Search

This study examines the potential determinants of LIFO liquidations using a multivariate tobit model.1 Unlike the univariate analysis used in prior LIFO liquidation studies (e.g., Davis, Kahn, and Rozen [1984] and Tse [1990]), our multivariate analysis takes into account both the correlations among these potential determinants and the magnitude of LIFO liquidations. Simple correlation analysis indicates that the sales changes and earnings changes of LIFO liquidation firms are highly correlated. Thus, while univariate support for both the declining-sales and declining-earnings explanations of LIFO liquidations could indicate that both explanations are correct, it could also be the case that support for either explanation stems from the high correlation between sales changes and earnings changes. Our multivariate approach helps avoid such ambiguities in interpretation and supports two conclusions. First

This study reexamines the LIFO adoption decision by focusing on what can be inferred about the sophistication of managers' and investors' decision making through analyzing the effect on stock prices of the resolution of publicly disclosed uncertainty about the LIFO adoption/nonadoption decision. I establish and test implications of three models of the LIFO adoption/nonadoption decision: the traditional tax benefit model; an extension of Kang's [1988] real value model; and a functional fixation model. Each model yields distinct predictions about the mean excess stock return when uncertainty about the LIFO adoption/nonadoption decision is resolved, and about the cross-sectional relations between individual excess stock returns and several explanatory

This paper presents a conceptual evaluation of the stock price effects of LIFO tax savings. In periods of inflation, adopting LIFO can create cash-flow benefits by deferring tax payments. Whether LIFO tax savings are valued by investors has been extensively investigated through examinations of the stock price reaction to initial announcements of LIFO adoptions.1 While evidence of positive abnormal returns at announcements of LIFO adoptions proportional to the projected nominal LIFO tax savings would be viewed as consistent with investors recognizing the expected tax benefits of LIFO, overall empirical evidence available to date does not indicate that LIFO-adoption-related abnormal returns are positively associated with the size of potential nominal LIFO tax savings.2

It is an acknowledged fact that inflation accounting is one of the most controversial issues in accounting research. The Anglo-Saxon accounting profession has made several (unsuccessful) attempts to agree on an inflation accounting method to be applied in company practice.' General price level adjusted (GPLA) accounting,2 while being far from generally accepted by accounting practitioners and academics, still has considerable support in the United States. If GPLA is to be used, the calculations involved should be consistent

This study presents a dynamic model of inventory accounting choice and thereby permits a systematic consideration of that choice over time. I also restrict the dynamic model sufficiently to generate a static, crosssectional model similar to those used in previous studies (e.g., Lee and Hsieh [1985] and Caster and Simon [1986]). Estimation of the dynamic model identifies correlates of the change in inventory methods. Empirical estimates of the cross-sectional model extend previous research on firm characteristics related to the use of different inventory accounting methods. Consistent with theories of optimal inventory accounting choice, I find that factor prices play an important role in the accounting choice. Size, change in gross national product, and to a small extent industry membership also play a role in the transition. Assuming that managers act to maximize firm values, I develop an econometric "threshold" model in which a dichotomous inventory method choice is determined by an underlying continuous variable. A change occurs when the continuous variable crosses a threshold. The model is simplified by assuming that changes of accounting "state" follow a first

This paper examines the effect that imperfectly informed capital market agents have on the equilibrium paths of output, investment, and asset prices of value maximizing firms. Though information is imperfect, rational expectations is imposed as an equilibrium condition. It is found that both asset prices and corporate decisions are simultaneously affected in such a way that the efficiency-optimality relationship between them is pre- served. The effect of a fuller information structure is to move the economy from one efficient markets equilibrium to another. elaborated. Extant beliefs regarding accounting information are crystallized at two polar extremes. Policy-making bodies like the SEC and the FASB and most prac- titioners believe that investors are naive in processing information, and are principally concerned with the earnings per share figure as reported by accoun- tants. Investor decision rules are invariant to the rules for computing income, and therefore if the "right" or "proper" measure of income were not adopted, investors would be misled. In recent years stock price researchers have rapidly accumulated evidence against this naive investor hypothesis and in favor of the efficient markets hypothesis. For example, Kaplan and Roll (12) demonstrated that firms switching from the deferral method to the flow-through method for investment credit and firms switching from accelerated to straight-line deprecia- tion, in each case increasing their reported income, did not experience abnormal price increases. Sunder (23) found that firms switching from the FIFO to the LIFO method of inventory valuation, thereby decreasing their reported income, did not suffer in stock price performance. These findings raise a serious question. Do accounting statements have any informational impact on security prices at all? Several studies tested and rejected the null hypothesis that the periodic financial statements issued by corporations have no information content for capital market agents. Ball and Brown (1) tested for information content of the annual earnings report and found that foreknowledge of the earnings number could be used to generate excess returns. Brown and Kennelly (5) replicated the

In this study we first investigated the hypothesis that the negative tax effect associated with LIFO inventory liquidations induces managers to avoid liquidations, compared to FIFO firms which do not face such tax incentives. Our results generally supported this hypothesis. We then investigated why all LIFO firms did not avoid inventory liquidations. The results suggest that economic conditions override the tax effects. In the next section we review previous literature in order to develop our hypotheses. Section 3 provides a description of the methods used to estimate inventory liquidations for both LIFO and FIFO firms. Next we describe the sample selection process (section 4), after which we present the results of our statistical tests on the liquidation differences between LIFO and FIFO firms (section 5). Section 6 provides the results of our study of factors which seem to account for the observed LIFO liquidations.

Although numerous inventory order quantity models have been pro- posed, few have incorporated inflation and taxes, two pervasive facets of our economic environment. When inventory costs are changing, optimal order quantity decisions will be influenced by tax incentives which depend on firms' cost flow assumptions. Optimal cost flow choices, in turn, will depend on ordering policies. Thus, firms have incentives to optimize simultaneously over ordering policies and cost flow assumptions. This study is the first to examine this joint optimization. Using a model developed in Biddle and Martin [1983b], results are obtained regarding the forms of optimal ordering policies under alternative cost flow as- sumptions, the patterns of year-end inventory levels, the determinants of accounting method choices, and the relative profit implications of costing method and ordering policy decisions. The three most widely used cost flow assumptions are the last-in, first- out (LIFO), first-in, first-out (FIFO), and weighted average cost (AC) methods.1 Of the 600 firms surveyed in the latest edition of Accounting Trends and Techniques (AICPA [1983]), 407 were using LIFO, 373 FIFO, and 238 AC to some extent (in their financial reports). The magnitudes of the cash flow implications of choices among these methods have been documented in recent years as a number of firms have switched from FIFO and AC to LIFO in response to available tax savings. (For estimates of these savings see, e.g., Halperin [1977], Biddle [1980], Biddle and Lindahl [1982], and Copeland and Weston [1983, p. 25].) To aid costing method choices, Sunder [1976a; 1976b] developed pres- ent value expressions for differences between the after-tax cash flows obtained under LIFO and FIFO. However, since Sunder's approach assumed that inventory levels are exogenously determined and equal under LIFO and FIFO, it does not allow for possible differences in ordering policies conditional on the choice of costing method; nor does it provide optimal ordering quantities once a particular method is chosen. It has long been recognized that LIFO provides incentives to avoid year-end inventory liquidations (see, e.g., Butters [1949], Levy [1974], Jannis and Johnson [1975], Halperin [1979] and Foss, Fromm, and Rottenberg [1980]). Biddle [1980] presented empirical evidence consist- ent with this notion by observing that 105 NYSE firms held larger inventories relative to sales after adopting LIFO when compared with similar non-LIFO firms. Because ordering policies affect the profits realized under alternative costing methods, they must be treated endogenously in costing method choices. A first step is the introduction of accounting tax incentives into order quantity models. Halperin [1977] developed a discounted present value expression for total inventory costs under LIFO for the determin- istic EOQ model. Cohen and Halperin [1980] subsequently developed a deterministic dynamic lot size model for LIFO, providing optimal order quantities over a multiyear horizon. Thus far, three stochastic models have been proposed. Cohen and Pekelman [1979], following Sunder [1976b], related LIFO layer age distributions to random walk ladder height processes. By restricting the set of ordering policies to base stock policies,2 they obtained a closed- form expression for discounted after-tax profits under LIFO (and FIFO). However, their expression was quite complex and concavity could not be shown. Biddle and Martin [1983a] developed a more general dynamic programming model in which the optimal order quantity was made a function of the LIFO layer cost configuration of each year's beginning inventory. Under reasonable conditions the optimal return function was shown to be concave. This study employs an alternative stochastic model (developed in Biddle and Martin [1983b]) which offers important advantages over those proposed by Cohen and Pekelman [1979] and Biddle and Martin [1983a]. Rather than optimizing with respect to a single order-up-to- level determined at the start of each year, the model permits a second order at year-end. This achieves greater descriptive validity by allowing intrayear (as well as interyear) cost changes and additional purchases after demand has been assessed. More important, there is a greater sensitivity to the effects of tax incentives on year-end procurement decisions and a smaller likelihood of year-end stockouts (which unreal- istically affect tax incentives by drastically altering inventory cost struc- tures). As a result, the model is uniquely suited for an examination of optimal choices among alternative inventory costing methods and the optimal ordering policies under each. In this study we extend the model to include not only LIFO and FIFO but also AC, a method not previously considered in order quantity models. In section 2 we present the model's notation and assumptions. Optimal choices between LIFO, FIFO, and AC are examined in section 3. We show that for any ordering policy and year-end inventory pattern (in- cluding liquidations), discounted expected profits under LIFO, FIFO, and -AC follow a strict ranking which depends only on changes in inventory costs. This ranking holds even when optimal ordering policies under each method are employed, taxes are increasing, and inventory costs are changing in a nonmonotonic fashion. Optimal ordering policies under LIFO, FIFO, and AC are then exam- ined in section 4. Differences in the forms of these policies hold important implications for managerial behavior. For example, under FIFO a base stock policy is optimal (see n. 2). In contrast, optimal order-up-to-levels under LIFO and AC vary with beginning inventory levels. The analysis also provides a counterintuitive result not obtainable from previous models, that is, optimal year-end inventory levels under LIFO may actually be less than those under FIFO and AC even when costs are increasing. Lower inflation, higher year-end premiums, and smaller be- ginning inventories increase the likelihood of this result. Simulations presented in section 5 illustrate the model's potential for aiding and explaining inventory management decisions. They indicate, for example, that the relative profit advantages of LIFO can increase with demand variance. The simulations also reveal the relative profit increments separately attributable to costing method and ordering policy choices. Section 5 provides empirical implications by identifying how various observable factors can potentially affect the discounted expected profits realized under LIFO, FIFO, and AC. Section 6 presents some concluding remarks.

Considerable attention in the accounting literature is devoted to dis- cussing possible motives for and effects of choices among alternative accounting methods. Although insights have been gained into investor (stock market) reactions to changes in accounting methods, little is known about either management motivations for observed accounting choices or the effects of these choices on subsequent operating decisions. One explanation for the dearth of empirical findings is that few of the proposed theories of accounting choice have offered strong links to real economic incentives. In addition, previous studies have concentrated on firm characteristics existing before and concurrent with the accounting choice. Since managers making accounting choices are likely to consider future conditions as well, this approach may overlook important deter- minants of accounting choices. This approach also precludes the detection of possible changes in operating policies induced by accounting choices. A choice among alternative inventory costing methods, especially between the last in, first out (LIFO) and first in, first out (FIFO) cost- flow assumptions, can generate potentially large changes in a firm's cash flows due to its impact on taxable earnings. These cash-flow effects provide economic incentives for choices between these methods. More- over, these cash-flow effects depend, in part, on the behavior of year-end physical inventories. Because year-end inventory levels are subject to management control, the LIFO-FIFO choice can both affect and be affected by subsequent inventory management policies. The LIFO-FIFO choice offers, therefore, the opportunity both to identify factors which influence accounting choices and to examine associations between these choices and subsequent operating decisions. This study investigates whether associations consistent with LIFO- FIFO tax incentives exist between management choices to adopt or not adopt the LIFO inventory costing method and characteristics of firms' year-end inventories. Both pre- and postchoice characteristics are ex- amined. Because the LIFO-FIFO choice is voluntary, a postchoice association would be consistent with managers both anticipating future inventory characteristics when making a LIFO-FIFO choice and chang- ing inventory management policies in response to that choice. Evidence consistent with the hypothesis that LIFO adoptions are associated with changes in inventory management policies would have important macroeconomic implications. Zarnowitz and Moore [1977] have argued that a failure to recognize the major shift in inventory costing methods which occurred in 1973 and 1974 (primarily FIFO to LIFO) resulted in an underestimation of inventory accumulations by the U.S. Department of Commerce. This underestimation resulted from the different proce- dures used under LIFO and FIFO to assign costs to inventory units. While this effect of LIFO-FIFO choices can bias macroeconomic mea- surements and forecasts, an associated change in inventory management policies by a large number of firms could directly affect underlying macroeconomic stocks and flows. The research design involves comparisons between a treatment group sample of firms which adopted LIFO and a matched pair control sample of firms which continued to use FIFO (or an average cost method). Two primary measures of inventory properties are derived from available financial statement disclosures. One measure approximates physical in- ventory levels, while the other estimates what the difference would have been each year between each firm's cost-of-goods-sold computed under the LIFO and FIFO alternatives. The empirical results based on these measures indicate statistically significant associations between LIFO adoption decisions and inventory properties and reveal striking cash-flow consequences associated with LIFO-FIFO choices.

Accounting research on choices of inventory valuation methods has focused on various consequences of two extreme methods: LIFO and FIFO. The main consequence studies relate to effects of the differences in taxes payable between the two methods on security prices. However, tax consequences appear to provide an incomplete explanation for managerial decisions to stay on FIFO or to change to LIFO. In this paper, I provide an analysis of another possible incentive of executives to stay on FIFO despite the apparent tax advantages of switching to LIFO.1 In

Biddle's study is an attempt to demonstrate empirically an association between LIFO adoption decisions and properties of year-end inventory levels. My comments on the study are divided into five main topics. In the first part, I examine the hypotheses generated. The second part consists of a discussion of the model used to generate these hypotheses. In the third part, I examine the experimental design, and in the fourth, I discuss the test results. I conclude by giving some suggestions for further tests of the hypotheses.

To provide a frame of reference, I should preface my comments with several observations on my personal perspective with respect to the review of Mr. Biddle's study.-First, as a business executive, I have had direct experience with the considerations involved in making a choice between the use of LIFO or FIFO inventory valuation methods. I might add that this involvement has also necessitated the ongoing review and analysis of the impact of such an election. Second, I am personally interested in the economic consequences of accounting choices, both from the standpoint of a businessman and from the viewpoint of a participant in the standard-setting process. Finally, I should note that I do not have the perspective of a statistician experienced in sample structure and mathematical analyses. As a result, my comments are directed toward the approach taken in the study and the possible significance of the results, rather than to the statistical methodology employed. I conclude with a discussion of areas in which future research may be useful.

Much of the discussion focused on the expected stock market reaction to the announcement of a LIFO liquidation. In particular, the issue of tax consequences arose, and the question was asked whether the timing of a LIFO liquidation might be casually linked with the availability to a firm of tax items such as net operating loss carryforwards or unused investment tax credits. The link would depend on the particular circumstances of a firm having tax items available. For example, if the firm makes use of net operating loss carryforwards that would otherwise not have been used for one or more years, then the firm is able to postpone the LIFO liquidation tax liability, and the present value of the tax cost is lowered (assuming constant tax rates). If a firm's existing tax carryforwards are sufficient to offset fully the LIFO liquidation tax cost and would have otherwise expired unused, the marginal tax rate on the liquidation-induced taxable earnings will be zero. A tax-induced market reaction becomes difficult to predict when, for instance, tax rates will change, or when the market's expectations reflect a nonzero probability of a LIFO liquidation. In the former case, a currentperiod inventory liquidation might reduce the likelihood of a future one.

Evidence exists indicating that annual earnings numbers convey information to investors that is relevant to the pricing of common stocks for publicly traded companies (e.g., Ball and Brown [1968] and Beaver, Clarke and Wright [1979]). However, comparatively little is known about the information content of the more comprehensive disclosures regarding earnings and earnings components that are provided in the complete text of a firm's financial statements and the accompanying footnotes.' These

The decision to select one of the two accounting methods for inventory valuation-first-in, first-out (FIFO)' and last-in, first-out (LIFO) involves consideration of several factors of which potential tax effect of the accounting method is often the most important.2 While the general relationships of inflation, inventory levels, rate of taxation, etc., to the potential tax savings have been well understood for a long time, realistic decision models which quantify the impact of alternative accounting procedures on the economic value of the firm are conspicuously absent. It is difficult, then, for a manager who is weighing the choice of the two accounting methods explicitly to determine the impact of such a decision on the value of the firm. The purpose of this study is to present procedures to estimate the difference between the net present value of tax payments under the two inventory valuation methods. Sunder [1976] developed a simple estimation procedure under conditions of certainty and level year-end inventories. Three extensions are presented here by relaxing various assumptions. First, the assumption of level year-end inventories is relaxed to compute the present value of cash-flow differences with changing inventory level within a deterministic framework. In a second extension, deterministic

While few, if any, early accounting studies were based on psychological theories of behavior, Birnberg and Nath [1967] conjectured that the functional fixation hypothesis proposed by Ijiri, Jaedicke, and Knight [1966] could lie at the heart of studies concerning alternative accounting techniques (e.g., LIFO and FIFO). From that point, until Ashton's recent paper at the 1976 Empirical Research Conference, only one study (Dopuch and Ronen [1973]) attempted to deal explicitly with the phenomenon. Their results confirmed the earlier conjectures. In this paper, we will reexamine the underpinnings of the functional fixity problem and offer some additional evidence concerning the existence of a fixity phenomenon. A final section will outline the implication for future research.

Between June 1974 and May 1975 over 400 firms on the New York and American stock exchanges adopted the last-in, first-out (LIFO) inventory costing method. Responding to the first round of double-digit inflation since WWII, these firms were able to assign their newer and thus higher inventory costs to units sold, thereby realizing substantial tax savings.1 However, Ricks [1982] found that stock returns near the earnings disclosure dates of 1974 LIFO adopters were negative and significantly lower than returns near the earnings disclosure dates of firms not using LIFO. Given that firms adopting LIFO in 1974 were voluntarily switching to an accounting method providing often significant tax savings, it is not obvious why investors would have reacted negatively. Nor is it likely that investors were unaware of many of the firms' LIFO adoption decisions; Stevenson [1987] found that at least two-thirds of the firms in Ricks' sample had disclosed their LIFO adoptions prior to their earnings disclo- sure dates. If investors were reacting to the effects of the LIFO adoptions, why did they react negatively given the increased cash flows and prior disclosures? If investors were not reacting to the effects of the LIFO adoptions, what caused the significant negative excess returns near the earnings disclosure dates of firms adopting LIFO in 1974? This study presents evidence suggesting that the negative excess re- turns observed by Ricks [1982] were associated with, and possibly due to, analysts' systematic overestimates of earnings of firms adopting LIFO in 1974. We are aware of no previous study documenting systematic errors in analysts' earnings forecasts conditional on a voluntary account- ing method change.' As a first step, we assess the generality of Ricks' finding by examining excess returns near the preliminary earnings disclosure dates (hereafter, prelim. dates) of all New York (NYSE) and American Stock Exchange (AMEX) firms adopting LIFO between 1973 and 1980. The results confirm negative excess returns around the prelim. dates of firms adopt- ing LIFO in 1974. There is little evidence of significant excess returns (negative or positive) near the prelim. dates of firms adopting LIFO in other years. To explain the negative returns of the 1974 adopters, we examine the association between excess returns and earnings forecast errors. We find that Standard & Poor's analysts systematically overestimated the earn- ings of these firms. Analyst forecast errors are found to be significantly correlated with both the excess returns around the prelim. dates and the earnings reductions due to the LIFO adoptions. Finally, we offer two possible explanations for the negative excess returns and the positive correlations between excess returns, analyst earnings forecast errors, and earnings effects of the 1974 LIFO adoptions. First, the limited availability of information regarding the likely earnings effects of the 1974 LIFO adoptions, combined with the relative novelty of this accounting change, may have led analysts and investors to underestimate the LIFO earnings effects and may have made it difficult for the market to discern immediately the cause of the resulting negative forecast errors. Evidence consistent with this "limited information" hypothesis is found in the predictions of Value Line analysts of LIFO earnings effects. The actual mean earnings-per-share effect of the LIFO adoptions ($.947) was 62% higher than the average Value Line estimate ($.584). However, tests for the stock price reversals which would be implied by the limited information explanation are inconclusive. A second possible explanation is that the magnitudes of LIFO earnings effects were caused by and/or provided implications regarding another factor which was important in valuing the firms, yet difficult for analysts and investors to predict. Inflation is suggested as the likely candidate. Since LIFO tax savings are a function of the rate at which inventory costs are increasing, firms experiencing higher rates of inflation would have greater incentives to adopt LIFO.4 Thus, the self-selected sample of firms which adopted LIFO in 1974 may include firms most affected by that year's surge of inflation. Since the extent to which inventory costs have increased is reflected in the magnitudes of the earnings effects of LIFO adoptions, the negative analyst forecast errors and the positive correlation between these forecast errors and the earnings effects of LIFO adoptions could be explained by analyst underestimates of the rates of inflation experienced by these firms. Evidence consistent with this "inflation" hypothesis is found in the fact that managers of firms adopting LIFO underestimated the impact of inflation on their own firms. In addition, negative excess returns and earnings overestimates are observed for a sample of established LIFO users, consistent with analysts and investors being surprised by the impact of inflation on these firms as well. However, aggregate tests based on cross-industry correlations between changes in wholesale prices and LIFO-related variables are generally inconclusive.

Many firms have adopted the last in, first out (LIFO) method of accounting for inventories in order to reduce taxable income and income tax liability in periods of rising prices. Lower tax liabilities under LIFO are not assured, however, since a LIFO company's "cost of goods sold" could include some very low costs from prior years, if that company's inventory purchases or production for the current year are insufficient to meet the year's demand. Moreover, even if the use of LIFO does result in a lower tax liability, the inclusion of any of the old, low costs in cost of goods sold under LIFO means that the tax liability could have been reduced even further by increasing purchases or production, such that these old costs would have remained in inventory. Increasing production or purchases, however, increases the cost of holding these extra units until the time at which they are depleted from inventory to meet demand. Such costs might include warehousing costs, interest costs, and the opportunity cost of having less funds available for other productive uses. Thus, when prices are rising, a trade-off exists between lower tax costs and higher holding costs associated with more production or purchases. This cost trade-off is further complicated by the fact that there may be a price or cost benefit to increasing purchases or production prior to the end of the time period at which tax liability is computed. In general, when

The introduction of a new product or production process often results in a distinct start-up phenomenon; as production and engineering personnel' gain experience with the innovation, persistent gains in productivity are achieved over periods as long as three years after initiation of manufacture. These productivity increases have implications for capital budgeting and project evaluation, budgeting sales and production operations, evaluation and control of operating costs, and inventory valuation and income determination. In this paper we argue that explicit recognition and estimation of the entire productivity path of a start-up will result in more effective accounting, whereas the application of a constant cost or productivity standard may result in misleading internal and external reporting. Our discussion is based on the "learning model" which has adherents in many diverse industries. Discussion of the accounting implications of start-ups and the use of the "learning curve model" is not unprecedented; both subjects have been explored previously by accountants in the airframe and electronics industries.' But this literature is deficient in several ways: (1) It considers only new product start-ups in labor-intensive, discontinuous forms

The traditional literature of financial statement analysis often emphasizes the desirability of adjusting the firm's financial ratios to predetermined targets which are usually based on industry-wide averages.' This study considers the question of whether there is empirical evidence to suggest that firms do adjust their financial ratios to such targets. The results of the test, using the partial adjustment model, indicate that financial ratios are periodically adjusted to their industry means. One way management can adjust the financial ratios to predetermined targets is to choose from the set of generally accepted accounting measurement rules (e.g., inventory valuation methods) those which affect the financial ratios in the desired direction. This technique-in the context of the income smoothing hypothesis-was proposed by Gordon as a means of maximizing management's utility

In inventory valuation, the practice of foregoing historical cost values in favor of some other values is an old one; the accounts of a certain Francesco di Marco of Prato for the year 1406 contained a write-down of inventories "because we no longer value them as above since their price has gone down." 1 Currently, a number of Dutch financial statements contain inventory values higher than cost because these statements reflect replacement value and were prepared following periods of rising prices. Between downward and upward adjustments from historical cost lies the rule of "cost or market whichever is lower." This rule has wide acceptance despite open opposition to it. Quoting from a 1939 publication

The research reported herein was undertaken to investigate relationships between (a) security evaluation and portfolio selection and (b) alternative inventory valuation and depreciation methods in financial reporting. First, a computer simulation model of a manufacturing firm was developed. In this simulation phase, the effects of alternative methods of financial statements and related measures (earnings per share, working capital, earnings margin, current ratio, inventory turnover, and other financial ratios along with corresponding rates of change and moving averages) were investigated under a wide range of operating conditions. In the second phase, an attempt was made to measure the effects of these accounting variations on evaluations by professional security analysts. Complete prospectuses were developed for two hypothetical companies named ETX Electronic Industries, Inc. and Rayco Electronics Corporation. Financial data for the companies were generated by the computer model. Four sets of financial reports with their related measures were generated for each company for a ten-year span. Two inventory methods (lifo and fifo) and two depreciation methods (straight-line and accelerated) were used. The four different financial reports for each company resulted m sixteen combinations of financial reports for the two companies. Participants in the study received an "information packet" which included an introductory letter, a return questionnaire, and one prospectus for each of the companies to be evaluated. The only variation in the in

This is the second occasion on which I have been asked to review a work of Carl Devine's. The first occurred 24 years ago when I reviewed his book on Inventory Valuation and Periodic Income for the American Economic Review-an assignment which enabled me to discover at an early stage his now widely known capacity for fine scholarship. The present work is as different from the first as the intervening years and the broadening interests of scholars in the field would suggest-perhaps more so. It deals with professional attitudes and endeavors to find background variables that explain or at least predict these attitudes. If identified with assurance, these variables, the paper suggests, might be used in selection of future professional accountants. Insofar as evidence adduced through such procedures as this one indicates that courses intended to inculcate professional attitudes or to improve them have failed to do so, educational methods might be changed. Similarly, educational efforts of the organized profession directed toward its own members might be changed. The effectiveness of any scientific exploration depends upon asking the right questions, upon setting up the relevant hypotheses. The hypotheses are the essence of Devine's paper, in the opinion of this reviewer, and this discussion will be concerned chiefly with them. This discussion also includes a suggestion for another approach to the subject than the one Professor Devine adopted when he chose background variables as the bases for his hypotheses. I will deal with the problem of the hypotheses on two levels: first, on the level adopted by Professor Devine, where the procedure shall be to indicate the questionable validity of specific hypotheses as, indeed, he did himself to some degree; and, second, on a more fundamental level, by considering, from an admittedly inexpert viewpoint, the validity of the use of background variables at all.

The Wall Street Journal (WSJ) in its Earnings Digest column and in feature stories annually summarizes the financial results of about 4,500 companies, including their changes in accounting methods.' Information about such changes is obtained by "scrutinizing the company's report." 2 Thus, changes in accounting methods, which constitute inconsistencies,3 are disclosed in the WSJ even if they were not sufficient to cause the company's auditor to issue a qualified opinion. I surveyed the Earnings Digest section of the WSJ for the period from January 1, 1967 through December 31, 1969. This period may be taken as an approximate surrogate for the calendar reporting years 1967, 1968, and 1969, though there is generally a lag of one to four months between the close of a company's fiscal year and the issuing of its annual report. All reported changes in accounting made by "industries" (see below) were noted and broken down into several major categories. Whether the change was made retroactive to the prior year was also noted. A comparison of absolute number of changes, between years and by categories, is summarized in Table 1. A relative comparison of categories of change, as a percentage of total changes in each year, is also summarized in that table. Of the 4,500 companies in the "universe," I would estimate that some 2,500 are industrials (that is, companies other than utilities, regulated

In their comments on my price-level paper,' Professors Chambers, Moonitz, and Winborne2 seem to have misunderstood my position on accounting for price changes as represented in my current valuation system. To a large extent the differences between us are differences of definition although Chambers also charges me with an error of logic and attacks the averaging procedures which I (and others) have used in measuring income and financial position. Because accounting information is needed for purposes of economic evaluation and decision-making, I suggested that financial statements should incorporate current value data which can assist in these tasks. The first and main requirement is that all items in the income statement for a particular period, and in the balance sheet at the end of the period, should be valued in terms of current prices of the period. In my current valuation system, I interpret current prices as pertaining to particular goods and services which are recorded in the financial statements in question, and not the prices of a hypothetical collection of goods and services which are valued by reference to a general purchasing power index. My income concept is restricted to the increment in wealth resulting from productive activities as represented by actual transactions, with both revenue and cost flows expressed in current prices. Income is not derived by reference to the change in balance sheet measures of generalized purchasing power between two points of

In recent years there have appeared reports of the comparative effects of using other than historical costs in preparing financial statements.' Generally, these papers described a comparison of the application of merely a single alternative income model against the application of generally accepted accounting principles (GAAP) to the same situation. As a result, very little information is available on comparisons of the applica

Net realizable value as a basis for inventory valuation has been advocated and adopted in practice under the following circumstances: 1. When estimates of cost are arbitrary and a ready market exists for the goods, as is typically the case in the extractive industries; 2. When joint products are prevalent, as in the meat packing industry, and 3. When the expected proceeds from the sale of the inventory items are below historical costs because of price changes, damage, or obsolescence. These circumstances represent cases in which valuation on a historical cost basis breaks down either because of the infeasibility of determining meaningful cost estimates or because historical cost overstates expected value. However, beginning with Canning,1 many have advocated using net realizable value for inventory valuation on its own merits and not just because of the shortcomings of the historical cost basis. When future cash flows could be estimated or approximated, Canning favored "direct valuation of inventories," his terminology for net realizable value. According to Canning

Since about 1950, the controversy over full vs. direct costing has generated a voluminous literature' largely aimed at the relative merits of each (1) for external financial reporting purposes or (2) for internal use in management planning and control.2 The arguments rest on the assumption that each method produces different results; obviously if each method produced exactly the same results then controversy is idle. The analyses to date have largely ignored the effect of the inventory costing method. A typical discussion of the generalized conditions which create differences between direct and full cost profit is: "When production exceeds sales (i.e., in-process and finished inventories increasing), absorp

The title of Sunder's paper refers to both "problems of measurement and some empirical evidence" bearing on the "relationship between accounting changes and stock prices." Largely because of these problems of measurement, the empirical evidence concerning abnormal price changes associated with changes in inventory costing assumptions is generally unconvincing. On the other hand, the empirical evidence concerning the shift in the relative risk of firms before and after such changes appears to be considerably more convincing and potentially quite important for future research involving the use of the market model for describing the normal relationship between returns on individual stocks and an index of overall market performance. While some of my comments will be specifically related to Sunder's research, much of the thrust of my remarks will be directed at the broader question of the utility of this approach to measuring the impact of accounting information.

INVENTORIES HAVE PLAYED A CRUCIAL ROLE in U.S. business cycles, for their perverse behavior has acted to magnify rather than dampen cyclical swings in demand. Indeed, fluctuations in inventory accumulation accounted on the average for 75 percent of the decline in real gross national product (GNP) experienced in the four recessions between 1948 and 1961. In addition, the erratic short-term behavior of inventory accumulationtypically representing half of the variation in quarterly GNP growthcreates a severe forecasting problem. The topic is of current interest for two reasons. First, although a reduced rate of inventory accumulation contributed to the slowdown in the first half of 1970, the magnitude of the swing in inventory investment has been modest by comparison with those in similar postwar periods. Thus current developments stand in sharp contrast with experience as recent as 1966, when the annual rate of inventory accumulation reached $18.6 billion in the fourth quarter with a major reversal down to $4.4 billion by the second quarter of 1967.1 Do the two periods differ because the current slackening in sales was anticipated to an extent that allowed production to adjust smoothly? Or must alternative explanations be sought for the current stability of inventory investment? Second, just as inventories have had a major influence during the contractionary phase of past cycles, they have been an important source of

Inventory valuation is usually considered as a problem of pricing.' All accounting data represent quantities multiplied by prices, and pricing is thus only a part of the inventory valuation problem. In this paper, attention will be directed to some less frequently considered difficulties of quantification and their effects on the pricing problem. It will be argued that valuation can only be approached from a study of the nonmonetary characteristics of inventory, i.e., the physical or time quantities of materials, labour and overheads and the position of these quantities in relation to the business operating cycle. Inventory valuation serves period accounting: the old period hands over its "values in suspense" to the new one. In this way, the old period is purged of matters which do not affect its results, and the new period is made to account for the values transferred to it. Considering the problem in this light may show how some conventions of accounting practice arise out of business conditions, and must therefore be applied differently in different firms.

In this paper we report some additional evidence regarding the effects of alternative inventory accounting techniques on the decisions of laboratory subjects. Briefly, we have attempted to incorporate various factors within the experimental design which may aid us in interpreting the findings from such laboratory experiments on the effects of alternative accounting techniques. At the outset, we wish to clarify our views about the entire issue of how accounting procedures may affect resource allocation decisions. First, there is an impressive body of evidence supporting the efficient market hypothesis in the assessment of the impact of new information on the prices of securities in capital markets.1 At the aggregate level, there is little reason to believe that the market is "fooled" by different accounting methods.2 However, the available evidence on the efficient market implies nothing about the ability of individual decision-makers, such as managers and credit officers, to adjust accounting reports across different valuation techniques. Individual decision-makers are our only concern. Granting this orientation, there is still some doubt about the implications of the findings of experimental studies like ours concerning the whole issue of alternative accounting techniques. As an example, suppose subjects-especially students-favor as an investment a firm which reports