ** 16.66 ** ( 2.09) ( 2.18) ( 2.18) BANKRUPT ? 0.32 *** 0.32 *** 0.30 *** (3.63) (3.78) (2.71) R square d 27.97% 31.44% 25.84% No. of obs. 449 449 449 W hites  heteroskedasticityadjusted t values are provided in parentheses below each coefficient. **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one tailed test if the sign is pre -determined) All Variables are defined in Appendix A.
64 CHAPTER 9 AN ALTERNATIVE MEASURE OF DISCLOSURE: MANAGEMENT EARNINGS GUIDANCE In this chapter I examine the association between disclosure and under -investment using an alternative measure of corporate disclosure. Motivated by models of asymmetric information, I investigate the effects of NFS disclosure on under -investment in the ma in analysis of this study. Since the theories linking disclosure to under investment provide little guidance as to what types or forms of disclosure should be relevant to reduce information asymmetry between managers and potential stakeholders1, I examine whether my results from the main analysis are robust to alternative measure of corporate disclosure: the frequency and consi stency of management earnings forecasts Management earnings forecasts represent s one of the key voluntar y disclosure mechanisms a manager use s to communicate to the capital market her private information about the firms earning prospects ; it is shown to alter analysts and investors expectations, (Ajinkya and Gift, 1984), affect stock prices (Pownall, et al. 1993), and reduce bidask spreads (Coller and Yohn 1997). If management earnings forecasts contain value relevant information they should mitigate the under -investment problem by reducing information asymmetry between a firm and its potential stakeholders. In this chapter I em pirically test this prediction. M anagement earnings forecasts ha ve various characteristics such as the form, horizon, level of detail, and historical pattern I focus on the historical pattern of management earnings 1 Empirical researchers have used a number of constructs to measure the quality of corporate disclosure, such as AIMR rankings (Lang and Lundholm, 1993, 1996; Welker, 1995; Healy Hutton, and Palepu, 1999, Nagar, Nanda and Wysocki, 2003, and Bens and Monahan, 2004), self constructed measures from annual re ports (Botosan, 1997, Hail, 2002), frequency and precision of management issued guidance (Pownall, Wasley, and Waymire, 1993; B aginski and Hassell, 1990; Coller and Yohn, 1997), and conference calls (Frankel, Johnson, and Skinner, 1999; Bushee, Matsumoto, and Miller, 2003).
65 forecasts because I am interested in the long -term effects of management earnings forecasts2 and therefore credibility matters Shielded by the Private Securities Litigation Reform Act of 1996, a manager may opportunistically issue optimistic forecast prior to equity offerin gs to boost stock price and maximize the proceeds from the offering. (Lang an d Lundholm, 2000) Anticipating such opportunistic behavior, a rational investor may discount an earnings forecast if it is made right before the equity offering. However, a firm that frequent ly and consistent ly provide s earnings forecasts may have lesser problem in reducing information asymmetry because the reputation of a transparent disclosure policy lends credibility to the cu rrent forecast (Hutton and Stocken, 2007) Following Ajinkya et al (2005), I use two variables to measure the historical pattern of earnings forecasts : the frequency and consistency of management earnings forecasts The first variable is frequency (FREQ) defined as the total number of annual earni ngs forecasts issued by a firm in the three years ending in 2001. 3 This variable measures how frequen tly a firm issues earnings forecasts The second variable is occurrence (OCCUR) defined as the number of years (out of three years ending in 2001) in wh ich management issued at least one annual earnings forecast. This variable measures how consistently a firm issues earnings forecasts. Those two variables are correlated but capture different characteristics of a firms forecast pattern A firm can be a fr equent but inconsistent forecaster such as one that issues sev eral forecasts in one year but none in the next. Alternatively, a firm can be a consistent but infrequent forecaster such as one that issues one forecast every year. 2 I am studying how management earnings guidance affects a firms investment and financing activities for th e subsequent two year period. 3 My sample years in measuring under investment are 2002 and 2003. Therefore, my sample years in measuring disclosure are the three years ending in 2001.
66 The purpose of using an a lternative measure of disclosure to test the theory of under investment is that NFS disclosure and management earnings forecasts complement each other in terms of generalizability and selection bias. The use of NFS disclosure from annual reports has less external validity due to the costs of hand collecting data, but the sample contains a more complete age and size portfolio.4 The use of management earnings forecasts has more external validity, but I lose younger firms by using thi s measure.5 To mitigate the endogeneity bias discussed in Chapter 8, 2SLS estimation is used.6 The empirical model is as follows: 01INVESTt 2SIZEt 1 3AB_RETt 1 4N_ANALt 1 5CORRt 1 + 6SURPt 1 7LOSSt 1 8 BIG_FIVEt 1 9 RET_STDt 1 0INST_OWNt 1 (9 1) 0 12TOBIN_Qt 1 3SIZEt 1 4LEVERAGEt 1 5SLACKt 1 6CF_SHOCKt + 7BANKRUPTt 1 8INST _OWNt 1 (9 2) The descriptive statistics are reported in Table 9 1. Out of the 2619 sample firms with available financial data, 53.84% (1410/2619) did not issue any forecasts in 19992001. Only 8.13% (213/2619) made at least one forecast every year for the three years ending in 2001. The results of the 2SLS estimations with investment as the dependent variable are reported in Table 9 2. The coefficients on FREQ and OCCUR are 1.08 and 3.99, both positive and statistically significant, suggesting that a firm that provide s earnings forecasts more frequently and consistently is less like to under invest. 4 I only required firms to have data for 20012003. 5 I require firms to have data from 19982003 since I need at least three years to identify firms that consistently update their earnings information. 6 I also treat institutional ownership as an endogenous variable to avoid the possible correlation between institutional ow nership and errors in the structural equation. See Chapter 8 for details.
67 Table 9 1 Management earnings forecasts and investment: d escriptive statistics FREQ No. Obs. OCCUR No. Obs. 0 1,410 0 1,410 1 5 996 1 661 610 165 2 335 1115 33 3 213 1620 12 > 20 3 Total Obs. 2,619 2,619 Table 9 2 Management earnings forecasts and under -investment: 2SLS regression results Predicted Sign FREQ OCCUR Intercept 17.59 *** 17.42 *** (2.78) (2.76) FREQ + 1.08 ** (2.07) OCCUR + 3.99 *** (2.53) TOBIN_Q 0.74 *** 0.78 *** (5.94) (6.22) SIZE + 2.14 *** 2.22 *** ( 6.00) ( 6.79) LEVERAGE 1.77 1.56 ( 1.54) ( 1.33) SLACK + 9.12 *** 9.71 *** (6.95) (6.97) CF_SHOCK + 28.70 *** 28.66 *** ( 25.34) ( 24.84) BANKRUPT ? 0.036 *** 0.036 *** (6.53) (6.44) INST_OWN + 2.45 *** 1.28 (2.73) (1.2) Industry Controls Omitted Omitted R square d 40.38% 38.10% No. of obs. 2619 2619 First Stage Statistics Instruments: AB_RET, N_ANAL, CORR, SURP, LOSS, BIG_FIVE, RET_STD Adj. R Squared 26.3% 27.4% Partial R Squared 2.55% 2.47% Partial F statistics 9.54 9.24 z -values are provided in parentheses below each coefficient. ** and *** indicate significance at the 10%, 5%, and 1% levels, respectively (one -tailed test if the sign is pre -determined) All variables are defined in Appendix A.
68 CHAPTER 10 CONCLUSION AND FUTUR E RESEA RCH Myers and Majluf (1984) and Stiglitz and Weiss (1981) posit that information asymmetry between a firm and its potential stakeholders may lead to under investment due to adverse selections in the equity and debt market. This study investigates whether d isclosure of NFS information mitigates the under investment problem. Moreover, the study empirically tests whether NFS disclosures that are relatively more relevant to equity and debt holders mitigate the under investment problem by reducing the informatio n asymmetry between managers and equity holders and that between managers and debt holders, respectively. To test these predictions, the study first examines the association between firms overall level of NFS disclosures and under -investment. The results indicate that a firm with higher level of NFS disclosures is less likely to under -invest. On average, a on e -standard -deviation increase in disclosure level increases investment by 2.92% of total assets. Next, the study investigates whether NFS disclosures that are relatively more relevant to equity and debt holders are both associated with the level of under investment. As expected, the results show that both equity and debt related NFS disclosures are negatively associated with the degree of unde r -investment. On average, a one -standard deviation increase in equityand debt -related disclosure levels in crease s investment by 2.32% and 1.50% of total assets, respectively. Finally, the study investigates the association between the levels of equity and debt related NFS disclosures and firms subsequent financing decisions. If equity and debt related NFS d isclosures reduce information asymmetry betw een managers and equity holders and that between managers and debt holders, one should expect equity and debt related disclosure levels to be associated with subsequent equity and debt financing, respectively. F or disclosures that are more relevant to equity holders, the results are consistent with the hypothesis. The study finds that the level of
69 equity -related NFS disclosures is positively associated with subsequent equity financing, suggesting that equityrela ted disclosures facilitate the flow of funds from equity investors to the firms. For debt related NFS disclosures, the results are not as expected. The study does not find a positive relation between the level of debt related disclosures and subsequent deb t financing. Instead, the level of debt related disclosures is positively associated with subsequent equity financing, suggesting that public disclosure of debt -related NFS information provides useful information to potential equity holders, but not to pot ential lenders. One of the limitations of the study is that due to the cost of hand collecting the disclosure data, the sample is restricted to firms in one industry (the electronic equipment industry). The results may not be generalizable to other indust ries with different disclosure practices and firm characteristics. On the other hand, this weakness is also the strength of the study. The study complements prior literature by including more size and age portfolios in the sample. The use of AIMR rankings and accrual quality tends to exclude firms that are young and small.1 Since the focus of this study is on financing constraints, it is important to include young and small firms in the sample. Another limitation of the study is that the NFS discl osures ex amined in this study are endogenous. The study uses 2SLS estimation in an attempt to mitigate the endogeneity problem and finds results consistent with the primary tests, but the fact that the instruments are weak and the sample size is small subjects the findings to careful interpretation. Overall, the study finds that disclosures of NFS information facilitate the flow of funds from investors to the firms by reducing information asymmetry between managers and potential stakeholders. The study also finds that the effect of NFS disclosure in mitigating the under 1 AIMR rankings tend to include larger firms because those firms are more likely to be followed by analysts. Estimating accruals quality requires a longer time series so that younge r firms are typically excluded from the sample.
70 investment problem is mostly attributed to its role in reducing information asymmetry between firms and equity investors Last the study finds that the effects of NFS disclosure on under investment are weaker in non -recessionary periods, as predicted by Tirole (2006) One possible avenue for future research would be to understand why public disclosure of debt -related information has limited effect in reducing information asymmetry between managers an d potential lenders. One possible explanation is that the inherent level of information asymmetry between borrowers and lenders is low. Banks, which are the dominant players in the credit market, are experts in assessing the default risks of their potentia l borrowers. A firm is unlikely to know more about its default risk than its bank does. A second possible explanation is that the communication between borrowers and lenders is carried out privately. If this is the case, what is the efficiency implication of this private communication? What prevents potential debt investors from demanding a level playing field as equity investors do by demanding Reg FD? In other words, what is the an economic rationale for allowing banks to be the information intermediaries in the debt market but prohibiting analysts or institutional investors from assuming similar role in the equity market?
71 APPENDIX A VARIABLE DEFINITIONS Variables Definitions INVEST [capital expenditure + r&d expenditure + acquisitions sale of ppe] / total assets expressed as percentage points DISC_ALL levels of overall nonfinancial statement disclosures in firms annual reports DISC_E levels of equity related nonfinancial statement disclosures in firms annual reports DISC_D levels of debt related nonfinancial statement disclosures in firms annual reports TOBIN_Q [(common shares outstanding*price+total assets) (book value of equity+deferred tax)] / total assets SIZE log (sales) LEVEAGE (long term debt + short term debt) / (long term de bt + short term debt + (price*common shares outstanding) + preferred stock) SLACK cash / total assets CF_SHOCK ( cash flow from operating activities t cash flow from operating activities t 1 )/ total assets BANKRUPT 1 z_score = 1 [(1.2*working capital/total assets) + (1.4*retained earnings/total assets) +(3.3*pretax income/total assets) +(0.6*price*common shares outstanding / total liabilities) + (sales/total assets)] FIN_CHOICE = 1, if e_issue = hi and d_issue = hi FIN_CHOICE = 2, if e_issu e = hi and d_issue = lo FIN_CHOICE = 3, if e_issue = lo and d_issue = hi FIN_CHOICE = 4, if e_issue = lo and d_issue = lo E_ISSUE is hi if firms level of net equity issuance, (data108 data115)/total assets, is higher than the industry median and is lo otherwise. D_ISSUE is hi if firms change in long term debt, (long term debt lag (long term debt))/total assets, is higher than the industry median and is lo otherwise. RANK_ALL ranks of disc_all RANK_E ranks of disc_e RANK_D ranks of disc_d Q_ QTR2 = 1 if tobin_q is in the second quartile of the sample Q_QTR3 = 1 if tobin_q is in the third quartile of the sample Q_QTR4 = 1 if tobin_q is in the fourth quartile of the sample CF average cash flow of 2002 and 2003 GROWTH log (total assets of 2001 / total assets of 2000) LAG_INVEST invest for year 2001 INVEST_02 [capital expenditure + r&d expenditure + acquisitions sale of ppe ]/ total assets for year 2002 DISC_ALL_S disc_all, constructed using stanga's (1976) weightings DISC_E_S disc_e, constructed using stanga's (1976) weightings DISC_D_S disc_d, constructed using stanga's (1976) weightings
72 E_FIN level of net equity issuance, deflated by total assets, expressed as percentage points N_ANAL number of analysts following SURP the absolute value of the difference between 2000 eps and 2001 eps, deflated by price. BIG_FIVE equals 1 if the firm hires a big five auditor, and 0 otherwise INST_OWN percentage of institutional ownership AB_RET buy and hold returns for 2001, adjusted by subtracting CRSP value weighted market index CORR earnings returns correlation for the 8 quarters of 2000 and 2001 LOSS equals 1 if 2001 earnings are negative, and 0 otherwise RET_STD standard deviation of monthly returns for year 2000 and 2001. obs must be > 20 OUT_DIREC percentage of outside directors FREQ the total number of annual earnings forecasts issued by a firm in the three years ending in 2001 OCCUR the number of years (out of three years ending in 2001) in which management issued at least one annual earnings forecast
73 APPENDIX B DISCLOSURE SCORES: INDEX, CALCULATION, A ND EXAMPLES Disclosure Scores: Index Non -Financial statement disclosures that are more relevant to equity holders DISC_E Main Categories Sub Categories Scores Company Background 2 Industry Background 2 Customer Names of the Customers 1 Customers by Product Lines 1 Proportion of Sales Attributed to Major Customers 1 Products Names of the Products 1 Product Lines 1 Functions of the Products 1 Technologies Employed in the Products 1 Proportion of Sales Attributed to Products /Product Lines 1 Products in Development 1 Employees Number of Employees 1 Employees by Functions 1 Legal Proceedings Yes / No 1 No Information Provided 1 Competitors Name of the Competitors 1 Competitors by Product Lines 1 Backlog Amount of the Backlog 1 Backlogs by Product Lines 1 R&D & Patents Information on R&D 1 R&D Forecast 1 Number of Patents 1 Information on Patents 1
74 Disclosure Scores: Index Non -Financial Statement Disclosures that are more relevant to debt holders DISC_D Main Categories Sub Categories Scores Contractual Obligations 2 Term Loan Name of the Lending Institution 1 Amount of the Loan 1 Interest Rate 1 Maturity Date 1 Collateral/Covenants 1 Revolving Credit /Credit Facility Name of the Lending Institution 1 Amount of the Loan 1 Interest Rate 1 Expiration Date 1 Remaining Balance / Available Credit 1 Collateral/Covenant 1 Forward Contract/ Swap Notional Amount of the Contract 1 Terms of the Contract 1 Expiration 1 Public Debt Amount of the Issuance 1 Provision of the Issuance 1 Interest Rate 1 Maturity Date 1 Investing Activities Acquisition Activity 1 Capex Forecast 1 Information on Capex 1 The Disclosure Score s : Calculation To minimize human bias, the scoring system is based on counting the number of items disclosed. Each disclosure sub-category counts as one point, with the exception of company background, industry background, and contractual obligations. Those three main di sclosure categories do not have sub categories and thus receive higher weights. They count as two points. The final scores for DISC_ALL, DISC_E and DISC_D are calculated by dividing the raw counts by the maximum possible counts each company could have obtained. The maximum possible raw counts of DISC_E are 24, whereas the maximum possible raw counts of DISC_D vary from company to company. If a firm does not have Term Loans, Revolving Credits, Forward Contracts/Swaps, or Public Debts, the scores und er each main category is excluded when calculating the maximum possible raw count .
75 APPENDIX C DISCLOSURE SCORES: D ESCRIPTIVE STATISTICS Panel A DISC_E Main Categories Mean Sub Categories Mean Cond. Mean Company Background 95.65% Industry Background 49.76% Customer 79.71% Names of the Customers 69.08% 86.67% Customers by Product Lines 25.12% 31.52% Proportion of Sales Attributed to 65.70% 82.42% Major Customers Products 98.55% Name of the Products 84.54% 85.78% Product Lines 87.44% 88.73% Functions of the Products 95.17% 96.57% Technologies Employed in the Products 48.79% 49.51% Proportion of Sales Attributed to 36.23% 36.76% Products /Product Lines Products in Development 28.50% 28.92% Employees 90.34% Number of Employees 90.34% 100.00% Employees by Functions 50.24% 55.61% Legal Proceedings 75.36% Yes 52.17% 69.23% No 23.67% 31.41% Competitors 75.85% Names of the Competitors 75.85% 100.00% Competitors by Product Lines 48.31% 63.69% Backlog 51.69% Amount of the Backlogs 51.69% 100.00% Backlogs by Product Lines 9.18% 17.76% R&D & Patents 72.46% Information on R&D 47.34% 65.33% R&D Forecast 4.35% 6.00% Number of Patents 47.83% 66.00% Information on Patents 25.12% 34.67% The means for the main categories and sub -categories are number of firms that disclose each respective disclosure category divided by 207, the number of firms in the sample The conditional means for the sub -categories are number of firms that dis close the respective sub -category divided by number of firms that disclose the main category to which the sub category belongs.
76 Panel B DISC_D Main Categories Mean Sub Categories Mean Cond. Mean Contractual Obligations 42.51% Term Loan 18.36% Name of the Lending Institution 3.86% 21.05% Amount of the Loan 18.36% 100.00% Interest Rate 14.01% 76.32% Maturity Date 14.49% 78.95% Collateral/Covenants 10.14% 55.26% Revolving Credit / Credit Facilities 60.39% Name of the Lending Institution 13.04% 21.60% Amount of the Credit Line 59.90% 99.20% Interest Rate 36.23% 60.00% Expiration Date 42.51% 70.40% Remaining Balance / Available Credit 48.31% 80.00% Collateral/Covenant 42.51% 70.40% Forward Contract/ Swap 5.31% Notional Amount of the Contract 5.31% 100.00% Terms of the Contract 2.90% 54.55% Expiration 3.38% 63.64% Public Debt 15.94% Amount of the Issuance 15.94% 100.00% Provision of the Issuance 14.49% 90.91% Interest Rate 14.98% 93.94% Maturity Date 13.04% 81.82% Investing Activities 68.60% Acquisition Activity 51.21% 74.65% Capex Forecast 24.15% 35.21% Information on Capex 24.15% 35.21% The means for the main categories and sub -categories are number of firms that disclose each respective disclosure category divided by 207, the number of firms in the sample The conditional means for the sub -categories are number of firms that disclose the respective sub -category divided by number of firms that disclose the main category to which the subcategory belongs .
77 APPENDIX D THE DISCLOSURE SCORE S: EXAMPLES DISC_E 1 Company Background: (Abbreviated) Catalyst Semiconductor Inc. (Catalyst, we, us or Registrant) designs, develops and markets a broad range of programmable IC products s erving the micro -controller applications market. We have sought to enhance our internal design and process technology expertise through strategic relationships with leading semiconductor manufacturers and we currently subcontract the fabrication of our semiconductor wafers through Oki Electric Industry C o., Ltd. (Oki) in Japan and X Fab Texas, Inc. (Xfab) in Lubbock, Texas Our business is highly cyclical and has been subject to significant downturns at various times which have been characterized by reduced product demand, production overcapacity and si gnificant erosion of average selling prices. The market for certain Flash and EEPROM devices, which comprise the majority of our business, is currently experiencing an excess market supply relative to demand which is resulting in a significant downward tre nd in prices. Total revenues for the quarter and the fiscal year ended April 30, 2002 were $12.5 million and $42.8 million, respectively, compared to revenues of $17.2 million and $98.0 million for the comparable periods of the prior year Our nu mber of full time equivalent employees decreased to 67 in April 2002 from 71 in April 2001. The decrease was principally in the sales and administrative areas in reaction to our decreased revenues On June 30, 2001, we repaid the $2.0 million outsta nding balance owed to a bank and cancelled the related borrowing agreement. At this time, we believe that we have sufficient cash on hand and will not immediately need to enter into another borrowing agreement for the near term We market our produc ts through a direct sales force and a worldwide network of independent distributors and sales representatives. For the year ended April 30, 2002, international sales represented 69% of our product sales Catalyst Semiconductor Inc., Excerpts from Annual R eports on Form 10-ks, 2001 2 Industry Background: (Abbreviated) Growth of Digital Video Imaging Multimedia technology and its uses have grown in the past decade. A significant driver of this growth in multimedia has been the growth of video technologies. Many large industries including the movie, television, publishing and computer industries depend directly on video technologies to create and deliver their products. Traditionally all video, still image and sound products were based on analog technol ogies The Internet and Miniaturization of Electronics Fuel Demand for Video Imaging
78 Video and image capture on PCs was first used in videoconferencing applications. However, early videoconferencing applications were expensive, and suffered from poo r image quality and inadequate network infrastructure. Video conferencing grew rapidly as image quality improved and cameras became more affordable. According to a study published by the Cahners In -Stat Group in July 2001, the market for PC cameras is predicted to grow from approximately 10.5 million units in 2000 to approximately 42.5 million units by 2005. As cameras became readily available on PCs, applications other than videoconferencing quickly followed Miniaturization has moved computing fr om the desktop to a wide assortment of portable and hand held devices including laptops, personal digital assistants, electronic games and mobile phones. These battery operated devices are creating an opportunity for small, low power, low cost digital still cameras to be integrated directly into the portable device so that images can be captured for transfer to computer systems using wired or wireless methods The more recent digital still cameras use a live video image sensor to display a real time im age on a miniature, built in display which serves as a viewfinder. Still images captured by the same image sensor and stored in the camera are transferred to a computer system for viewing, editing, transmitting and printing Advances in Image Sensor Techn ology Image sensors are at the center of all electronic cameras. Image sensors capture an image through a lens and convert that image into electronic signals. Charged couple device, or CCD, technology has dominated the image sensor market for over 25 years. However, production is concentrated with relatively few, large, primarily vertically integrated manufacturers. According to the Cahners July 2001 study, the top six CCD image sensor manufacturers, Fuji Corporation, or Fuji, Matsushita Electric Indu strial, or Matsushita, Nippon Electric Corporation or NEC, Sharp Corporation, or Sharp, Sony Corporation, or Sony, and Toshiba Corporation, or Toshiba, account for approximately 97.1% of total CCD image sensor production. A newer, easier to use semico nductor technology, complementary metal oxide semiconductor, or CMOS, has been adopted for most common integrated circuits. Although CMOS technology has been available for image sensor designs for over 20 years, until recently it has not been used in commercial products because of poorer image quality. Recent improvements in CMOS, including smaller size circuits, better current control, and a more stable fabrication process, have made it possible to design CMOS image sensors that provide high image qua lity and that have many advantages over CCD image sensors. Omnivision Technologies Inc., Excerpts from Annual Report on Form 10-k, 2001 ]
79 3 Customers: 3.1 Names of the Customers or Distributors: ESD enters into arrangements with distributors to broaden d istribution channels, to increase its sales penetration to specific markets and industries and to provide certain customer services. Distributors are selected based on their access to the markets, industries and customers that are candidates for ESD produc ts. Major domestic distributors include Aurum Technology, EDS, Fiserv, Nextira One, Norstan, Siemens Business Communications, Sprint, Symitar Systems and Verizon. Major international distributors include Adamnet (Japan), IBM (Europe, Argentina), Informatio n Technology & Data (Turkey), IVRS (Hong Kong, China), Loxbit (Thailand), OLTP Voice (Venezuela), Norstan (Canada), Promotora Kranon (Mexico), Siemens AG (Worldwide), Switch (Chile), Tatung (Taiwan), and Telia Promotor (Sweden) Intervoice Inc., Excerpts from Annual Report on Form 10-k, 2001 3.2 Customers by Product Lines The Companys customers include the following, among others: Communications Serial Communications Video and Imaging ADC Telecommunications, Inc. Alcatel Alsthom S.A. Cisco Systems, Inc. Lucent Technologies, Inc. Marconi Communications Plc. NEC Nokia Corporation Siemens Corporation Tellabs, Inc. AMI Inc. Cisco Systems, Inc. Digi International, Inc. LM Ericsson T elephone Co. Rose ElectronicsTella bs, Inc. Hewlett Packard Company Logitech International S.A. Sharp Electronics Corp. Exar Inc., Excerpts from Annual Report on Form 10-k, 2001
80 3.3 Proportion of Sales Attributed to Major Customers/ Distributors Domestic Channel. In the United States, the primary distributors of our products to resellers are Ingram Micro, Inc. and Tech Data Corp. Ingram Micro accounted for 34% of our total revenue in 1999, 32% in 2000 and 27% in 2001. Tech Data began distributing our Internet security products in February 1999, and in the years ended December 31, 1999, 2000 and 2001, accounted for approximately 12%, 20% and 19%, respectively, of total revenue. Sonicwall Inc., Excerpts from Annual Report on Form 10-k, 2001 4 Products: 4.1 Names of the Products (Abbreviated) GigaMux'r' DWDM Optical Transport Our GigaMux'r' optical transport product utilizes EPC'TM' Sub Rate Access Multiplexer Electric Photonic Concentrator, JumpStart'TM' CWDM Transport JumpStart is our entry level solution TeraManager'TM' Element Management System TeraManager'TM' is our TL1 based intelligent element management software platform TeraMatrix'TM' Op tical Switching We have completed development of a fully Sorrento Networks Corp., Excerpts from Annual Report on Form 10-k, 2001 4.2 Product Lines (Abbreviated) ANALOG AND MIXED SIGNAL PRODUCTS We design, manufacture and market high -per formance analog and mixed signal integrated circuits for computing, consumer, communications and industrial applications DISCRETE PRODUCTS We design, manufacture and market discrete devices for computing, automotive, communications and industrial applications. Discrete devices are individual diodes or transistors that perform basic signal amplification and switching functions in electronic circuits INTERFACE AND LOGIC PRODUCTS We design, develop, manufacture and market high performance interface and logic devices utilizing three wafer fabrication processes: CMOS, BiCMOS and Bipolar OTHER PRODUCTS Included within the "Other" reporting segment are non-volatile memory
81 products and optoelectronic products. Non -Volatile Memory Products We design, manufacture and market non -volatile memory integrated circuits, which are storage devices that retain data after power to the device has been shut off. We offer an extensi ve portfolio of high performance serial EEPROM and EPROM products Optoelectronic Products Optoelectronics covers a wide range of semiconductor devices that emit, sense and display both visible and infrared light Fairchild Semiconductor INTL., Exc erpts from Annual Report on Form 10-k, 2001 4.3 Functions of the Products Automotive Electronics The applications span not only traditional AM/FM radio, but also the emerging class of entertainment, wireless and telematics applicat ions that are adding value to mobile consumers, including digital satellite radio, in -car or in -flight video, Internet access, route guidance and navigation, emergency assistance, 'hands -free' mobile phone operation and other wireless systems including Bluetooth based appliances, wireless dashboard interfaces, and keyless entry. Microtune Inc., Excerpts from Annual Report on Form 10-k, 2001 4.4 Technologies Employed in the Products S EMICONDUCTOR ARCHITECTURE -Netergy's semiconductors are based on programmable processor architectures that enable implementation of IP telephony and videoconferencing applications in a highly efficient manner. Netergy's semiconductor architectures employ 32-bit reduced instruction set computer, or RISC, microprocessor cores, which execute the embedded applications software. Some of Netergy's semiconductors also employ a 64 -bit Single Instruction Multiple Data, or SIMD, digital signal processor, or DSP, to accelerate the execution of signal processing intensive operations. Furthermore, Netergy's Audacity T2 and T2U semiconductors benefit from the unique feature of not requiring any external SRAM or DRAM to operate. Netergy's RISC processor cores use a prop rietary instruction set specifically designed for multimedia communication applications. The RISC cores control the overall chip operation and manage the input/output interface through a variety of specialized ports that connect the chip directly to external host, audio, and network subsystems. The cores are programmable in the C programming language and allow customers to add their own features and functionality to the device software provided by Netergy. Netergy's DSP core architecture is a SIMD processor that implements computationally intensive video, audio, and graphics processing routines as well as certain digital communication protocols. The DSP cores are programmable with a proprietary
82 instruction set consisting of variable -length 32 -bit and 64 bit microcode instructions that provide the flexibility to improve algorithm performance, enhance audio and video quality, and maintain compliance with changing digital audio, video, graphics, and communication protocol standards. The DSP cores access their in structions through an internal bus that interfaces to on -chip SRAM and read -only memory, or ROM, that is pre -programmed with video and audio processing subroutines. Netergy Microelectronics, Inc, a subsidiary of 8x8 Inc., Excerpts from Annual Report on F orm 10-k, 2001 4.5 Proportion of Sales Attributed to Products /Product Lines Since 1992, we have focused on developing a leadership portfolio of CPLD products and increasing the percentage of our overall revenue derived from this attractive market. Dur ing 2001, approximately 76% of our revenue was derived from CPLD products, as compared to 66% in calendar 1999 and essentially zero in 1992. Lattice Semiconductor Corp., Excerpts from Annual Report on Form 10-k, 2001 4.6 Products in Development We ar e currently developing the Compact Node Type 90071, which is a dual high level output node that allows operators to considerably reduce the number of amplifiers needed per node, and the Compact Fiber Deeper Node Type 90090, which is designed to eliminate t he need for amplifiers after the node and to help operators plan for their future network speed needs. Based on our belief at the time, we previously announced that Compact Distribution Node Type 90071 and Compact Fiber Deeper Node Type 90090 would be avai lable during the spring 2001. Although the 90071 node shipped in June 2001, it is expected that the 90090 node will not ship before October 2001. In addition, based on our belief at the time, we previously announced that the new Surge -Gap(TM) Power Distribution Unit product and the new Surge Gap Tap product were scheduled for availability in June 2001 and July 2001, respectively. It is now expected that they will become available in August 2001. Scientific -Atlanta Inc., Excerpts from Annual Report on Form 10-k, 2001 5 Employees 5.1 Number of Employees As of December 31, 2001, we employed 487 individuals on a full time basis, all but thirty -three of whom reside in the United States. Silicon Storage Technology, Excerpts from Annual Report on Form 10 -k, 2001 5.2 Employees by Functions
83 Of these 487 employees, 90 were employed in manufacturing support, 263 in engineering, 74 in sales and marketing and 60 in administration and finance. Silicon Storage Technology, Excerpts from Annual Report on Form 10-k, 2001 6 Legal Proceedings 6.1 Yes IPO Class Action Lawsuit On August 6, 2001, a putative securities class action, captioned Beveridge v. Avanex Corporation et al., Civil Action No. 01 CV 7256, was filed against us, certain company officers and dir ectors (the "individual defendants"), and four underwriters in our initial public offering ("IPO"), in the United States District Court for the Southern District of New York. The complaint alleges violations of Section 11 of the Securities Act of 1933 ("Securities Act") against all defendants, a violation of Section 15 of the Securities Act and Section 20(a) of the Securities Exchange Act of 1934 ("Exchange Act") against the individual defendants, a violation of Section 10(b) of the Exchange Act (and Ru le 10b5, promulgated thereunder) against us and the individual defendants, and violations of Section 12(2) of the Securities Act and Section 10(b) of the Exchange Act (and Rules 10b3 and 10b5, promulgated thereunder) against the underwriters. The compla int seeks unspecified damages on behalf of a purported class of purchasers of common stock between February 3, 2000 and December 6, 2000 (the "class period"). Avenex Corp., Excerpts from Annual Report on Form 10-k, 2001 6.2 No Item 3. Legal Proceedings None C-Cor Inc., Excerpts from Annual Report on Form 10-k, 2001 6.3 Yes, but no Information provided The Company is a party to various legal proceedings and administrative actions, all of which are of an ordinary or routine nature incidental to the o perations of the Company. Although it is difficult to predict the outcome of any legal proceeding, in the opinion of the Company's management, such proceedings and actions should not, individually or in the aggregate, have a material adverse effect on the Company's financial condition, results of operations or cash flows. AVX Corp., Excerpts from Annual Report on Form 10-k, 2001
84 7 Competitors 7.1 Names of the Competitors The market for wireless tracking and management of enterprise assets is relatively n ew, constantly evolving, and competitive. Although the Companys current competitors do not provide the exact same capabilities, they do offer subsets of the Companys system capabilities or alternate approaches to the issues the Company's products resolv e. Those companies include both emerging companies with limited operating histories, such as WhereNet Corp., Remote Equipment Systems, Inc., Media Recovery, Inc., and companies with longer operating histories, greater name recognition and/or significantly greater financial, technical and marketing resources than the Company, such as Savi Technology, Symbol Technologies, Inc., and Intermec Technology Corp. The Company expects that competition will intensify in the near future. There a re, however, significant barriers to entry for the Company's potential competitors. I D Systems Inc., Excerpts from Annual Report on Form 10-k, 2001 7.2 Competitors by Product Lines As of December 31, 2000, we had more than 55% market share in terms of AMR meter modules shipped. Our largest competitor in the AMR meter module market is Schlumberger's Resource Management Services division, which acquired a former competitor, CellNet Data Systems in March 2000. In addition to being a competitor, Schlumberger also acts as a reseller and integrator of our solutions. There are also a few new communications providers, radio-based, Internet based, and public network based, most of whom are narrowly focused, that have recently been awarded pilot systems at utilit ies, including companies such as Nexus and Innovatec. These companies currently offer alternative solutions and compete aggressively with us. In our EIS market, there are many market participants that may be both competitors and potential partners. W e face competition from a number of companies such as ABB, Siemens, Lodestar, ICF Kaiser, and Accenture. In competitive wholesale markets in California and Ontario, Canada, we have partnered with ABB and Ernst & Young to offer a total integrated system sol ution. We will continue to partner with some of these companies to address future competitive energy markets. Itron Inc., Excerpts from Annual Report on Form 10-k, 2001 8 Backlog 8.1 Amount of the Backlogs Our backlog was approximately $10.2 million and $5.2 million as of the first business day of June, 2002 and 2001, respectively. Ditech Networks Inc., Excerpts from Annual Report on Form 10-k, 2001 8.1 Backlogs by Product Lines (Abbreviated) EOSG booked $264.2 million in fiscal 2002, including c ontracts valued at approximately $122.1 million from the U.S. Army to provide Second Generation
85 Thermal Imaging Acquisition and Targeting Systems used on its M2 Bradley Fighting Vehicles, M1 Abrams Battle Tanks and Long Range and HMMWV Scouts ESG sec ured $194.9 million in new contracts, including significant awards of $106.7 million for production and engineering of the AN/UYQ 70 FSCG received a total of $109.2 million in new awards in fiscal 2002, including $43.6 million in contracts for communications and surveillance systems, $37.1 million in awards for high-speed cameras and flight and mission data recorders, and $28.5 million for advanced electronic manufacturing services for major aerospace prime contractors. DRS Ahead Technology, included in the operations of Other, booked $8.9 million in new orders for magnetic burnish, glide and test verification heads used in the manufacture of computer disk drives. DRS Technologies Inc., Excerpts from Annual Report on Form 10-k, 2001 9 R&D & Patents 9.1 Information on R&D Current research and development in the mobile communications area is focused on key components, radio frequency subsystems and cellular systems for emerging CDMA2000, GPRS and EDGE applications as w ell as 3G WCDMA systems that provide greater bandwidth and will enable new possibilities for accessing the Internet using mobile communications platforms. Conexant Systems Inc., Excerpts from Annual Report on Form 10-k, 2001 9.2 R&D Forecast Our curren t outlook forecasts a decline in R&D spending of roughly $1.3 million dollars in 2002. KVH Industries Inc., Excerpts from Annual Report on Form 10-k, 2001 9.3 Number of Patents The Company currently holds 30 United States patents and 14 foreign patents covering a wide range of electronic systems and circuits, of which 19 United States patents and 10 foreign patents were obtained in the Company's acquisition of Scientific -Atlanta, Inc.'s interdiction business during1998. Blonder Tongue Labs Inc., Excer pts from Annual Report on Form 10-k, 2001 9.4 Information on Patents (Abbreviated) One patent, expiring in March 2011, relates to the Criterion series of survey lasers providing coverage of forestry applications that include height and diameter measureme nt of trees
86 In fiscal 1996, a fifth patent was issued, expiring in June 2013, relating to our Survey and Mapping product line, which incorporates our proprietary Walkabout software that enables field data collection in the G.I.S. mapping process During fiscal 1997, two patents were issued on our proprietary technology related to the consumer range -finding instrumentation developed for Bushnell During fiscal 1998, one patent was issued, expiring in December 2014, which was a continuation on an existing patent related to our consumer range -finder... Laser Technology Inc., Excerpts from Annual Report on Form 10-k, 2001 DISC_D 1 Contractual Obligations Total outstanding commitments at December 31, 2001 were as follows: (In thousands) Total Less than 1 year 1 to 3 years 4 to 5 years After five years Long term debt $ 15,378 $ $ $ 15,378 $ Operating leases 900 133 128 98 541 Purchase obligations 23,900 23,900 Letters of credit 6,760 6,760 Total $ 46,938 $ 30,793 $ 128 $ 15,476 $ 541 Cobra electronics Corp., Excerpts from Annual Report on Form 10-k, 2001 2 Term Loan 2.1 Name of the Lending Institution 2.2 Amount of the Loan 2.3 Interest Rate 2.4 Maturity Date 2.5 Collateral/Covenants In August 2000, the Registrant secured a $795,000 (2.2) mortgage loan with EAB (2.1), secured by the recently purchased facility at 11 13 Stepar Place, Huntington Station,(2.5) New York. The loan is now with Citibank, as successor to EAB. The term of the loan is 10 years to be repaid in 120 equal installments. (2.4) the mortgage is subject to certai n financial covenants, including maintenance of asset and liability percentage ratios. The mortgage loan bears interest at 1 1/2% above the six month LIBOR rate. (2.3) American technical Ceramics., Excerpts from Annual Report on Form 10-k, 2001
87 3 Revolvi ng Credit / Credit Facility 3.1 Name of the Lending Institution 3.2 Amount of the Loan 3.3 Interest Rate 3.4 Expiration Date 3.5 Remaining Balance / Available Credit 3.6 Collateral/Covenant Mobitec has an agreement with a bank in Sweden (3.1) from which we may currently borrow up to a maximum of 10,000,000 krona (SEK) or $945,984. (3.2) At December 31, 2001, $760,092 was outstanding, (3.5) resulting in additional borrowing availability to us of $185,892. 3,000,000 krona (SEK) of the 10,000,000 krona (SEK) maximum borrowing is secured by cash on deposit with the bank. The terms of this agreement require our payment of an unused credit line fee equal to 0.5 percent of the unused portion and interest at 5 percent of the outstanding balance. (3.3) This agreem ent is secured by substantially all assets of Mobitec AB. (3.6) Our line of credit agreement expires on December 31, 2002 (3.4) and is renewable at that date for another year. Mobitec Holding AB. A subsidiary of Digital Recorders Inc., Excerpts from Annual Report on Form 10 -k, 2001 4 Forward Contract/ Swap 4.1 Notional Amount of the Contract 4.2 Terms of the Contract 4.3 Expiration The Company has entered into an interest rate swap agreement with a major U.S. bank which expires November 30, 2004, (4.3) to hedge its exposure to variability in expected future cash flows resulting from interest rate risk related to 25% of its long-term debt. The interest rate under the swap agreement is fixed at 6.8% (4.2) and is based on the notional amount, which wa s $7.5 million at December 31, 2001. (4.1) The swap contract is inversely correlated to the related hedged longterm debt and is therefore considered an effective cash flow hedge of the underlying long -term debt. The level of effectiveness of the hedge is measured by the changes in the market value of the hedged long -term debt resulting from fluctuation in interest rates. During fiscal 2001, variable interest rates decreased resulting in an interest rate swap liability of $147,000 as of December 31, 2001. A s a matter of policy, the Company does not enter into derivative transactions for trading or speculative purposes. Diodes Inc., Excerpts from Annual Report on Form 10-k, 2001 5 Public Debt 5.1 Amount of the Issuance 5.2 Provision of the Issuance 5.3 Inter est Rate 5.4 Maturity Date
88 On July 13, 2000, the Company sold $550 million (5.1) principal amount of 41/4% Convertible Subordinated Notes due 2007. (5.4) The interest rate is 41/4% per annum on the principal amount, payable semi annually in arrears in ca sh on January 15 and July 15 of each year, beginning January 15, 2001. (5.3) The notes are convertible into shares of the Company's common stock at any time on or before July 15, 2007, at a conversion price of $73.935 per share, subject to adjustment if ce rtain events affecting the Company's common stock occur. (5.2) The notes are subordinated to all of the Company's existing and future senior indebtedness and to all debt and other liabilities of the Company's subsidiaries. The Company may redeem any of the notes, in whole or in part, on or after July 18, 2003, as specified in the notes and related indenture agreement. 6 Investment Activity 6.1 Acquisition Activity During 2001, the Company spent approximately $91 million for several small acquisitions and in vestments. The acquisitions were accounted for as purchases and, accordingly, the results of operations of each acquired company are included in the consolidated income statement from the date of acquisition ITT Corp., Excerpts from Annual Report on For m 10 -k, 2001 6.2 Capex Forecast We expect to spend approximately $6.0 million to purchase capital equipment during the next twelve months, principally for the purchase of design and engineering tools, additional test equipment and computer software and h ardware. Integrated Silicon Solution, Excerpts from Annual Report on Form 10-k, 2001 6.3 Information on Capex The fiscal 2001 expenditures were primarily related to additional FSA attachment equipment and the Minnesota and Thailand facility additions related to the move of operations from Arizona to Minnesota and Thailand. The capital expenditures in both fiscal 2000 and 1999 include additional equipment to increase the capacity of the automated flexible circuit production facility in Litchfield, Minne sota and the costs to construct and equip a material manufacturing facility. Innovex Inc., Excerpts from Annual Report on Form 10-k, 2001
89 LIST OF REFERENCES Akerlof, G 1980. The market for lemons : Qualitative uncertainty and the market mechanism Quart erly Journal of Economics 84: 488 500. Almeida, H., M. Campello, and M. S. Weisbach. 2004. The cash flow sensitivity of cash. Journal of Finance 59: 1777 1804. Altman, E.I. 1968. Financial ratios, dis criminant analysis, and the prediction of corporate bankruptcy. Journal of Finance 23: 589 609. Alti, A. 2003. How sensitive is investment to cash flow when financing is frictionless? Journal of Finance 58: 707 722. Ajinkya, B. B., and M. J. Gift. 1984. C orporate managers' earnings forecasts and symmetrical adjustments of market expectations. Journal of Accounting Research 22: 425 444. Ajinkya, B., S. Bhojraj, and P. Sengupta. 2005. The association between outside directors, institutional investors and the properties of management earnings forecasts. Journal of Accounting Research 43: 343 376. American Institute of Certified Public Accountant (AICPA). 1994. Improving Business Reporting -A Customer Focus. New York. AICPA. Amir, E., and B. Lev. 1996. Value rel evance of nonfinancial information: The wireless communications industry. Journal of Accounting and Economics 22: 3 30 Baginski S. P., and J M. Hassell 1990. The Market Interpretation of Management Earnings Forecasts as a Predictor of Subsequent Financ ial Analyst Forecast Revision. The Accounting Review 65: 175 190 Baker, M., and J. Wurgler. Market timing and capital structure. Journal of Finance 57: 1 32. Barry, C. B., and S. J. Brown. 1985. Differential Information and Security Market Equilibrium. The Journal of Financial and Quantitative Analysi s 20: 407 422. Bens, D. A., and S. Monahan. 2004. Disclosure quality and the excess value of diversification. Journal of Accounting and Economics 42: 691 730 Bergstresser, D. 2006. Discussion of Overinvest ment of free cash flow. Review of Accounting Studies 11: 191 202. Biddle, G. C., and G. Hilary. 2006. Accounting quality and firm level capital investment. The Accounting Review 81: 963 982. Biddle, G. C., G. Hilary, and R. S. Verdi. 2008. How does financ ial reporting quality improve investment efficiency? Working Paper.
90 Botosan, C. 1997. Disclosure level and the cost of equity capital. The Accounting Review 72: 323 49. Botosan, C., and M. A. Plumlee. 2000. A re examination of disclosure level and the exp ected cost of equity capital. Journal of Accounting Research 40: 21 40 Bound, J., D. A. Jaeger, and R. M. Baker. 1995. Problems with Instrumental Variables Estimation When the Correlation between the Instruments and the Endogenous Explanatory Variable Is Weak. Journal of the American Statistical Association 90: 443 450. Bushee, B., D. Matsumoto, and G. Miller. 2003. Open versus Closed Conference Calls: The Determinants and Effects of Broadening Access to Disclosure. Journal of Accounting and Economics 34: 149 180. Bushman, R. M., J. D. Piotroski, and A. J. Smith. 2006. Capital allocation and timely accounting recognition of economic losses. Working Paper. Cleary, S. 1999. The relationship between firm investment and financial status, Journal of Finance 54: 673 692. Cole, C. J., and C. L. Jones. 2004. The usefulness of MD&A disclosures in the retail industry. Journal of Accounting, Auditing and Finance 19: 361 388. Cole, C. J., and C. L. Jones. 2005. Management discussion and analysis: A review and implications for further research. Journal of Accounting Literature 24: 135 174. Coller, M. and T. L. Yohn. 1997. Management forecasts and information asymmetry: An exami nation of bid ask spreads. Journal of Accounting Research 35 : 181 191. Cooper, R., J. Haltiwanger, and L. Power. 1999. Machine replacement and the business cycle: Lumps and bumps, American Economic Review 89: 921 946. Diamond, D., and R. Verrecchia. 1991 Disclosure, liquidity, and the cost of capital. The Journal of Finance 66: 1325 1355. Doms, M., and T. Dunne. 1998. Capital adjustment patterns in manufacturing plants. Review of Economic Dynamics 1: 409 429. Dye, R. 2001. An evaluation of Essays on Di sclosure and the disclosure literature in accounting. Journal of Accounting and Economics 32: 181 235. Erickson, T., and T. M. Whited. 2000. Measurement error and the relationship between investment and Q. Journal of Political Economy 108: 1027 1057. Fa ma, E. F., and K. R. French. 1997. Industry costs of capital. Journal of Financial Economics 43: 153 193.
91 Fazzari, S.M., R. G. Hubbard, and B. C. Petersen. 1988. Financing constraints and corporate investment. Brookings Papers on Economic Activity 1: 141 206 Francis, J., and K. Schipper. 1999. Have financial statements lost their relevance? Journal of Accounting Research 37: 319 352. Francis, J., K. Schipper, and L. Vincent. 2003. The relative and incremental explanatory power of earnings and alternative ( to earnings) performance measures for returns. Contemporary Accounting Research 20: 121 164. Frankel, R., M. Johnson, and D. Skinner. 1999. An Empirical Examination of Conference Calls as a Voluntary Disclosure Medium. Journal of Accounting Research 37, 133 150. Frankel, R., M. McNichols, and G. P. Wilson. 1995. Discretionary disclosure and external financing. The Accounting Review 70: 135 150. Gertler, M., and S. Gilchrist. 1994. Monetary policy, business cycles, and the behavior of small manufacturing fir ms. The Quarterly Journal of Economics 109: 309 340. Greenwald, B. C., J. E. Stiglitz, and A. Weiss. 1984. Informational imperfections in the capital markets and macro-economic fluctuations. American Economic Review 74: 194 99. Hail, L. 2002. The impact of voluntary corporate disclosures on the exante cost of capital for Swiss firms. European Accounting Review 11: 741 773. Hall, B. 2002. The financing of research and development. Oxford Review of Economic Policy 18: 35 51. Heal y, P., A. Hutton, and K. Palepu. 1999. Stock Performance and Intermediation Changes Surrounding Sustained Increases in Disclosure. Contemporary Accounting Research 16: 485 520. Healy, P., and K. Palepu. 2001. A review of the empirical disclosure literature Journal of Accounting and Economics 31: 405 440. Heaton, J. B. 2002. Managerial optimism and corporate finance. Financial Management 31: 33 45. Hope, O., and W. B. Thomas. 2008. Managerial empire building and firm disclosure. Journal of Accounting Resear ch 46: 591 626 Hoshi, T., A. Kashyap, and D. Scharfstein. 1991. Corporate Structure, Liquidity, and Investment: Evidence from Japanese Industrial Groups. Quarterly Journal of Business and Economics 106: 33 60. Hubbard, R. 1998. Capital -market imperfections and investment. Journal of Economic Literature 36: 193 225.
92 Hutton, A. P., and P. C. Stocken. 2007. Effect of Reputation on the Credibility of Management Forecasts. Working Paper Jensen, M. C. 1986. Agency costs of free cash flow, corporate finance, and fakeovers. The American Economic Review 76: 323 329. Kaplan, S. N., and L. Zingales. 1997. Do investment -cash flow sensitivities provide useful measures of financing constraints? Quarterly Journal of Economics 112: 169 216. Kanodia, C. 2007. Accounting disclosure and real effects. Now Publishers Inc Kim, O., and R. Verrecchia. 1994. Market liquidity and volume around earnings announcements. Journal of Accounting & Economics 17: 41 68. Kiyotaki, N., and J. Moore. 1997. Credit cycles. Journal of Political Economy 105: 211 248. Lang, M., and R. Lundholm. 1993. Cross -sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research 31: 246 71. Lang, M ., and R. Lundholm. 1996. Corporate Disclosure Policy and Analyst Behavior. The Accounting Review 71: 467 92. Lang, M., and R. Lundholm. 2000. Voluntary disclosure and equity offerings: Reducing information asymmetry or hyping the stock? Contemporary Accounting Research 17: 624 62. Leone, A. J., S. Rock., and W. Willenborg. Disclosure of Intended Use of Proceeds and Underpricing in Initial Public Offerings. The Accounting Review 45: 111 153. Leuz, C., and P. Wysocki, 2008. Economic consequences of financia l reporting and disclosure regulation: A review and suggestions for future research. Working Paper. McNichols, M. F., and S.R. Stubben. 2008. Does earnings management affect firms investment decisions? The Accounting Review Forthcoming McNichols, M. F., and B. Trueman. 1994. Public disclosure, private information collection and short term trading. Journal of Accounting and Economics 17: 69 94. Modigliani, F., and M. H. Miller. 1958. Cost of capital, corporation finance, and the theory of investment. The American Economic Review 48: 261 297. Murphy, K. J. 1985. Corporate performance and managerial remuneration: An empirical analysis. Journal of Accounting and Economics 7: 11 42. Myers, S. C. 1977. Determinants of corporate borrowing. Journal of Financial E conomics 5: 147 175. Myers, S. C., and N. S. Majluf. 1984. Corporate financing and investment decisions when firms have information that investors do not. Journal of Financial Economics 13: 187 221.
93 Nagar, V., D. Nanda, and P. Wysocki, 2003. Discretionary Disclosure and Stock Based Incentives. Journal of Accounting and Economics 34: 283 309. Oliner, S. D., and G. D. Rudebusch. 1992. Sources of the Financing Hierarchy for Business Investment. The Review of Economics and Statistics 74: 643 654. Pownall, G., C. Wasley, and G. Waymire. 1993. The Stock Price Effects of Alternative Types of Management Earnings Forecasts. The Accounting Review 68: 896 912. Rauh, J. D. 2006. Investment and financing constraints: Evidence from the funding of corporate pension plan s. Journal of Finance 61: 33 71. Richardson, S. 2006. Over investment of free cash flow. Review of Accounting Studies 11: 159 189. Sengupta, P. 1998. Corporate disclosure quality and the cost of debt. Accounting Review 73: 459 74. Stanga, K.G. 1976. Disclosure in published annual reports. Financial Management 5: 42 52. Stiglitz, J. E., and A. Weiss, 1981. Credit rationing in markets with incomplete information. American Economic Review 71: 393 409. Tirole, J. 2006. The Theory of Corporate Finance. Princeton University Press Tobin, J. 1969. A general equilibrium approach to monetary theory. Journal of Money, Credit and Banking 1: 15 29. Verrecchia, R. 2001. Essays in disclosure. Journal of Accounting & Economics 32: 97 180. Welker, M. 1995. Disclosure policy, information asymmetry and liquidity in equity markets. Contemporary Accounting Research 11: 801 828. Whited, T. M. 1992. Debt, Liquidity Constraints, and Corporate Investment: Evidence from Panel Data. Journal of Finance 47: 1425 1460. White, H. 1980. A heteroskedasticity-consistent covariance matrix and a direct test for heteroskedasticity. Econometrica 48: 817 38.
94 BIOGRAPHICAL SKETCH Hung Yuan (Richard) Lu was born in Taipei, Taiwan. He completed his bachelo rs degree in applied mathematics at the National Chung Hsing University in 1998. After serving in the army for two years, he came to the United States to pursue a masters degree in accounting at California State University, Fullerton. After finished the foundation courses, he transferred and completed his masters degree in accounting at the Ohio State University. He joined the PhD program at the University of Florida in 2004 and completed his Ph.D. in August 2009. He will start his academic career as an associate professor at California State University, Fullerton.