Accounting for Generated vs. Contributed Change in the Enterprise Value Peter Easton Center for Accounting Research and Education University of Notre Dame Peter Vassallo Australian School of Management University of New South Wales Eric Weisbrod School of Business Administration University of Miami March 2015 We thank Bill Beaver, Demetris Christodoulou, Mincherng Deng, Joseph Gerakos, Aloke Ghosh, Adel Ibrahim, Steve Lilien, Steve Monahan, Sundaresh Ramnath, Theodore Sougiannis, Oktay Urcan, Terry Walter, Bill Wright and seminar participants at Baruch College, the University of Illinois at Urbana-Champaign, the University of Miami and the University of Sydney for helpful comments on an earlier draft. Abstract We demonstrate, both analytically and empirically, that the manner and degree to which the accounting system records change in enterprise value depends upon both the sign (positive vs. negative) and source (generated vs. contributed) of the value change. Absent accounting conservatism, the accounting measure, enterprise profit after tax, equals the cum-distribution change in enterprise market value generated during the period (i.e., generated value change). Changes in enterprise value can also occur through contributions from/distributions to the owners of the firm (i.e., contributed value change). We show that accounting conservatism differentially records these two sources of value change, leading to differences between accounting and market-based measures of enterprise performance. Our empirical approach identifies both conditional and unconditional conservatism in the same regression framework. This allows us to identify empirical manifestations of accounting conservatism that are quite different from those in the extant literature. Keywords: Accounting Conservatism, Firm Growth, Free Cash Flows, Earnings Return Relation 2 1. Introduction The focus of our paper is on the recording of changes in the market value of a corporate enterprise in the financial statements of the corporation. We use the word “enterprise” to identify the entity owned by the equity and debt holders, which engages in the firm’s primary revenuegenerating activities.1 From the perspective of the owners of the enterprise, a key measure of investment performance, or investment growth, is the change in market value generated by the enterprise before taking into account contributions from/distributions to owners during the period. We refer to this measure as EVg, Enterprise Value Generated. By comparison, the accountant’s corresponding cum-distribution measure of the performance of the enterprise is enterprise profit after tax (EPAT). Prior literature suggests that, due to accounting conservatism, negative EVg (i.e., a decrease in enterprise value) is more likely to be recorded in EPAT than positive EVg (i.e., an increase in enterprise value).2 A key premise of our paper is that accounting conservatism may also affect the way EPAT treats contributions from/distributions to the owners of the firm (i.e., enterprise cash flows, ECF).3 Because of this, the relation between EPAT and 1 We follow the terminology of Gode and Ohlson [2013] to distinguish between enterprise activities and financing activities. Enterprise activities are sometimes referred to as operating activities in practice. See Appendix A.1: Terminology for a quick reference of common terms employed in the paper. Enterprise/operating activities include activities such as research, development, purchasing inputs, production, marketing, and distribution of products and services (e.g., coffee in the case of Starbucks, passenger miles in the case of United Airlines, and brand-name household consumer products in the case of Procter and Gamble). 2 This argument is usually attributed to Basu [1997], who focused on the relation between earnings and cumdividend change in equity. We focus on the relation between enterprise profit and cum-dividend change in the value of the enterprise because we argue and show that most conservatism occurs at the enterprise/firm level rather than at the equity level. 3 The variable, which we label enterprise cash flow, ECF, is often referred to as free cash flow. We, like Gode and Ohlson [2013], use the term enterprise cash flow because we feel it is more descriptive and more encompassing; e.g., it includes contributions from the owners of an acquired company where the acquisition currency is shares in the newly-formed enterprise (e.g., the 2006 acquisition of Gillette by Procter and Gamble). Several textbooks on financial statement analysis and valuation calculate free cash flow using precisely the same method as we follow in this paper for calculation of ECF; see, for example, Easton et al. [2015], Penman [2012], Wahlen et al. [2015], and Gode and Ohlson [2013]. Our calculation of this variable amounts to the same calculation as seen in other texts such as Damodoran [2012] and White et al. [2003], where earnings after taxes are adjusted for accruals and for capital expenditure. To see the equivalence, note that changes in accruals are in both EPAT and change in book value of net enterprise assets (ΔNEA) and capital expenditure is part of ΔNEA; the calculation, ECF = EPAT - ΔNEA the change in market value of the enterprise (ΔEV) depends not only on whether enterprise performance is positive or negative, but also on changes in enterprise value due to transactions with owners during the period. We demonstrate, both analytically and empirically, that the manner and degree to which the accounting system records ΔEV in EPAT and ΔNEA (change in the book value of net enterprise assets) depends upon both the sign (positive vs. negative) and source (EVg vs. ECF) of value change. In principle, if all value generated by the enterprise during the period was realizable and verifiable, and enterprise transactions with owners of the firm were treated similarly by investors and accountants, EPAT would be equal to EVg, and ECF would provide no incremental explanatory power for EPAT.4 In practice, however, markets often incorporate unverifiable changes in unrealized enterprise value, and accountants often expense value-enhancing investments from owners, which are instead capitalized by investors.5 Our empirical analyses are based on a regression of the accountants’ measure of change in enterprise value (EPAT) on the market measure of value generation/loss (EVg) and a measure of net contributions from/distributions to the owners of the enterprise (ECF), with each of these variables deflated by beginning-of-year enterprise value (EV). We show that estimates of the coefficients relating EPAT to EVg and EPAT to ECF vary in predictable ways according to the signs of ΔEV (i.e., increase vs. decrease in enterprise value), EVg (i.e., internally generated removes the accruals from EPAT and the remainder is “free cash flow.” In our empirical analyses, EPAT and ΔNEA are calculated following the appendix to Nissim and Penman [2001], who refer to these variables as operating income and net operating assets. We use the terms enterprise profit after tax and net enterprise assets to underscore the focus on the entity (the enterprise), which is owned by the equity and debt holders. 4 This effectively implies that the accounting would be mark-to-market for all enterprise assets (including such things as human capital, asset synergies, etc.). 5 EPAT may also differ from EVg in the treatment of distributions made to owners. For example, if enterprise assets are liquidated to provide for distributions to owners, ECF will provide incremental explanatory power for EPAT when the book value of liquidated assets differs from fair market value (via the gain or loss on sale recorded in EPAT). 2 growth vs. contraction), and ECF (i.e., cash provided by vs. cash distributed to the owners of the enterprise). We also demonstrate that ECF provides significant incremental explanatory power for EPAT beyond EVg, and that the mapping from EVg to EPAT depends on the sign of ECF. Our results show that, in most cases, the association between ECF and EPAT is as large as, and sometimes larger than, the association between EVg and EPAT. We find that, on average, a one standard deviation change in ECF (EVg) is associated with approximately a 0.27 (0.24) standard deviation change in EPAT. In other words, we show that accounting measures of enterprise performance are often more strongly associated with transactions between the enterprise and its owners during the period than with value generated by the enterprise during the period. To understand this result, we note that financial statements capture the entire amount of the net contributions from/distributions to the owners of the firm (ECF), in either the income statement (i.e., EPAT) or as change in the book value of net enterprise assets (i.e., ΔNEA), in the fiscal period in which the contribution/distribution is made. Value generation/loss (EVg) is, however, captured in EPAT in the fiscal period in which the value is generated/lost only to the extent that the expected change in value is realized during the period; the remaining value generation/loss, which is captured in change in the market value of the enterprise, reflects changes in expectations about future profitability. In any instance where the financial statement treatment (e.g., capitalization vs. expensing) of ECF differs from the market valuation of ECF, EPAT will differ from EVg, and the extent of the difference will depend on the nature of ECF. For example, suppose that R&D is highly successful and results in increased sales in the current year and several future years; enterprise 3 value will increase by an amount equal to the present value of the expected net benefits from the R&D in the period in which the R&D expenditure occurs. In contrast, under U.S. GAAP, the full cost of the R&D will be expensed in EPAT in the period in which the R&D expenditure occurs, while only the portion of the increased sales from R&D that accrue in the current year will be captured in EPAT of the current period. The deferral of the future benefits from R&D will decrease the coefficient on EVg in our model, while the accelerated expensing (relative to market) of the ECF invested in R&D will increase the coefficient on ECF in our model. Indeed, in our empirical analyses, we interact EVg and ECF with multiple measures of asset tangibility, including R&D and advertising intensity, and consistently find that the coefficients on EVg (ECF) decrease (increase) with measures of the enterprises’ investment in intangibles. Our empirical analyses lead to several key conclusions: (1) conditional and unconditional accounting conservatism (as defined in the extant accounting literature) interact to determine the mapping from EVg to EPAT;6 (2) consistent with the notion of conditional conservatism, only the portion of EVg that relates to creation of value within the fiscal period is reflected in EPAT of the fiscal period -- a greater portion being captured in current EPAT when EVg is negative (i.e., value is lost); (3) consistent with the notion of unconditional conservatism, ECF is completely captured in the accounting record (in either ΔNEA and/or EPAT) in the period in which the contribution is made, but the accounting treatment (i.e., the portion of ECF recorded in ΔNEA vs. EPAT) depends on the use of contributed funds;7 and, (4) the mapping from EVg and ECF to ΔNEA and EPAT varies with a number of enterprise characteristics including; a) intangibles 6 We define and elaborate on the terms conditional and unconditional conservatism in section 2. If the ECF expenditure is in a non-zero NPV activity, this NPV will be captured in EVg but not in EPAT of the period (unless the non-zero NPV effect affects EPAT of the current period), contributing to a coefficient relating EPAT to EVg that is less than one. This is the form of conservatism modelled by Feltham and Ohlson [1995] and described empirically by Easton and Pae [2004]. 7 4 intensity, b) capital intensity, c) the amount of debt, and d) the ratio of the book value of the enterprise to the market value of the enterprise. Our paper is related to the extant literature falling under the oft-used label, “conservative accounting”, which has followed two paths: (1) studies of conditional conservatism focusing on the different mapping of earnings to returns conditional on the sign of returns; and, (2) studies of unconditional conservatism focusing on the pre-determined understatement of earnings. The conditioning variable in the extant conditional conservatism literature is cum-dividend change in the value of equity whereas our conditioning variable is cum-free cash flow change in the market value of the enterprise; our conditioning variable brings the focus to the firm/enterprise where the transactions that are recorded in the accounting system occur. Our conditioning variable also recognizes change in the value of debt as a potentially important element of generated value. Our analysis of the accounting for net contributions by the owners of the enterprise captures ideas that are implicit in studies of unconditional conservatism. We demonstrate that unconditional conservatism affects the mapping between earnings and market returns both in the cross-section and over time due to variation in the amount of capital contributed by/distributed to owners each period as well as cross-sectional variation in the uses of invested capital. Further, we extend prior literature by capturing both forms of conservatism in the same regression framework. Our inclusion of an analysis of the effects of change in value due to both EVg and ECF, contributes to the extant literature, which has afforded little attention to the analysis of 5 accounting for changes in value via ECF.8 This is particularly important in the empirical literature on conditional accounting conservatism (also referred to as asymmetric timely loss recognition), which, as far as we are aware, has not considered the accounting for this form of change in market value at all.9 The remainder of the paper proceeds as follows. We begin, in section 2 by describing and providing reasons for our dissection of change in enterprise value into generated value change vs. contributed value change. We show that the accounting for generated value change in income is the same as the accounting for this source of value change on the balance sheet and only a portion of generated value change is recorded in the financial statements of the current period. On the other hand, the entire amount of contributed value change is either expensed in income or capitalized on the balance sheet. We provide two illustrations: contributed value invested in R&D and distributed value from the sale of property, plant, and equipment. Section 3 describes our empirical method and predictions. We describe our sample selection procedure in section 4 and we provide and discuss selected descriptive statistics. Section 5 analyzes and compares partitions of the data based on whether enterprise value is increasing or decreasing and on the source of this value increase/decrease; we provide evidence 8 Penman and Yehuda [2004] is one of the few examples. Penman and Yehuda focus on the value relevance of free cash flow (i.e., analysis of the relation between change in the value of the enterprise and free cash flow), which differs from our focus on the accounting for change in enterprise value via free cash flow (i.e., we analyze the relation between free cash flow (labelled ECF in our analyses) and the recording of this free cash flow in the financial statements). Collins et al. [2014] observe that the manifestation of asymmetric timeliness in cash flow from operations adds noise or bias to tests of conditional conservatism. There are two vast differences between Collins et al. and our study. First their focus is on cash flow from operations, while our focus is on ECF. Second, their focus is on eliminating cash flow effects from measures of conditional conservatism, while we demonstrate that including ECF in regressions of EPAT on EVg allows researchers to capture conservatism in the accounting treatment of contributions from/distributions to owners. 9 Much theoretical work has been done on generated vs contributed change in enterprise value and conservatism (see, for example, Zhang [2000], Easton [2009], and Ohlson [2009]) but we are unaware of empirical work on this issue. 6 from analyses of a 2 x 2 x 2 partition of the data (i.e., increase vs. decrease in enterprise value, internally generated growth vs. contraction, and cash provided by vs. cash distributed to the owners of the enterprise).10 We show that the recording of generated and contributed value varies considerably across these partitions and that the direction of enterprise cash flow (growth through cash inflow and contraction via cash outflow) affects the amount of generated value change that is recorded in current enterprise profit after tax, EPAT. Finally we show that a large portion (sometimes most) of the association between EPAT and change in the market value of the enterprise is due to contributed value change during the period (enterprise cash flow), hitherto largely ignored in the extant literature, rather than enterprise value generated during the period. In Section 6, we examine whether the empirical manifestation of change in enterprise value in the financial statements differs as expected across firm-year characteristics. To summarize, we predict and show that, when there is cash inflow during the period, the estimates of the coefficients relating EPAT to ECF, increase (decrease) with investment in R&D and advertising (capital expenditures). On the other hand, when there is value generated, the estimates of the coefficients relating EPAT to EVg decrease (increase) with investment in R&D and advertising (capital expenditures). These results are consistent with conservative accounting principles such as the immediate expensing of investments in R&D and advertising (accompanied by the deferred recognition of the benefits of R&D and advertising) compared with the capitalization, depreciation, and subsequent gain or loss recognition on investments in fixed assets. These aspects of accounting conservatism are captured only indirectly in the Basu framework. 10 The two empty sets are those with: (1) positive ΔEV and negative EVg and ECF outflow: and (2) negative ΔEV and positive EVg and ECF inflow. 7 We also examine the effect of debt on the degree of accounting conservatism observed in financial statements, which has been a focus of prior literature (see, for example, Ball, Robin, and Sadka [2008] and Kahn and Watts [2009]). Consistent with this literature we find that the coefficient relating EPAT to EVg increases with leverage. Finally, we examine the effect of balance sheet conservatism (proxied by the enterprise-level book-to-market ratio) on our measures of conservatism in EPAT. We find that more conservative balance sheets are associated with more conservative income statements (and vice versa). Section 7 presents a summary and conclusions. 2. Change in Enterprise Value and Conservative Accounting 2.1. Sources of Change in Enterprise Value and Links to the Extant Literature Our empirical analyses incorporate the key elements of a vast empirical literature on accounting conservatism. Our emphasis on: (1) the enterprise, which is the entity for which accounting transactions are recorded; (2) change in the market value of the enterprise; and, (3) sources of change in market value of the enterprise, permits analyses of the key elements of the extant literature within one framework. The extant empirical literature takes two primary approaches/perspectives -- the study of conditional conservatism and the study of unconditional conservatism. Ryan [2006, p. 511] describes these approaches as follows: Conditional conservatism involves the more timely recognition of bad news than good news in earnings (often referred to as asymmetric timeliness), as occurs with impairment accounting…Conditional conservatism is distinct from unconditional conservatism, which involves the predetermined understatement of the book value of net assets, as occurs with the immediate expensing of the costs of most intangibles. 8 Prior literature has largely focused on conditional conservatism and on the news-dependent nature of accounting. For example, Ryan [2006, p. 513] goes on to state that “Basu does not perceive unconditional conservatism to have any usefulness in contracting because of its newsindependent nature.” Similarly, Ball, Kothari and Nikolaev [2013] state that the key difference between conditional and unconditional conservatism is that conditional conservatism carries new information. Our focus on change in the value of the enterprise instead hones in on factors that will affect current and future enterprise profitability and the recording of the related transactions in enterprise earnings of either the current period or future periods. Some of the changes in the market value of the enterprise are reflected in the accountant’s measure of change in value, and others are not. Our perspective is that earnings record transactions in the current period while change in the market value of the enterprise captures the effect of changes in expectations related to transactions of the current period as well as changes in expected future transactions. This perspective is, in principle, very similar to that of Basu, but differs subtly inasmuch as our focus is on earnings recording the transactions themselves whereas Basu’s perspective is that earnings records (or fails to record) the information about generated value creation/loss that is (at the same time) reflected in the prices of equity shares. Our perspective also captures a source of conservatism analyzed by Feltham and Ohlson (1995) – the existence of positive NPV projects.11 11 Easton and Pae [2004] is an empirical analysis of accounting conservatism as modelled by Feltham and Ohlson [1995]. Following the form of earlier studies, where returns are the dependent variable as in Easton and Harris [1991], rather than the independent variable as in Basu [1997], Easton and Pae [2004] argue and show that, when enterprise cash flows and lagged net enterprise assets are added to a regression of returns on earnings and earnings changes: (1) a positive coefficient on enterprise cash flows captures conservatism associated with investments in positive net present value projects, the effects of which will not flow into the accounting statements until the expected future benefits are realized; and (2) a positive coefficient on change in lagged net enterprise assets implies accounting conservatism associated with the application of accounting rules to net enterprise assets in place. Our results are consistent with those in Easton and Pae [2004] although our method is quite different; and much more akin to the current literature on conditional and unconditional conservatism. 9 They point out that a firm may be perceived to have the opportunity, in expectation, to undertake strictly positive NPV projects. Such projects do not affect earnings in the current period but the same is not true for change in market value of the enterprise. The impetus for change in the market value of the enterprise is either, events that affect the current and future profitability of enterprise assets in place, or contributions from/distributions to the owners of the firm. We view change in market value (which is akin to the conditioning variable in the conditional conservatism literature) as a manifestation of the internally generated change in value, which is or is not recorded in current earnings, rather than as the impetus in and of itself for accounting conservatism.12 As an important aside, we recognize that change in the market value of debt is also a manifestation of enterprise value change.13 Put another way, generated value will be reflected in change in the value of the enterprise, and some of this value change will be recorded in the current earnings of the enterprise; that is, the conditioning variable is enterprise value change. Externally sourced (contributed/distributed) change in enterprise value has not been considered in the literature on conservative accounting and fits best into Ryan’s description, “pre-determined understatement of earnings,” whereby, for example, contributions used to fund 12 Some examples of generated value change (growth) include increases in expected asset turnover or profit margin, new product developments, increased brand recognition, and increased growth options due to weakening of competition or increases in the size of the economy as a whole. Stated differently, we define generated value change as any increase in the market value of the enterprise beyond the value change (growth) due to cash inflows from owners during the period. Similarly, a generated decrease in enterprise value could be driven by the same economic factors cited as growth examples. Notably, our definition of generated growth includes growth from what Roychowdhury and Watts [2007] refer to as “rents” and Ball Kothari and Nikolaev [2013] label as the g component of stock returns (which is associated with “revisions in the value of growth options or un-booked intangibles”). Generated value change may also be associated with the investment of ECF (i.e., contributed growth) in non-zero NPV projects, modelled in Feltham and Ohlson [1995]. 13 Empirically, and as expected, much of generated growth is captured in change in the market value of equity. This is due to the fact that the equity holders are the residual claimants to the value of the enterprise; they reap most of the benefits of growth and pay much of the cost of contraction because the return to debt holders is largely through contractual payments. As a practical empirical matter, our estimate of generated value change (growth) is highly correlated with equity returns. 10 capital expenditure will be capitalized while those used to fund R&D will be expensed. Thus, by focusing on the enterprise rather than equity and considering the separate and joint manifestation of generated and contributed change in value we are analyzing the effects of both conditional conservatism (mostly via the recording of generated value change) and unconditional conservatism (mostly via the recording of contributions/distributions). 2.2. The Empirical Manifestation of Change in Enterprise Value in the Financial Statements; Accounting Conservatism Our empirical analyses are based on a regression of EPAT on the corresponding market measure of enterprise value generation/loss (EVg) and a measure of net contributions from/distributions to the owners of the enterprise (ECF), with each of these variables deflated by beginning of year enterprise value (EV). Conservatism will be manifested in the coefficient relating EPAT to EVg being less than one or the coefficient relating EPAT to ECF being greater than zero because, absent conservatism, EPAT = EVg. Change in enterprise value, ΔEV is estimated as the sum of the market value of equity plus the market value of debt (estimated as the book value of debt). Net contributions, ECF, are calculated from the income statement and balance sheets as enterprise profit after tax, EPAT, minus the change in net enterprise assets, ΔNEA. Generated value creation/loss, EVg, is calculated as the sum of ΔEV and ECF; i.e., EVg is the change in the market value of the enterprise before net distributions to/contributions from the owners of the enterprise (alternatively this variable may be described as the cum-free cash flow increase in the market value of the enterprise, which is analogous to the cum-dividend change in the market value of equity). 11 The identity, ECF = EPAT - ΔNEA captures the fact that the sum of the effects of a dollar of contributions/distributions on the income statement and the balance sheet will be one dollar. Cash inflow from the owners of the firm will be used to pay expenses of the period or invested in an enterprise asset, which will be expensed (i.e., depreciated) in future periods as the investment generates future sales. Re-writing the identity as EPAT = ECF + ΔNEA, highlights the fact that, in the absence of conservatism in accounting (i.e., ΔNEA = ΔEV), EPAT = ECF + ΔEV. Accounting conservatism will be manifested in ΔNEA less than ΔEV and a correspondingly higher EPAT; e.g., when ECF is spent on R&D, the corresponding ΔNEA is, ceteris paribus, zero and EPAT = ECF, yet ΔEV equals ECF. The identity, ECF = EPAT - ΔNEA also makes it clear that, absent ECF, EPAT will equal ΔNEA; that is, value generated/lost via the enterprise assets in place will, unlike growth/contraction via contributions from the owners of the enterprise, have the same effect on the income statement and the balance sheet. The difference between the accounting for generated value change and value change via ECF may be illustrated by continuing the R&D example. Value creation due to the success of R&D may affect current sales and hence current EPAT as well as future sales and profitability; the only effect on current ΔNEA will be through current EPAT; the portion of generated value that is captured in both the income statement and the balance sheet will be the same and will range from none if all of the benefits of the R&D accrue in future periods to one if all of the benefits accrue in the current period. As another example, consider generated value change which enhances/reduces the usefulness of enterprise assets; again, the only effect on current ΔNEA will be through current EPAT (i.e., the portion of generated value change that is captured in both the income statement and the 12 balance sheet will be the same). But there is an asymmetry in the recognition of enterprise value increase vs. enterprise value decrease in both the income statement and in the balance sheet. Only the portion of value increases that affects current profitability is captured in current EPAT and current ΔNEA whereas expenses (write-downs and restructuring charges) associated with reduction in expected future profitability (i.e., value decreases) will be taken in current EPAT and current ΔNEA. This is the form of conservatism investigated by Basu and characterized as “recognize losses in value immediately but recognize gains in value with a lag.” To summarize, our perspective on conservative accounting introduces accounting for ECF as well as generated value creation/destruction, EVg; this perspective adds to the literature because these ECF effects have, by and large, not been hitherto considered in the empirical accounting literature, and they matter. Our perspective also focuses on the enterprise as a whole, which is the entity where transactions that are recorded by accounting arise. 3. Empirical Method and Predictions We begin with a simple relation between the accounting measure of the change in enterprise value and the change in the market value of the enterprise: it = it eit Then, we note that its owners, it ( it (A) it has two components: (1) net distributions from the enterprise to to the owners of the enterprise is signed positive while it from the owners of the firm is signed negative) and, (2) generated change in the value of existing assets of the enterprise it . That is: 13 it = it Substituting for (B) it it it = in (A): it it eit (C) This relation motivates our core regression: it = α10t + α11t α12t it ε1it (1) That is, it it = α10t + α11t it = α10t + α11t it it = α10t + α21t it α12t α12t it ε1it it 1 ε1it That is: where the variable it α22t is calculated as ε1it it + (1a) it . it is contributed value change and is value change generated by the existing enterprise assets. In other words, it increase/decreases with positive/negative it . it it , it it and increases/decreases with negative/positive For expositional convenience in the remainder of the paper, we will refer to the variables it and it as EPAT, EVg and ECF. 14 Note that α11t α21t ; i.e., the mapping from EVg to EPAT is the same as the mapping from EVg to ΔNEA. Also note that α22t ‐α12t 1; i.e., ECF goes to/comes from either EPAT of the period or a decrease/increase in net enterprise assets. Much of our analyses consider partitions of the data according to the sign of ECF. Absent conservatism, the accounting measure of value generated, EPAT, will be equal to the market measure of value added, EVg, and it follows that the coefficient relating EPAT to EVg will be one and the coefficient on ECF will be zero; but, with conservatism the estimate of the coefficient on EVg may be less than one and the estimate of the coefficient on ECF may be greater than zero. Conservatism in the accounting treatment of EVg will likely differ from conservatism in the accounting treatment of ECF. We will discuss each form of conservatism separately. Internally generated increase in value (i.e., positive EVg) due to such things as discovery of new technology, acquisition of new contracts, which may be serviced with the existing enterprise assets, effective cost-cutting, re-investment of internally generated cash in positive NPV projects, etc., may affect both current EPAT and future EPAT. The change in enterprise value will reflect the present value of the effect of growth on current and future EPAT. The relation between current EPAT and this internally generated value change (i.e., α21t ) will depend on the extent to which the growth effect is expected to be more transitory (with a relation that is close to one-to-one) vs. more permanent with a relation that is much less than one-to-one. The effect of generated value change on NEA will be equal to the effect on EPAT. Internally generated decrease in value (i.e., negative EVg) due to such things as loss of comparative technological advantage, loss of market share, cost increases, etc., may affect both 15 current EPAT and future EPAT. Since these effects are more likely to be transitory rather than permanent on average (otherwise the firm will go out of business), the relation between EPAT and internally generated contraction (i.e., α21t ) is likely to be closer to one-to-one than the relation between EPAT and internally generated value creation (i.e., positive EVg). Further, generally accepted accounting principles place greater verification thresholds on increases in value recorded in EPAT than on decreases in value recorded in EPAT, which can also contribute to a greater portion of EVg being recognized in current-period EPAT when EVg is negative. In short, consistent with prior literature, we predict that the accounting system is more likely to record internal contraction than internal growth (i.e., we will observe a higher coefficient α21t for internal contraction). In other words, the accounting system records more of decreases in EVg in the current EPAT than increases in EVg, consistent with the notion of conditional conservatism. If there is conservatism in the accounting treatment of ECF and there is no conservatism in the accounting treatment of internally generated value increase/decrease (EVg) (i.e., there is no reason why the difference between EPAT and EVg is systematically related to EVg), the estimate of coefficient relating EPAT to EVg will be one and the estimate of the coefficient relating EPAT to ECF will capture the conservatism in the recording of ECF in EPAT. We return to the R&D example to illustrate this point. In the absence of conservatism in the accounting treatment of ECF, EPAT = ΔNEA + ECF = ΔEV + ECF. So, if we were to regressed EPAT on ΔEV and ECF in the absence of 16 conservatism, the coefficients would both be one. In regression (1a) we regress EPAT on ΔEV + ECF (that is, EVg) and ECF; in the absence of conservatism, EPAT is identically equal to EVg and it follows that the coefficient on EVg will be one and the other variable, ECF, can have no incremental explanatory power (i.e., the coefficient relating EPAT to ECF will be zero). Conservatism in the recording of ECF introduces variation in EPAT but does not affect EVg. The effects of $100 of ECF invested in a zero NPV enterprise will be: (1) ΔEV + ECF = $100 + (-$100) = EVg = 0; and, (2) R&D will all be expensed so that EPAT = - $100. ECF will completely explain EPAT (because EVg is zero), with a coefficient of 1; i.e., EPAT = 1*0 + 1*($100).14 Suppose, however, that the R&D investment has positive NPV of $10, all of which is recorded in current period sales, then the effects will be: (1) ΔEV + ECF = $10; and, (2) EPAT = - $90. Now, EPAT = 1*$10 + 1*(-$100) = -$90. The R&D example is a most vivid example but the logic follows for less conservative accounting actions (such as accounting depreciation being greater than economic depreciation). Expanding this example further, suppose that there is conservatism in the accounting treatment of ECF and in the accounting treatment of internally generated value change; in this example this may mean that some of the effects of the positive NPV is expected in future periods rather than in the current period and the coefficient on EVg will be less than one. As a further illustration, suppose that there is $100 of ECF outflow funded entirely by the sale of a depreciated enterprise asset. If the book value and the salvage (sale) value are the same (i.e., the accounting has not been conservative), EVg = EPAT=ΔEV + ECF =ΔNEA + ECF = $100 + $100 = 0 and EPAT = 1*EVg+0*ECF = 1*0 + 0*$100 =0. But, if the book value is 14 If partial capitalization of R&D was permitted under the accounting rules, the coefficient on ECF will be less than one. For example, if only half of the R&D in this example was expensed, ΔNEA would be $50, EPAT would be $50, which would equal 1*0 + 0.5*(-$100) = -$50. 17 $50 (i.e., book value is depreciated beyond economic salvage value), ΔNEA = $50 and EPAT = $50 because there is a gain on sale (i.e., EPAT = 1*0 + 0.5*$100). 4. Sample Selection and Selected Descriptive Statistics We obtain annual financial statement data from the Compustat (Xpressfeed) database for fiscal years 1963 – 2012. We match this with stock return data from the CRSP database. The sample period begins in 1963 in order to ensure data availability in Compustat and for consistency with prior studies examining accounting conservatism (e.g., Basu 1997; Ball, Kothari and Nikolaev 2013). We exclude foreign incorporated firms (we require that Compustat FIC=USA), financial institutions (SIC codes 6000 – 6900), utilities (SIC codes 4900 – 4999), observations with negative market value or total assets (potential data errors), and observations with beginning-of-fiscal-year stock prices less than one dollar. We require that all observations included in the sample have sufficient data available for the calculation of all variables in Table 1. To mitigate the influence of extreme observations on our results, we truncate observations that fall in the top or bottom 2.5 percent of any of the variables included in the primary regression equation (equation 1a) as well as price-deflated earnings and annual returns. Table 1 provides descriptive statistics for our sample of 110,144 observations. Initial evidence of conservatism in accounting is seen in the fact that the mean (median) change in enterprise value as a percentage of opening enterprise value (ΔEV) is greater (0.109 (0.040)) than the mean (median) change in the book value of the enterprise as a percentage of opening enterprise value (ΔNEA) (0.056 (0.036)). The mean EPAT is less than the mean ΔNEA (0.040 compared with 0.056) and the mean (median) ECF is -0.016 (0.004). The mean (median) R&D plus advertising (RDADV) is 0.047 (0.018) of enterprise value; 66.23 percent of observations 18 have non-zero (positive) R&D and advertising.15 Mean (median) capital expenditures are 0.080 (0.050) of enterprise value. The mean (median) ratio of the market value of net financial liabilities to opening enterprise value is 0.126 (0.110) and 32.85 percent of sample observations exhibit negative values of net financial liabilities, indicating that these observations have net financial assets.16 Table 2 reports Spearman correlations among key variables. We discuss some highlights. The correlation between our dependent variable EPAT and the dependent variable in Basu (X/P, i.e., price scaled earnings) is 0.931 and the correlation between EVg and RET is 0.966. Thus, it is not at all surprising that the estimates of our coefficient relating EPAT to EVg are generally very similar to the estimates of the coefficient relating earnings to equity returns. As expected the correlations between EPAT and EVg and EPAT and ECF are both positive (0.408 and 0.274) and highly significant. The correlation between EVg and ECF is significantly positive (0.158); that is, in general, higher generated value change is associated with more cash outflow. We will refer back to this table when correlations among other variables become pertinent to subsequent analyses and discussions. 5. Initial Empirical Analyses The basic premise of our paper is that the accounting for change in the market value of the enterprise depends upon the source of the value change. To shed light on this premise we first examine samples of observations in which the value of the enterprise is either increasing or 15 This result is not tabulated. That is, the equity holders own the enterprise and a “pile of cash” (as in Microsoft, for example). It follows that calculating EV as MVE + NFL (where negative NFL implies net financial assets) remains valid. 16 19 decreasing. Then we further partition according to the source of increase/decrease in enterprise value, resulting in a total of six sub-samples: (1) Enterprise Growth, Value Creation, Net Cash Inflow (+ΔEV, +EVg, ECF in): there is growth on every dimension (i.e., all is going well and there is a further injection of cash); (2) Enterprise Growth, Value Creation, Net Cash Outflow (+ΔEV, +EVg, ECF out): the enterprise is growing but there is net cash outflow (i.e., all is going well and there is a removal of cash); (3) Enterprise Growth, Value Loss, Net Cash Inflow (+ΔEV, -EVg, ECF in): the enterprise is growing because of net cash inflow despite negative internal growth (i.e., the enterprise growth is due to the injection of cash by its owners): (4) Enterprise Contraction, Value Creation, Net Cash Outflow (-ΔEV, +EVg, ECF out): the enterprise is contracting because net cash outflow is greater than value generated (i.e., generated growth is not large enough to replace the value that the owners are taking out of the enterprise); (5) Enterprise Contraction, Value Loss, Net Cash Inflow (-ΔEV, -EVg, ECF in): the enterprise is contracting despite net cash inflow (i.e., the enterprise is fairing badly and there is an injection of cash) and, (6) Enterprise Contraction, Value Loss, Net Cash Outflow (-ΔEV, -EVg, ECF out): there is contraction on every dimension (i.e., the enterprise is fairing badly and there is removal of cash). 20 5.1. Growth vs. Contraction A summary of the results from estimation of regression (1a) for the entire sample of observations and for the sub-samples with enterprise value increasing and with enterprise value decreasing is provided in Table 3, Panel A. We also present results from the estimation of earnings-return regressions to provide evidence of manifestations of conservatism that are observed when taking our enterprise value change perspective but are not observed when implementing the news/information perspective. Since the results for the full sample are more or less an average of the results from the enterprise value increasing and the enterprise value decreasing sub-samples, we focus our discussion on the results for these sub-samples. There are two striking differences in the results across these sub-samples. First, the estimate of the coefficient relating EPAT to EVg is not significantly different from zero for enterprises that are increasing in value (-0.001 with a tstatistic of -0.07) while this estimate is significantly positive for the enterprises that are decreasing in value (0.141 with a t-statistic of 14.41).17 That is, none of the generated increase in value is captured in contemporaneous EPAT while 14.1 percent of the enterprise value loss is captured in contemporaneous EPAT. Second, the estimates of the coefficients relating EPAT to ECF are quite similar for each of the groups of enterprises (0.192, with a t-statistic of 11.18, for enterprises with increasing value and 0.179, with a t-statistic of 8.94, for enterprises with decreasing value). That is, conservatism in the reporting of ECF in EPAT is similar across enterprises that are increasing in value and for enterprise that are decreasing in value, on average. 17 t-statistics reported throughout our analyses are calculated using two-way clustered standard errors, clustered by firm and year. 21 In Panel B of Table 3, we present the results of the Basu-style regression of earnings on returns with a dummy variable capturing the sign of the return. The results are similar to those in the extant literature; the estimate of the coefficient on returns when returns are positive is not significantly different from zero (0.018 with a t-statistic of 1.25) and the estimate of this coefficient when returns are negative is significantly positive (0.182 with a t-statistic of 9.13). For comparison, we also run a Basu-style regression at the enterprise level. We regress EPAT on EVg interacted with a dummy variable capturing the sign of EVg. The results are quite similar to those in the earnings on returns regression. The estimate of the coefficient on EVg is not significantly different from zero (-0.009 with a t-statistic of -0.98) and the estimate of this coefficient when returns are negative is significantly positive (0.189 with a t-statistic of 14.89). The higher t-statistic on negative EVg and the slightly higher adjusted R2 (0.148 cf. 0.137) relative to these statistics in the regression of earnings on returns are consistent with our motivation to examine the enterprise as a whole, which is the entity where transactions that are recorded by accounting arise. Throughout our analyses, we also report the number of negative return observations in each estimation sample. While the full sample is roughly equally split between positive and negative return observations, observations within sub-samples (1) – (6) are either almost all positive (when EVg is positive) or almost all negative (when EVg is negative) consistent with the high correlation between EVg and RET reported in Table 2.18 Due to the homogenous nature of returns within each sub-sample, we report, in the analyses that follow, simple regressions of earnings on returns for each sub-sample (as opposed to piecewise asymmetric timeliness regressions), and then compare earnings/return coefficients across sub-samples. 18 Results are qualitatively very similar if we run regressions with only the positive return observations when EVg is positive and with only negative return observations when EVg is negative. 22 5.2. Sources of Change in Enterprise Value Descriptive statistics and regression results for each of the six enterprise value change sub-samples are summarized in Table 4. Panel A presents descriptive statistics for the subsamples. Panel B presents simple Spearman correlations. Panel C presents regression results from estimating regression (1a). Panel D reports the results from a regression of earnings on returns along with the number of negative return observations in each sub-sample. (i) Enterprise Value Change and Value Generation We begin with a comparison of the two sub-samples of observations where the enterprise value is increasing and there is value generated from the existing enterprise assets (i.e., subsamples (1) and (2). The results for these sub-samples are summarized in the first two columns of Table 4. In sub-sample (1), with 26,020 observations, the owners have contributed to the enterprise value change via ECF inflow while in subsample (2), which has 28,709 observations, value has been distributed to the owners of the enterprise via ECF outflow. Sub-sample (1) is comprised of enterprises that are increasing in value due to both creation of value via the enterprise assets in place and cash inflow from the owners of the enterprise. Consistent with this, the median ΔEV (0.397) reported in Panel A is the highest among the six sub-samples. Sub-sample (2) enterprises have a similar median EVg (0.289 cf. 0.262), but are distributing some of the enterprise value back to owners, resulting a slightly lower median ΔEV of 0.209, which is still the second-highest enterprise value change among the six subsamples. Panel A also shows that, while both sub-samples with increasing enterprise value report positive median current-period enterprise profit, they are still investing in both intangibles (the median RDADV for sub-sample (1) is similar to the median for sub-sample (2), 0.018 cf. 0.020) and 23 fixed assets (the median PPE for sub-sample (1) is similar to the median for sub-sample (2), 0.307 cf. 0.302).19 Most of the cash inflow in sub-sample (1) comes from debt holders (median ECF of -0.084 and ΔNFL of 0.081) and much of the cash outflow in sub-sample (2) goes to debt holders (median cash flow of 0.059 and ΔNFL of -0.036). The enterprises that are distributing cash to the owners are more than twice the size of those receiving cash from the owners (median EV of $0.246 billion cf. $0.113 billion). Turning to the simple correlations presented in Panel B, it is notable that the correlation between EPAT and EVg is not significantly different from zero (-0.005) for sub-sample (1), but positive and significant (0.221) for sub-sample (2). The correlation between EPAT and ECF for sub-sample (1) is the lowest (0.049) among the six sub-samples, while the correlation between EPAT and ECF for sub-sample (2) is the highest among the six sub-samples (0.295). A summary of the results from the estimation of regression (1a) is presented in Panel C. We continue the comparison of the two sub-samples of observations where enterprise value is increasing and there is value generated from the existing enterprise assets. The estimate of the coefficient on EVg in sub-sample (1) shows a statistically significant 3 cent net expense in the current period per dollar of value generated. This may, for example, be an indication of current period accrual expenses, which have been made to enhance future profitability, but were not funded by current period cash flow. More importantly, the coefficient on EVg for sub-sample (1) is the smallest (i.e., least positive) of the six sub-samples consistent with the notion that, in this sample where growth is most evident, accounting captures net expenses (the estimate of the coefficient relating EPAT to EVg is significantly negative, -0,030 with a t-statistic of -4.19), 19 All analyses and comparisons based on PPE have been repeated replacing PPE with capital expenditure. Results are very similar, which is not surprising because PPE and capital expenditure are highly correlated (0.680 for the full sample). 24 which are associated with the generation of profits in future periods rather than in the current period. Further, note that, while the estimate of the coefficient on EVg in significantly negative for sub-sample (1) where there is cash inflow, it is significantly positive for sub-sample (2) (0.015) when there is cash outflow. The estimate of the coefficient on ECF (i.e., 0.108) in sub-sample (1) implies that the effect of conservative accounting for ECF is 0.108 per dollar of ECF inflow; in other words, the effect of ECF inflow on EPAT is overstated by 10.8 cents per dollar of ECF. The estimate of the coefficient on ECF in sub-sample (2), where ECF is positive, (i.e., there is net cash outflow) is much higher than in subsample (1) (0.290 vs. 0.108); the higher coefficient in subsample (2) is consistent with accounting conservatism having a greater effect on the reporting of cash outflows of enterprises that are increasing in value than on the reporting of cash inflows for enterprises that are increasing in value. The interpretation of these coefficients is that the difference between accounting EPAT and economic EPAT is 0.108 per dollar of ECF when there is ECF inflow and 0.290 when there is ECF outflow; in other words, the over-statement of the EPAT (i.e., net expense) effect of the ECF inflow is less than the overstatement of the EPAT (i.e., net profit) effect when there is ECF outflow. The characteristics of these sub-samples (see Panel A) provide some indication of the reasons for this difference. A possible explanation is the fact that, although the enterprise value is increasing in both sub-samples, the observations in sub-sample (1) have a median increase in NEA of 14.4 percent of enterprise value while those in sub-sample (2) have virtually no change in NEA (0.016); i.e., much of the cash inflow is going to build enterprise assets but, not surprisingly (in light of the fact that the enterprises in sub-sample (2) are also growing), cash 25 outflow is not coming from sale of enterprise assets – rather it is coming from EPAT of the period. For comparison, Panel D of Table 4 also reports the Basu-style estimates of earnings/return coefficient for each of the six sub-samples. The coefficient of 0.003 on RET in sub-sample (1) is not significantly different from zero. Similar to the coefficient on EVg, the coefficient on RET for sub-sample (1) is the smallest earnings/return coefficient among the six sub-samples, and also the only one that is not significantly positive. On the other hand, for sub-sample (2), where there is cash outflow, the estimate of the coefficient on RET is significantly positive (0.39 with a tstatistic of 2.54). (ii) Enterprise Contraction and Value Loss Next we compare the two sub-samples (5 and 6) of observations where the enterprise value is decreasing and the existing enterprise assets are losing value. In sub-sample (5), with 22,098 observations, the owners have contributed to enterprise value change via ECF inflow; while in subsample (6), with 22,081 observations, ECF outflow has been distributed to the owners of the enterprise. The results for sub-samples (5) and (6) are summarized in the last two columns of Table 4. The enterprises in sub-sample (5) are contracting due to loss in value of the enterprise assets, but they continue to receive support from the owners of the enterprise via ECF inflows. These are relatively small enterprises (median EV of $0.105 billion compared with $0.160 for sub-sample (6)). These enterprises are, on average, the most unprofitable enterprises across all six sub-samples (median EPAT of 0.010); they have higher intangible intensity and much lower 26 property plant and equipment than enterprises in sub-sample (6) (median RDADV of 0.022 cf. 0.014 and mean PPE of 0.161 cf. 0.229). These are growth-oriented (vs. value-oriented) enterprises with the lowest mean BTM (0.388) of any of the six sub-samples. Consistent with the contraction experienced by these enterprises, they exhibit the highest correlation between EPAT and EVg of the six sub-samples (0.394). Moving to Panel C, the estimate of the coefficient on EVg in regression (1a) of 0.156 for subsample (5) indicates that there is a 15.6 cent loss in the current period per dollar of internal contraction, while the estimate of this coefficient for sub-sample (6) is smaller (0.118); in words, more of the value lost relates to current earnings for the sub-sample of enterprises for which owners are contributing cash (presumably this cash contribution leads to expenses of the period (for example R&D, advertising and capital expenditure used to stem future losses)) relative to those where the owners are removing cash. The striking feature in the comparison across the results from sub-samples (5) and (6) is the difference in the coefficients on ECF across these two samples of contracting enterprises. Net cash inflow appears to go to support EPAT of the current period (coefficient of 0.359), whereas little of net cash outflow comes from current EPAT. For comparison, Panel D of Table 4 also reports the earnings/return coefficient for each subsample. The coefficients of 0.193 and 0.208 for sub-samples (5) and (6), respectively, are very similar to each other. Consistent with extant literature, these coefficients are significantly positive and much higher than the coefficients for sub-samples (1) and (2). Notably, the similar earnings/returns coefficients for these two sub-samples mask the differences in ECF conservatism that become apparent when the sources of enterprise value loss are considered. 27 (iii) Enterprise growth, value loss, net cash inflow The results for sub-sample (3) are summarized in column (3) of Table 4. While enterprise growth (contraction) is driven by generated value increase/decrease for most enterprises, the enterprises in sub-sample (3) are experiencing value increase due to large injections of cash by the owners of the enterprise despite experiencing value loss/contraction during the fiscal year. This is an unusual situation, evidenced by the fact that sub-sample (3) is the smallest of the six sub-samples and only contains 5.6 percent of the observations in the full sample (see Table 1, Panel B). The mean levels of EV in Panel A indicate that sub-sample (3) contains, on average, the smallest enterprises in the sample, consistent with these enterprises’ small market capitalization serving as a contributing factor in their ability to grow the value of the enterprise by attracting additional capital despite experiencing value loss. These enterprises are relatively unprofitable (median EPAT of 0.042), they invest little in intangibles (lowest median RDADV of 0.013), and heavily in fixed assets (median PPE of 0.350). Because of their high asset tangibility, these enterprises are able to raise a high percentage of their market value from ECF inflows (most negative median ECF across all subsamples, -0.181), resulting in large increases in leverage (median ΔNFL of 0.170). The results from estimation of regression (1a) are summarized in Panel C. The estimate of the coefficient on EVg (0.251) indicates that there is a 25.1 cent EPAT loss in the current period per dollar of generated enterprise value loss. It is also interesting to note that the annual equity return (RET) is negative (i.e., in Basu parlance, there is bad news) for most of the observations in sub-sample (3) (6,255 of 6,606). Consistent with prior literature (e.g., Basu 1997), the estimates of the coefficients on EVg and the earnings/return coefficient are much higher (0.251 and 0.196, 28 respectively, with t-statistics of 7.06 and 4.77) for this sub-sample than for sub-samples (1) and (2) where EVg was positive. The estimate of the coefficient on ECF (0.064) implies that there is little conservatism in the accounting for ECF for these observations. This is consistent with the high levels of PPE in these enterprises and indicates that the majority of the financing raised by these enterprises goes towards investments that are capitalized into NEA (the implied coefficient relating ΔNEA to ECF is 0.936) rather than covering expenses of the current period. These enterprises are capitalizing ECF at a slightly higher rate than those in sub-sample (1) and again we see that the capitalization rate in this sub-sample is much higher than the liquidation rate in sub-sample (2), where the implied coefficient relating ΔNEA to ECF is 71.0 cents per dollar. (iv) Enterprise contraction, value generation, net cash outflow The results for the analysis of sub-sample (4) are summarized in column (4) of Table 4. Similar to sub-sample (3) this subsample is comprised of enterprises where the overall growth pattern runs counter to its internal growth pattern. In this case, the enterprise is contracting despite internal growth, a somewhat rare occurrence indicated by the fact that only 6.70 percent of our observations are in sub-sample (4). Sub-sample (4) enterprises have high beginning-ofperiod leverage (median NFL of 0.370, which is the highest among the six sub-samples). Despite experiencing positive median EVg of 0.052, these enterprises are contracting due to large cash outflows to the owners of the enterprise, generally resulting in a deleveraging of the enterprise by returning capital to debt holders (median ΔNFL of -0.086). The results from estimation of regression (1a) are reported in Panel C. The estimate of the coefficient on EVg (0.229) indicates that there is a 22.9 cent profit in the current period per dollar 29 of generated value change. It is interesting to note that the coefficients on EVg and the earnings/return coefficient are significantly positive and relatively high (0.229 and 0.218, respectively, with t-statistics of 8.55 and 6.24); this, perhaps, seems odd because EVg and the majority of returns are positive (“good” news) and this would suggest lower EVg and earnings/return coefficients. In un-tabulated analysis, we penetrated this result further by running the earnings/return regression as specified by Basu (i.e., with an intercept and slope dummy, which is one if returns are negative, zero otherwise). The estimate of the coefficient on positive returns is, contrary to the prediction in Basu, significantly positive (0.162 with a t-statistic of 3.08). That is, we have isolated a sample of observations where equity returns are positive and there is a substantial estimated coefficient relating earnings to return. Note that the EVg and earnings/return coefficients are much lower in the other sub-samples in which there is value generation (i.e., sub-samples (1) and (2)). The key characteristic of this sample is that the enterprises are contracting due to cash outflow despite positive internal growth, highlighting the importance of considering overall enterprise growth as well as the direction of ECF in the analysis of accounting conservatism. The estimate of the coefficient on ECF (i.e., 0.069) implies that for each dollar of cash outflow there is little conservatism in the accounting for EPAT for these observations. Perhaps the most relevant comparison for this coefficient is that of sub-sample (2). Both sub-samples are comprised of enterprises with internal value creation who are returning cash to owners; enterprises in sub-sample (4), however, are somewhat less profitable and make larger payouts that shrink the size of the enterprise. Accordingly, enterprises in sub-sample (2) are able to source a larger percentage of cash outflows from current EPAT relative to sub-sample (4) enterprises. 30 5.3. Putting it all together We have shown that the recording of enterprise value increases/decreases in EPAT differs according to whether the value increase/decrease is generated via existing enterprise assets or contributed by/distributed to the owners of the enterprise. We have also shown differences across sub-samples based on the sign of the change in the market value of the enterprise, the sign of value generation/loss, and the sign of enterprise cash flow. Since generated enterprise value change is highly correlated with equity returns, which is the variable that is the focus of much of the extant work on conservatism in accounting, many of our empirical results from the analysis of the relation between EPAT and EVg are similar to those from the analysis of the relation between earnings and returns. Our primary contribution to the empirical understanding of the effects of growth on conservative accounting is through the introduction of cash flow to/from the owners of the enterprise. There are several key results. First, as highlighted in the comparison of sub-samples (5) and (6) where the enterprise is contracting, there is loss in value of the existing enterprise assets, and there is either cash outflow or cash inflow, the extent to which cash flow affects the recording of EPAT varies a great deal (0.359 when there is cash inflow and not significantly different from zero when there is cash inflow). Second, the direction of cash flow affects the sign and magnitude of the coefficient relating EPAT to EVg; in other words identifying the direction of contributed/distributed value helps our understanding of conservatism in accounting for internally generated value change. This effect is best seen in the comparison of the estimates of the coefficients relating EPAT to EVg across sub-samples (1) and (2), where there is enterprise growth and generation of value from existing 31 enterprise assets, and across sub-samples (5) and (6), where there is enterprise contraction and loss of value of the existing enterprise assets. Partitioning growing enterprises (sub-samples (1) and (2)) on the sign of enterprise cash flow facilitates identification of a significantly negative relation relating EPAT to EVg when there is cash inflow (coefficient estimate of -0.030 with a tstatistic of -4.19) and a significantly positive relation when there is cash outflow (coefficient estimate of 0.015 with a t-statistic of 2.45). Partitioning contracting enterprises (sub-samples (5) and (6)) on the sign of cash flow facilitates identification of a significantly higher coefficient relating EPAT to EVg when there is cash inflow (0.156) than when there is cash outflow (0.118); the t-statistic for the differences between these two coefficient estimates is 2.95. Third, our results highlight the role of overall enterprise growth/contraction in accounting conservatism as opposed to positive or negative equity return “news.” This point is best illustrated by sub-sample (4), which isolates a sample of observations where equity returns are generally positive, but the estimated coefficient relating earnings to return is actually the largest of the six sub-samples. The key characteristic of this sample is that there is overall enterprise contraction due to cash outflow despite positive EVg and generally positive equity returns. 5.4. The relative magnitude of the effects of EVg and ECF on EPAT. The magnitude of the effect of ECF on recorded EPAT is, in many cases, equal to or greater than the magnitude of the effect of generated value change on EPAT. We summarize these effects in Figure 1. We plot the marginal effects of one-standard deviation changes in EVg and ECF for each regression sample, relative to the standard deviation of EPAT in each sample.20 In all but sub-sample (6), where the enterprise is contracting on all dimensions and the 20 Note that this is equivalent to normalizing the regression (1a) variables to have a mean of zero and standard deviation of 1 within each regression sample and then plotting the absolute value of the normalized regression coefficients for EVg and ECF. 32 coefficient relating EPAT to ECF is not significantly different from zero, a one standard deviation change in ECF contributes substantially to EPAT. In fact, for the full sample, the estimated effect of a one-standard deviation change in ECF is slightly larger than the estimated effect of a one standard-deviation change in EVg, such that a one standard deviation change in ECF is associated with a change of 27.10 percent of a standard deviation of EPAT, while an equivalent change in EVg is only associated with a change of 23.62 percent. These estimated marginal effects are not statistically different from each other, indicating that ECF explains a roughly equal amount of EPAT variation as EVg for the full sample. As further illustrated in Figure 1, the marginal effect of an ECF change is significantly lower than that of an EVg change in sub-samples (3) and (4), roughly equivalent to that of an EVg change in sub-samples (1) and (5), and significantly greater than (almost four times as much as) an EVg change in sub-sample (2). In short, variation in ECF, which has been, by and large, omitted from previous studies of accounting conservatism, accounts for much of the observed variation in EPAT across enterprises. 6. Accounting for Change in Enterprise Value and Enterprise Characteristics Thus far, we have demonstrated the effects of conservative accounting in recording the effects of value change generated from the existing net enterprise assets vs. contributed/distributed value change. We have shown that each dollar of ECF is recorded in either EPAT or in ΔNEA; that is, if the amount of value change due to ECF captured in EPAT is low, the amount recorded in change in the book value of enterprise assets is high and vice-versa. Importantly, the entire amount of the ECF is recorded in either EPAT or ΔNEA. For generated growth, however, the coefficient relating EPAT to EVg is the same as the coefficient relating 33 ΔNEA to EVg; that is the manifestation of conservatism on the income statement is the same as the manifestation of conservatism in consecutive balance sheets. In this section, we further demonstrate these aspects of our framework by examining how the EVg and ECF coefficients in regression (1a) vary with enterprise-year characteristics. This additional analysis serves to validate the inferences we have made thus far regarding the effects of accounting conservatism on the estimated coefficients in regression (1a), and highlights three possible contributions of our work: (1) enhancing our understanding of the nature of conservative accounting as we illustrated via discussion of the R&D and change in enterprise asset value examples, above; (2) focusing our understanding of the effects of the presence of and the amount of debt on conservative accounting; and, (3) enhancing our understanding of the link between balance sheet conservatism and income statement conservatism. These analyses are based on regressions where we add interaction terms to regression (1a). We undertake four sets of analyses based on four (separately considered) interaction terms. We independently partition the data into deciles of: (1) investment in intangibles (i.e., the ratio of R&D plus advertising expense to beginning of year market value of the enterprise); (2) investment in fixed assets (i.e., the ratio of property, plant and equipment to beginning of year market value of the enterprise);21 (3) the amount of debt (i.e., the ratio of beginning of year net financial liabilities to market value of the enterprise); and, (4) the enterprise-level book-tomarket ratio (i.e., the ratio of the book value of the enterprise at the beginning of the year to the market value of the enterprise). 21 Because the correlation between (annual capital expenditure (i.e., CAPEX/EV) and the balance in property, plant and equipment (i.e., PP&E/EV) is large (Pearson correlation of 0.608)), we conduct our analyses with partitions based only on PP&E; we obtain very similar results if instead we partition on CAPEX. 34 Specifically, for each characteristic, by fiscal year, we rank all observations in the full sample into deciles. We then scale the decile rankings to have a mean of zero and a range of one. This decile rank transformation mitigates the effect of extreme observations and facilitates interpretation of the regression coefficients. We then run regression (1a) including interaction terms between the characteristic variable and EVg as well as the characteristic variable and ECF, as follows: it = 1+ 2 it 3 it 4 _ 5 it ∗ _ it 6 it ∗ _ it εit where DEC_X indicates a decile transformation of variable X. All other variables are as defined in regression (1a). The results of these regressions are summarized in Table 5 for each of the sub-samples (1) to (6) described in section 5. The results for the partition on RDADV are summarized in Panel A of Table 5.22 As expected, for enterprise that are increasing in value and generating value internally (sub-samples (1) and (2)), the coefficient relating EPAT to EVg decreases with DEC_RDADV (coefficient estimates of -0.059 and -0.042 with t-statistics of -6.07 and -5.60); although the R&D leads to expected future profits, the effect on current EPAT is negative due to the expensing of R&D. For contracting firms (sub-samples (5) and (6)), the coefficient relating EPAT to EVg increases with DEC_RDADV (coefficient estimates of 0.075 and 0.046 with t-statistics of 3.74 and 2.60); for these observations the expensed R&D is associated with negative EVg – it seems that the R&D is being made in an attempt to counteract the expected decline in future profits reflected in the negative EVg. As expected, the interaction between DEC_RDADV and ECF is highest for the 22 For consistency with the rest of the analyses in this section, we rank RDADV into deciles in the same way we rank other characteristics despite the fact that almost exactly one third of the observations in the full sample have zero RDADV. For robustness, we performed an alternate analysis where we partitioned RDADV into no, low, and high RDADV groups with roughly equal observations in each tercile. All inferences remain unchanged. 35 three sub-samples of observations where there is ECF inflow; the greater the R&D the greater the amount of ECF expensed in current EPAT. The negative estimate of the coefficient on the interaction between DEC_RDADV and ECF (-0.120) in sub-sample (4) where the enterprise value is declining because ECF outflow is greater than enterprise value generated is interesting; we postulate that this reflects the success of R&D, generating both future growth (positive EVg) as well as surplus ECF in the current period from prior positive EVg projects.23 Since higher intangibles intensity may imply lower tangibles intensity and since we observe a negative correlation between RDADV and PPE (0.203 for the full sample), we expect the estimates of the interactions based on DEC_PPE to be, more or less, the mirror image of those based on RDADV. This is seen for some of the sub-samples. For example, for sub-samples (1) and (2), the estimates of the coefficients on the interaction of DEC_PPE and EVg are significantly positive (0.034 and 0.035 with t-statistics of 2.02 and 2.73) and, for each of the subsamples where there is ECF inflow, the estimates of the coefficients on the interaction of DEC_PPE and ECF are significantly negative (-0.348, -0.258 and -0.617). Notable exceptions are the significantly positive estimates of coefficient on the interaction of DEC_PPE and EVg for the sub-samples of contracting firms (sub-samples (5) and (6)); this result may reflect the existence of more write-offs in EPAT for these observations simply because there are more assets to be written off. Much work has been done on the effect of debt on the coefficient relating earnings to returns, focusing particularly on this coefficient when returns are negative (i.e., news is bad); see, for example, Ball, Robin, and Sadka [2008] and Kahn and Watts [2009]. This work generally observes a larger coefficient relating earnings to negative returns when the debt level is higher. 23 Future work will penetrate this postulate further. 36 We observe a similar pattern for the coefficient relating EPAT to EVg. This is evidenced by the positive coefficient on EVg*DEC_NFL in sub-sample (5) and (6), see Panel C, where EVg is negative (coefficient estimates of 0.050 and 0.100). Interestingly, the estimates of this coefficient are also significantly positive (0.064 and 0.037) for sub-samples (1) and (2) where EVg is positive. This result is new to the literature as far as we are aware; the extent to which change in generated enterprise value is captured in current EPAT increases with debt level when the enterprise is doing well and when the enterprise is doing poorly. The estimates of the coefficient on ECF*DEC_NFL are negative and significant for all six sub-samples. This suggests that when enterprises have higher levels of debt, ECF inflows are more likely to be capitalized into NEA than expensed to EPAT, and that cash outflows are more likely to come from liquidating NEA than from current EPAT. Our emphasis to this stage has been on the effects of conservative accounting on the income statement and on changes in consecutive balance sheets. The effects of conservatism will, however, accumulate on the balance sheet leading to lower book-to-market ratios. The estimate of the coefficients on EVg*DEC_BTM reported in Panel D are positive for all sub-samples and much higher for the sub-samples where enterprise value is declining and EVg is negative; for these observations, the higher book-to-market will mean that the likelihood of a write-off affecting current EPAT will be higher because there will be less of a buffer between the book value of assets and the market value of assets. The estimate of the coefficient on ECF*DEC_BTM is negative and significant in all sub-samples. Recall that the lower the estimate of the coefficient relating EPAT to ECF, the higher the coefficient relating ΔNEA to 37 ECF; it follows that, again, we observe that the less conservative the balance sheet, the less conservative the income statement. 7. Summary and conclusions We introduce an alternate perspective on accounting conservatism, focusing on the recording of change in enterprise value in the financial statements. We demonstrate, both analytically and empirically, that the manner and degree to which the accounting system records change in enterprise value depends upon both the sign (positive vs. negative) and source (internal vs. external) of the value change. Our framework allows researchers to measure an aspect of accounting conservatism that has not been examined in prior studies: conservatism in the accounting treatment of enterprise cash flows (i.e., contributed value change). We show that this source of value change explains a considerable portion of EPAT; in fact, it explains almost four times that explained by variation in internal growth for a sub-sample of observations where there is enterprise growth due to internally generated growth yet there is net cash outflow to the owners of the firm. We validate our framework’s ability to capture novel aspects of accounting conservatism by demonstrating that the degree to which the accounting system captures both generated and contributed enterprise value change varies with enterprise characteristics such as intangible intensity, capital expenditures, debt level, and enterprise-level book-to-market ratio. 38 References Ball, R., S. Kothari, and V. Nikolaev (2013), “On Estimating Conditional Conservatism”. The Accounting Review, 88, 3, 755-787. Ball, R., A. Robin, and G. Sadka (2008), “Is Financial Reporting Shaped by Equity Markets or by Debt Markets? An International Study of Timeliness and Conservatism”. Review of Accounting Studies, 13: 168-205. Basu, S. (1997), “The Conservatism Principle and the Asymmetric Timeliness of Earnings”. Journal of Accounting and Economics, 24: 3-37. Collins, D., P Hribar, and X. Tian (2014), “Cash flow asymmetry: Causes and implications for conditional conservatism research”. Journal of Accounting and Economics, 58: 173 – 200. Damodaran, A. (2012), “Investment Valuation: Tools and Techniques for Determining the Value of Any Asset”. 3rd Edition. John Wiley & Sons, Inc. Easton, P. (2009), Discussion of “Accounting Data and Value: The Basic Results”. Contemporary Accounting Research, 26: 261-272. Easton, P., and T. Harris (1991), “earnings as an Explanatory Variable for Returns”. Journal of Accounting Research, 29: 19-36. Easton, P., M. McAnally, G. Sommers, and X. Zhang (2015), “Financial Statement Analysis and Valuation”. 5th Edition. Cambridge Business Publishers. Easton, P., and J. Pae (2004), “Accounting Conservatism and the Relation between Returns and Accounting Data”. Review of Accounting Studies, 9: 495-522. Feltham, G., and J. Ohlson (1995), “Valuation and Clean Surplus Accounting for Operating and Financial Activities”. Contemporary Accounting Research, 11: 689-731. Gode, D., & J. Ohlson, (2013). “Financial Statement Analysis and Valuation.” Retrieved December 28, 2014, from http://www.godeohlson.com/ Khan, M., and R. Watts (2009), “Estimation and Empirical Properties of a Firm-year Measure of Accounting Conservatism”. Journal of Accounting and Economics, 48: 132-150. 39 Lawrence, A., R. Sloan, and Y. Sun (2014), “Why are Losses Less Persistent than Profits? Curtailments versus Conservatism”. Working paper, University of California, Berkeley. Nissim, D. and S. Penman (2001), “Ratio Analysis and Equity Valuation: From Research to Practice”. Review of Accounting Studies, 6, 109-154. Ohlson, J. (2009), “Accounting Data and Value: The Basic Results”. Contemporary Accounting Research, 26: 231-259. Penman, S. (2012), “Financial Statement Analysis and Security Valuation”. 5th Edition. McGraw-Hill Education. Penman, S., and N. Yehuda (2009), “The Pricing of Earnings and Cash Flows and an Affirmation of Accrual Accounting”. Review of Accounting Studies, 14:453-479. Roychowdhury, S., and R. Watts (2007), “Asymmetric Timeliness of Earnings, Market-to-Book and Conservatism in Financial Reporting”. Journal of Accounting and Economics, 44: 2– 31. Ryan, S., (2006), “Identifying Conditional Conservatism”, European Accounting Review, 15, 4: 511-525. Wahlen, J., S. Baginski, and M. Bradshaw (2015), “Financial Reporting, Financial Statement Analysis, and Valuation”. 8th Edition. Cengage Learning. White, G., A. Sondhi and D. Fried (2003), “The Analysis and Use of Financial Statements”. 3rd Edition. John Wiley & Sons, Inc. Zhang, X. (2000), “Conservative Accounting and Equity Valuation.” Journal of Accounting and Economics 29: 125-149. 40 Appendix A: Variable Measurement A.1: Terminology Throughout our paper, we adopt the terminology of Gode and Ohlson [2013] for distinguishing between the productive enterprise of the firm and the financing of that enterprise. As noted by Gode and Ohlson [2013], enterprise activities are often referred to as operating activities. We do not use the word “operating” because there is ambiguity about its precise meaning. In particular, GAAP operating cash flows differ from enterprise cash flows. We adapt the following from Gode and Ohlson [2013] as a terminology reference: Ours Enterprise Value [EV] Enterprise profit after tax [EPAT] Enterprise assets [EA] Enterprise liabilities [EL] Net enterprise assets [NEA = EA – EL] Enterprise cash flows [ECF = EPAT – Change in NEA] Financial assets [FA] Financial liabilities [FL] Net financial liabilities [NFL = FL- FA] Alternatives used in practice Firm Value [FV] Market Value of the Firm [MVF] Net operating profit after tax [NOPAT] Earnings before interest but after taxes [EBIAT] Unleveraged NI, prefinancing NI Operating assets [OA] Operating liabilities [OL] Net operating assets [NOA = OA – OL] Free cash flows to the firm [FrCF or FCF] Unlevered cash flows “Cash” or cash and marketable securities Debt, Finanical Obligations [FO] Net debt = Debt – “Cash” Net Finanical Obligations [NFL] A.2: Variable Definitions # = Compustat Data Item. Δ = Change between current and prior fiscal year. Enterprise Value (EV) = Market Value of Equity (MVE) plus Net Financial Liabilities (NFL). Market Value of Equity (MVE) = Fiscal year end price (#PRCC_F) times common shares outstanding (#CSHO), from Compustat. Scaled by beginning-of-period enterprise value (EVt-1). Net Financial Liabilities (NFL) = Financial Liabilities (FL) minus Financial Assets (FA). Scaled by beginning-of-period enterprise value (EVt-1). Financial Liabilities (FL) = Debt in current liabilities (#DLC) plus long term debt (#DLTT) plus preferred stock (#PSTK) minus preferred treasury stock (#TSTKP) plus preferred dividends in arrears (#DVPA). Financial Assets (FA) = Cash and short term investments (#CHE) plus investments and advances-other (#IVAO). 41 Net Enterprise Assets (NEA) = Net Financial Liabilities (NFL) plus Common Stockholders’ Equity (CSE) plus Minority Interest (#MIB). Scaled by beginning-of-period enterprise value (EVt-1). Common Stockholders’ Equity (CSE) = Common equity (#CEQ) plus preferred treasury stock (#TSTKP) minus preferred dividends in arrears (#DVPA). Enterprise Profit After Tax (EPAT) = Comprehensive Net Income (CNI) plus Net Financial Expense (NFE) plus minority interest in income (#MII). Scaled by beginning-of-period enterprise value (EVt-1). Comprehensive Net Income (CNI) = Net Income (#NI) minus preferred dividends (#DVP) plus Clean Surplus Adjustment to Net Income (CSA). Clean Surplus Adjustment to Net Income (CSA) = marketable securities adjustment (#MSA) minus lag marketable securities adjustment (#MSAt-1) plus cumulative translation adjustment (#RECTA) minus lag cumulative translation adjustment (#RECTAt-1). Net Financial Expense (NFE) = After-tax interest expense ((#XINT)*(1 – marginal tax rate)) plus preferred dividends (#DVP) minus after tax interest income ((#IDIT)*(1-marginal tax rate)) plus unusual financial expense ((#MSAt-1)-(#MSA)). Comprehensive Net Financial Income (CNFI) = Comprehensive Net Income (CNI) minus Enterprise Profit After Tax (EPAT). Marginal Tax Rate = The top statutory federal corporate tax rate plus 2% average state corporate tax rate. The top statutory federal corporate tax rate was 52% in 1963, 50% in 1964, 48% in 1965 – 1967, 52.8% in 1968 – 1969, 49.2% in 1970, 48% in 1971 – 1978, 46% in 1979 – 1986, 40% in 1987, 34% in 1988 – 1992 and 35% in all sample years thereafter. Enterprise Value Generated (EVg) = Change in Enterprise Value (ΔEV) plus Enterprise Free Cash Flow (ECF). Scaled by beginning-of-period enterprise value (EVt-1). Enterprise Free Cash Flow (ECF) = Enterprise Profit After Tax (EPAT) minus change in Net Enterprise Assets (ΔNEA). Scaled by beginning-of-period enterprise value (EVt-1). Earnings scaled by Price (X/P) = Earnings before extraordinary items (#IB) scaled by beginning of period Market Value of Equity (MVEt-1). Stock Return (RET) = Cumulative buy-and-hold fiscal year stock return, from CRSP. Research, Development, and Advertising (RDADV) = R&D Expense (#XRD) plus Advertising Expense (#XAD). Scaled by beginning-of-period enterprise value (EVt-1). Property Plant and Equipment (PPE) = Property plant and equipment, net of accumulated depreciation (#PPENT). Scaled by beginning-of-period enterprise value (EVt-1). Enterprise Book-to-Market (BTM) = Beginning-of-period Net Enterprise Assets (NEAt-1) divided by beginning-of-period Enterprise Value (EVt-1). 42 Table 1 Descriptive Statistics Panel A: Full sample descriptive statistics N Mean σ EV ($ Bil.) 110,144 1.917 12.368 110,144 0.109 0.482 ΔEV 110,144 0.056 0.156 ΔNEA 110,144 0.040 0.105 EPAT 110,144 0.092 0.468 EVg 110,144 -0.016 0.149 ECF 110,144 0.037 0.127 X/P 110,144 0.084 0.477 RET 110,144 0.047 0.084 RDADV 110,144 0.080 0.095 CAPX 110,144 0.373 0.358 PPE 110,144 0.126 0.334 NFL 110,144 0.020 0.138 ΔNFL 110,144 0.660 0.468 BTM p1 0.003 -0.739 -0.333 -0.343 -0.795 -0.518 -0.422 -0.745 0.000 0.000 0.005 -0.804 -0.312 -0.024 p25 0.036 -0.172 -0.016 0.014 -0.168 -0.077 0.005 -0.233 0.000 0.022 0.106 -0.056 -0.049 0.306 Median 0.140 0.040 0.036 0.057 0.047 0.004 0.054 0.032 0.018 0.050 0.268 0.110 0.006 0.603 Panel B: Frequencies among enterprise growth sub-samples (1) (2) (3) (4) (5) + ΔEV + ΔEV + ΔEV - ΔEV - ΔEV + EVg + EVg - EVg + EVg - EVg ECF in ECF out ECF in ECF out ECF in N 26,020 28,709 5,932 7,391 21,080 Pct (%) 23.62 26.06 5.39 6.71 19.14 p75 0.651 0.295 0.115 0.092 0.278 0.064 0.095 0.321 0.060 0.102 0.532 0.333 0.075 0.936 (6) - ΔEV - EVg ECF out 21,012 19.08 p99 32.071 1.781 0.576 0.251 1.657 0.316 0.321 1.589 0.373 0.458 1.555 0.814 0.483 2.043 Total 110,144 100 This table presents descriptive statistics for the full sample of observations used in our analysis. Panel A presents sample statistics. σ and p denote the sample standard deviation and percentiles, respectively. Panel B presents frequencies of observations within each enterprise growth sub-sample used in our empirical analyses. As described in Section 5 of the text, we partition the full sample into six enterprise growth sub-samples based on the sign of ΔEV, EVg, and ECF. A positive (negative) sign denotes that observations in the sub-sample are restricted to those where the corresponding variable is greater than or equal to zero (less than zero). All variables are defined in Appendix A.2. . 43 Table 2 Spearman Correlations Among Key Measures [1] [1] EV [2] ΔEV [3] ΔNEA [4] EPAT [5] CNFI [6] EVg [7] ECF [8] X/P [9] RET [10] RDADV [11] PPE [12] NFL [13] ΔNFL [14] BTM [2] [3] [4] [5] 0.201 0.018 0.367 0.044 0.323 0.366 0.033 -0.054 -0.044 -0.165 0.215 0.919 0.112 0.408 -0.062 0.072 -0.174 -0.699 0.274 -0.062 0.059 0.302 0.318 0.931 -0.156 0.214 0.882 0.105 0.414 -0.039 -0.045 0.014 -0.070 -0.061 0.160 -0.030 0.156 0.155 0.345 -0.491 0.034 0.005 -0.077 0.137 -0.827 -0.032 0.119 0.682 -0.106 -0.103 -0.216 0.080 -0.031 0.318 -0.605 [6] [7] [8] [9] [10] [11] [12] [13] 0.158 0.395 0.282 0.966 0.166 0.419 0.014 -0.019 -0.067 0.022 0.171 0.088 0.345 0.170 -0.188 0.067 0.198 0.193 0.057 -0.218 0.468 -0.173 -0.828 -0.123 -0.182 -0.037 0.099 -0.059 0.151 0.242 0.336 0.149 -0.092 0.694 0.612 -0.030 This table presents Spearman rank correlations among key variables in the sample. All variables are defined in Appendix A.2. Correlations presented in bold typeface are statistically significant at the 1% level. 44 Table 3 OLS regressions of Earnings Measures on Market Measures Panel A: Regressions of EPAT on Generated vs. Contributed Enterprise Value Changes Adj. R2 N Intercept EVg ECF *** *** *** Full Sample 0.038 0.053 0.191 0.138 110,144 (7.20) (4.28) (11.34) Enterprise Growth Sample 0.069*** -0.001 0.192*** 0.097 60,661 (12.09) (-0.07) (11.18) 0.141*** 0.179*** 0.227 49,483 Enterprise Contraction Sample 0.045*** (8.44) (14.41) (8.94) Panel B: Basu Model Replication Intercept RET D(RET<0) 0.018 -0.010** X/P 0.065*** (8.96) (1.25) (-2.34) EPAT Intercept 0.071*** (12.46) EVg -0.009 (-0.98) D(EVg<0) -0.018*** (-4.24) RET*D(RET<0) 0.182*** (9.13) EVg*D(EVg<0) 0.189*** (14.89) Adj. R2 0.137 Adj. R2 0.148 N 110,144 N 110,144 N RET <0 51,533 N EVg <0 48,024 This table presents estimates from OLS regressions of accounting measures of changes in enterprise and equity value on market-based measures of changes in enterprise and equity value. Panel A presents estimates of regressions of EPAT on generated vs. contributed changes in enterprise value (EVg and ECF, respectively). The growth sample is composed of firm-year observations in which enterprise value increased during the fiscal year (ΔEV >=0), while the contraction sample is composed of firm-year observations in which enterprise value decreased during the fiscal year (ΔEV < 0). Panel B presents estimates of OLS regressions of earnings on returns, following Basu [1997]. D(RET<0) is an indicator variable equal to one when RET < 0 and zero otherwise. For comparison, Panel B also presents regressions of EPAT on EVg, interacted with an indicator variable [D(EVg<0)] equal to one when EVg < 0 and zero otherwise. In each analysis, we report the number of sample observations for which the indicator variable is equal to one. All other variables are defined in Appendix A.2. t-statistics reported in parenthesis are calculated using two-way clustered standard errors, clustered by firm and year. *, ** and *** indicate (two-tailed) significance at the 10%, 5% and 1% levels respectively. 45 Table 4 Summary of Results by Sub-Sample (1) (2) (3) (4) + ΔEV + ΔEV + ΔEV - ΔEV + EVg + EVg - EVg + EVg ECF in ECF out ECF in ECF out (5) - ΔEV - EVg ECF in (6) - ΔEV - EVg ECF out Panel A: Sample Medians for Key Measures EV ($ Bil.) ΔEV ΔNEA EPAT EVg ECF RDADV PPE NFL ΔNFL BTM 0.113 0.397 0.144 0.065 0.262 -0.084 0.018 0.307 0.101 0.081 0.595 0.246 0.209 0.016 0.080 0.289 0.059 0.020 0.302 0.132 -0.036 0.667 0.079 0.081 0.197 0.042 -0.071 -0.181 0.013 0.350 0.152 0.170 0.654 0.122 -0.055 -0.052 0.074 0.052 0.127 0.014 0.407 0.370 -0.086 0.965 0.105 -0.234 0.062 0.010 -0.331 -0.061 0.022 0.161 0.010 0.059 0.388 0.160 -0.235 -0.007 0.046 -0.156 0.049 0.014 0.229 0.137 -0.028 0.618 Panel B: Spearman Correlations between Key Measures EPAT x EVg EPAT x ECF EVg x RDADV EVg x PPE EVg x NFL EVg x ΔNFL EVg x BTM ECF x RDADV ECF x PPE ECF x NFL ECF x ΔNFL ECF x BTM PPE x RDADV PPE x NFL PPE x ΔNFL PPE x BTM NFL x ΔNFL NFL x BTM ΔNFL x BTM -0.005 0.049 0.095 -0.163 -0.292 -0.190 -0.223 -0.012 -0.100 0.030 -0.470 -0.058 -0.223 0.467 0.395 0.690 0.217 0.588 0.423 0.221 0.295 0.116 -0.094 -0.260 -0.402 -0.068 0.077 0.171 0.079 -0.705 0.306 -0.122 0.460 0.080 0.707 0.080 0.589 -0.037 0.172 0.106 -0.043 0.040 0.211 -0.291 0.134 -0.025 -0.049 0.143 -0.483 0.041 -0.243 0.422 0.331 0.620 0.141 0.564 0.344 0.268 0.216 0.047 0.147 0.114 -0.529 0.318 0.036 0.189 0.226 -0.845 0.452 -0.085 0.338 -0.068 0.577 -0.148 0.565 -0.322 0.394 0.218 -0.143 0.345 0.414 -0.046 0.372 -0.110 0.012 0.181 -0.497 0.066 -0.257 0.473 0.307 0.708 0.068 0.568 0.303 0.256 0.134 -0.067 0.318 0.353 -0.009 0.315 -0.016 0.188 0.226 -0.752 0.383 -0.164 0.432 0.019 0.658 -0.077 0.608 -0.157 Observations 26,020 28,709 5,932 7,391 21,080 21,012 46 Table 4 (continued) Summary of Results by Sub-Sample (1) (2) (3) (4) + ΔEV + ΔEV + ΔEV - ΔEV + EVg + EVg - EVg + EVg ECF in ECF out ECF in ECF out Panel C: Estimates of Regression (1a) Intercept 0.073*** 0.057*** EVg ECF Adjusted R2 (12.14) -0.030*** (-4.19) 0.108*** (4.76) (14.86) 0.015** (2.45) 0.290*** (6.80) 0.043*** (5.54) 0.251*** (7.06) 0.064*** (2.94) 0.040 0.098 0.062 Panel D: Estimates of Earnings-on-Returns Regressions Intercept 0.045*** 0.071*** 0.021*** (5) - ΔEV - EVg ECF in (6) - ΔEV - EVg ECF out 0.046*** (11.30) 0.229*** (8.55) 0.069*** (3.40) 0.067*** (10.20) 0.156*** (12.67) 0.359*** (9.15) 0.058*** (19.53) 0.118*** (9.45) -0.027 (-0.65) 0.074 0.225 0.054 0.047*** (7.13) 0.193*** (13.03) 0.074*** (13.62) 0.208*** (11.44) RET (5.25) 0.003 (0.21) (12.22) 0.039** (2.50) (2.64) 0.196*** (4.77) 0.056*** (5.86) 0.218*** (6.24) Adjusted R2 0.000 0.025 0.033 0.081 0.104 0.108 Observations 26,020 1,876 28,709 319 5,932 5,707 7,391 1,821 21,080 21,018 21,012 20,792 N RET <0 This table provides descriptive statistics and OLS regression estimates for each enterprise growth sub-sample. Each column presents results for one of the enterprise growth sub-samples (1) – (6) defined in Table 1. Panel A presents the sub-sample medians for key variables. Panel B presents univariate correlations between key variables within each sub-sample. Correlations presented in bold typeface are statistically significant at the 1% level. Panel C reports the estimated coefficients and adjusted R2 from estimating regression (1a) on each sub-sample. The dependent measure in these regressions is EPAT. Panel D reports the estimated coefficients and adjusted R2 from regressing earnings (X/P) on returns for each subsample. We also report the number of observations in each sub-sample with negative RET. All variables are defined in Appendix A.2. t-statistics reported in parenthesis are calculated using two-way clustered standard errors, clustered by firm and year. *, ** and *** indicate (two-tailed) significance at the 10%, 5% and 1% levels respectively. 47 Table 5 Effects of firm-year characteristics on the regressions of EPAT on generated vs. contributed enterprise value changes, by sub-sample (1) (2) (3) (4) (5) (6) + ΔEV + ΔEV + ΔEV - ΔEV - ΔEV - ΔEV + EVg + EVg - EVg + EVg - EVg - EVg ECF in ECF out ECF in ECF out ECF in ECF out Panel A: Effect of R&D Plus Advertising Intercept 0.069*** 0.055*** (11.94) 0.008 (1.12) -0.016*** (-2.88) -0.059*** (-6.07) 0.115*** (4.82) 0.312*** (6.01) (14.88) 0.014*** (3.22) 0.021*** (3.75) -0.042*** (-5.60) 0.286*** (6.86) 0.025 (0.78) 0.043*** (5.83) -0.019 (-1.44) 0.226*** (6.45) 0.132 (1.44) 0.084*** (3.66) 0.254*** (3.38) 0.101 0.103 0.135 0.075 0.270 0.056 (12.05) 0.024*** (3.98) -0.017* (-1.96) 0.034** (2.02) 0.181*** (8.19) -0.348*** (-10.98) 0.056*** (14.61) 0.010 (1.58) 0.019*** (2.62) 0.035*** (2.73) 0.292*** (7.90) -0.218*** (-3.92) 0.044*** (6.17) 0.038*** (4.36) 0.200*** (5.64) -0.014 (-0.18) 0.131*** (4.79) -0.258*** (-4.08) 0.047*** (11.62) -0.012** (-2.21) 0.230*** (9.11) -0.058 (-0.86) 0.067*** (3.11) 0.085 (1.58) 0.066*** (10.19) 0.012** (1.99) 0.164*** (10.96) 0.104*** (5.08) 0.316*** (8.86) -0.617*** (-16.76) 0.059*** (20.26) -0.006 (-1.18) 0.135*** (9.58) 0.070*** (4.10) -0.021 (-0.49) 0.044 (0.71) Adjusted R2 0.120 0.105 0.130 0.074 0.251 0.059 Observations 26,020 1,876 28,709 319 5,932 5,707 7,391 1,821 21,080 21,018 21,012 20,792 DEC_RDADV EVg EVg*DEC_RDADV ECF ECF * DEC_RDADV Adjusted R2 Panel B: Effect of Asset Tangibility Intercept 0.073*** DEC_PPE EVg EVg*DEC_PPE ECF ECF*DEC_PPE N RET <0 48 0.047*** (11.19) 0.015*** (2.75) 0.232*** (8.61) 0.105** (2.16) 0.065*** (3.11) -0.120*** (-3.23) 0.061*** (9.54) 0.014 (1.49) 0.140*** (13.21) 0.075*** (3.74) 0.311*** (9.16) 0.459*** (11.21) 0.057*** (19.26) 0.001 (0.11) 0.117*** (9.65) 0.046*** (2.60) -0.024 (-0.56) 0.084** (2.13) Table 5 (Continued) Effects of Firm-Year Characteristics (1) (2) (3) (4) + ΔEV + ΔEV + ΔEV - ΔEV + EVg + EVg - EVg + EVg ECF in ECF out ECF in ECF out Panel C: Effect of Debt Level Intercept 0.066*** DEC_NFL EVg EVg*DEC_NFL ECF ECF*DEC_NFL Adjusted R2 (5) - ΔEV - EVg ECF in (6) - ΔEV - EVg ECF out (10.69) -0.023*** (-3.09) -0.005 (-0.77) 0.064*** (6.86) 0.107*** (5.14) -0.370*** (-7.89) 0.059*** (15.67) -0.018*** (-4.17) 0.013*** (2.61) 0.037*** (6.79) 0.287*** (8.17) -0.402*** (-12.99) 0.037*** (4.68) 0.024 (1.63) 0.184*** (6.34) -0.029 (-0.29) 0.069*** (3.31) -0.193*** (-3.15) 0.051*** (12.51) -0.029*** (-4.29) 0.169*** (5.88) 0.100 (0.98) 0.159*** (5.54) -0.310*** (-5.30) 0.063*** (10.04) -0.047*** (-5.04) 0.174*** (14.42) 0.050* (1.81) 0.244*** (6.52) -0.585*** (-11.06) 0.065*** (21.41) -0.033*** (-5.21) 0.170*** (14.96) 0.100*** (5.75) 0.044 (1.08) -0.346*** (-4.95) 0.091 0.140 0.098 0.157 0.239 0.133 0.048*** (12.70) -0.035*** (-4.35) 0.198*** (6.68) 0.138** (2.17) 0.144*** (5.95) -0.197** (-2.43) 0.070*** (9.96) -0.010 (-1.09) 0.223*** (10.19) 0.251*** (5.72) 0.226*** (6.40) -0.729*** (-12.03) 0.057*** (19.71) -0.003 (-0.35) 0.165*** (12.30) 0.243*** (6.32) 0.153*** (5.26) -0.527*** (-7.00) Panel D: Effect of Balance Sheet Conservatism Intercept 0.070*** 0.054*** (11.31) 0.015** (2.29) -0.014 (-1.28) 0.020 (0.88) 0.137*** (6.63) -0.398*** (-9.71) (14.43) 0.004 (0.59) 0.017** (2.18) 0.034** (2.15) 0.345*** (12.71) -0.358*** (-5.06) 0.040*** (5.38) 0.020 (1.49) 0.195*** (5.55) 0.013 (0.13) 0.091*** (4.05) -0.250*** (-3.20) Adjusted R2 0.112 0.111 0.102 0.107 0.261 0.161 Observations 26,020 1,876 28,709 319 5,932 5,707 7,391 1,821 21,080 21,018 21,012 20,792 DEC_BTM EVg EVg*DEC_BTM ECF ECF*DEC_BTM N RET <0 This table presents estimates from OLS regressions of EPAT on EVg and ECF (regression 1a) for each sub-sample, including interaction terms between relevant firm-year characteristics and both EVg and ECF. DEC_ denotes that the corresponding variable has been ranked into deciles. We sort all observations in the full sample into deciles by fiscal year. Decile ranks are then scaled to have a mean of zero and range of one. All variables are defined in Appendix A.2. t-statistics reported in parenthesis are calculated using two-way clustered standard errors, clustered by firm and year. *, ** and *** indicate (two-tailed) significance at the 10%, 5% and 1% levels respectively. 49 Estimated marginal effect as a % of one EPAT std. dev. Figure 1 Normalized marginal effects of changes in enterprise value on estimated EPAT EVg 60% ECF 50% 40% 30% 20% 10% 0% Full Sample +EV +EVg ECF in +EV +EVg ECF out *** +EV -EVg ECF in ** -EV +EVg ECF out *** -EV -EVg ECF in -EV -EVg ECF out *** This figure presents the marginal effects of one standard deviation changes in enterprise value on EPAT. Each column plots the magnitude of the effect of a one standard deviation change in EVg or ECF on EPAT based on the estimated coefficients from regression (1a) for each regression sample listed along the horizontal axis. The regression sub-samples are defined in Table 1. The height of each column is scaled by the standard deviation of EPAT for the corresponding sample, facilitating comparison across sub-samples. This is equivalent to normalizing all regression variables within each regression sample and plotting the (absolute) normalized coefficients. For example, the leftmost column in the figure was calculated by multiplying the full sample EVg coefficient (0.053, Table 3, Panel A), by the standard deviation of EVg for the full sample (0.468, Table 1). This marginal effect of approximately 0.025 was then scaled by the standard deviation of EPAT in the full sample (0.105, Table 1). The height of the column indicates that a one standard deviation change in EVg in the full sample is associated with a 0.2362 standard deviation change in EPAT (23.62%). *, ** and *** indicate that the heights of the two columns for the corresponding sample are significantly different at the (two-tailed) 10%, 5% and 1% levels respectively, based on Wald tests of the corresponding normalized regression coefficients. 50
© Copyright 2026 Paperzz