Guide: Instructions for case analysis or articles 1. Summary: what is a main concept in the case or article?a. When you read this article you may be able to:i. Identify and differentiate the concepts fixed and variable costs.ii. Distinguish between value-adding and non-value-adding activities and costs.iii. Apply the appropriate method of analyzing cost behavior to a situationSituations that arise in the case or article.2. Possible solutions to such situations (applying the lesson of the day).3. Select one possible solution to the case. Explanation for how you select this solution (the best alternative is not always feasible to resolve the central or primary situation presented in the case or article).4. Reflexive analysis on developments in the case or article. You can use other references that approach the situation raised in this case or article from other perspectives.5. Conclusion or recommendation.6. If the case has questions, answer them.

Writea 700- to 1050-word paper in which analyses the concepts expose in the article.

FormatsuggestedAPA guidelines.

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0951-3574.htm Measuring general managers' performances Market, accounting and combination-of-measures systems General managers' performances 893 Kenneth A. Merchant Leventhal School of Accounting, University of Southern California, Los Angeles, California, USA Abstract Purpose - This paper discusses how to choose a measure or set of measures for the purposes of evaluating and rewarding general managers' performances. Design/methodology/approach - The paper describes a set of criteria that is useful for evaluating any measure or set of measures. Then it applies the criteria to an evaluation of three measurement alternatives in common use at general management organization levels: market measures, accounting measures, and combinations of measures. Findings - The paper shows that all of the measurement alternatives fail to satisfy one or more of the evaluation criteria and, hence, lead to less than optimal outcomes. But it also shows that some alternatives are better than others in specic situations. Originality/value - While comprehensive sets of evaluation criteria have been applied to nancial accounting choice issues, this is the rst such approach in management accounting. This approach can lead to improved performance measurement system choices. It can also be used to guide future research because the analysis also reveals major gaps in our knowledge about the qualities of performance measures in common use. Keywords Measurement, Performance, General management, Market yield, Accounting systems, Balanced scorecard Paper type General review A commonly cited management axiom is \"What you measure is what you get.\" This axiom works in practice because performance measures are linked to any of a number of incentives, both extrinsic and intrinsic, that employees value or penalties that they wish to avoid. People respond to these incentives/penalties. The measures, then, play valuable motivational, or decision inuencing, roles. But what performance measure (or measures) should be used? At general manager levels of prot-seeking rms - both at the corporate and operating unit levels - the organizational role that is the focus of this paper, the managers' responsibilities are both broad and varied[1]. Reecting that variety, we see that the list of measures used to motivate and evaluate general managers' performances is long. These measures can be Helpful comments were provided by numerous participants at the 4th Asia Pacic Interdisciplinary Research Conference (APIRA), held in Singapore, the International Management Accounting Conference (IMAC) III, held in Bangi, Malaysia, and the 2nd Advances in Management Accounting (AIMA) Conference, held in Monterey, California, where this paper provided the basis for plenary talks, and by Mark DeFond, Trevor Hopper, Lee Parker, Tatiana Sandino, K.R. Subramanyam, Wim Van der Stede, and Mark Young. Accounting, Auditing & Accountability Journal Vol. 19 No. 6, 2006 pp. 893-917 q Emerald Group Publishing Limited 0951-3574 DOI 10.1108/09513570610709917 AAAJ 19,6 894 classied into three broad categories. Two of these categories include single-number, summary measures of performance. One includes market measures; that is, those that reect changes in stock prices or shareholder returns. A second includes summary accounting-based measures, which can be dened in either residual terms (e.g. net income after taxes, operating prot, residual income, EVA) or ratio terms (e.g. ROI, ROE, RONA). The third broad measurement category includes combinations of measures. These combinations can involve either the use of both market and accounting measures or the use of one or both types of summary measures plus some disaggregated nancial measures (e.g. revenues, expenses) and/or non-nancial measures (e.g. market share, sales growth, inventory turnover, customer satisfaction). Looking at these alternatives suggests a number of questions, such as: Does it matter what measure(s) rms use? Can all of these approaches be correct? Does one or more have advantages either in general or in specic situations? Are measurement problems an important cause of some or all of the major nancial reporting frauds, such as occurred at Enron, WorldCom, or Parmalat? Are they a major cause of the seemingly persistent problem of management myopia (excessively short-term oriented behavior)? Why should a rm choose a market measure rather than an accounting-based measure? When should a rm use a single, summary measure of performance, either market- or accounting-based, rather than some combination of measures? Addressing these questions is the focus of this paper. The paper begins by describing generic criteria that are useful in evaluating any measure or set of measures. Then it discusses each of the three measurement alternatives described above - market measures, accounting measures, and combinations of measures - and critiques them by comparing them with the evaluation criteria. Not surprisingly, none of the measurement alternatives provides a perfect solution. All fail to satisfy one or more of the criteria, although some are better than others in specic situations. The paper concludes by discussing some management and research implications of this analysis. Criteria for evaluating measurement alternatives For motivational (decision-inuencing) purposes, a measure or a combination of measures should have all of the following qualities[2]. It should be: . congruent with the organization's objectives; . controllable by the manager whose behaviors are being inuenced; . timely; . accurate; . understandable, and . cost effective to produce. The following sections describe the importance of each of these qualities. Congruence For motivational purposes, congruence with the organization's objectives may be the single most important measurement quality[3]. A measure should go up when good actions are taken and, hence, the organization's objectives are most likely to be served, and go down when bad actions are taken. If a measure does not reect progress toward the desired ends, or if it does so incompletely, then motivating managers to work to improve the measure will be unproductive, no matter how well the measure satises the other measurement criteria (Otley, 1999). What is the objective of prot-seeking organizations? Most people, at least those in developed countries, have come to agree that the primary objective of these organizations is to maximize shareholder (or owner) value (e.g. Jensen, 2001; Society of Management Accountants of Canada, 1999; Wenner and LeBer, 1989; Treynor, 1981). As Jensen (2001, p. 11) phrases it, \"200 years' worth of work in economics and nance indicate that social welfare is maximized when all rms in an economy attempt to maximize their own total rm value.\" Ideally, then, to reect success properly, performance measures should go up when value is created and go down when it is destroyed[4]. What is value? Value is perhaps the most basic concept in business. The value of any economic asset, such as a business entity, can be calculated at any specic time by discounting to present value the future cash ows that the rm is expected to generate (e.g. Brealey and Myers, 2002). The values of some economic assets are also reected in the amounts for which the assets were sold in recent market transactions. The current value of a barrel of oil is reected in the values of recent oil sales. Similarly, the value of a publicly traded business entity is reected in the value of recent stock sales. Correlation statistics can be used to assess the extent to which a measure is congruent with value changes. A performance measure is highly congruent if it is highly correlated with changes in value. Most academics agree with this simple test of congruence. For example, Garvey and Milborn (2000, p. 210) concluded, \"The simple correlation between EVA or earnings and stock returns is a reasonably reliable guide to its value as an incentive contracting tool.\" Use of an incongruent measure can be counterproductive, actually motivating managers to do the wrong things (e.g. Kerr, 1975). For example, when managers are held accountable only for short-term prots, which is an incongruent measure of long-term value maximization, they are prone to engage in excessively short-term oriented (myopic) behaviors. They might, for example, cut research and development investments to boost short-term prots even at the expense of the long-term (e.g. Seglin, 2003; Jacobs, 1992; Merchant, 1990). Or if managers are held accountable only for increased sales, they might increase sales in some value-destroying ways, such as by extending the payment terms beyond what is protable, by selling to customers with questionable credit, or even by selling items that the customers did not need, as happened at Sears' Auto Centers (Kelly and Schine, 1992). Controllability While congruence is a necessary measurement quality to motivate managers effectively, it is not sufcient. For one thing, even a measure that is perfectly congruent with the organization's objectives will not motivate the right behaviors if the manager has no control over that measure. Controllability means that those whose behaviors are being inuenced are able to affect the measure in a material way in a given time span. If managers cannot signicantly inuence a measure, that measure does not provide information about the desirability of the actions that the managers took. These measures are said to lack \"informativeness\" (e.g. Banker and Datar, 1989; Feltham and Xie, 1994). If a results area is totally uncontrollable, the performance measures provide General managers' performances 895 AAAJ 19,6 896 no information about the desirability of actions that were taken. Partial controllability makes it difcult to infer from the results whether or not good actions were taken. Most measures are not inherently controllable or uncontrollable. Controllability depends on the authority given to the individual. Sales revenue, for example, might be highly controllable from the perspective of the sales and general managers of an organization, but it is largely uncontrollable from the perspective of the receptionists and janitors, and maybe even the plant managers, in that organization. Controllability typically also depends on the situation being faced. In some businesses, even sales and margins might be largely uncontrollable. They might be largely determined by competitive pressures or economic conditions. Timeliness Timeliness refers to the lag between the managers' actions and the measurement/feedback of results (and provision of incentives). Timeliness is a critical element in all motivational theories that include an element of feedback, such as path-goal theory (Locke and Latham, 1990) and self-efcacy theory (Bandura, 1977). Timely feedback and reward provide greater short-term performance pressure and stronger motivational reinforcement. All employees need consistent, short-term performance pressure to perform at their best (Merchant, 1989). The pressure helps ensure that they do not become lazy, sloppy, or wasteful. Short-term pressure can also stimulate creativity. It increases the likelihood that employees will be stimulated to search for new and better ways of improving their results. This is an application of the old adage: \"Necessity is the mother of invention.\" Measures and/or the incentives linked to them that are delayed for signicant periods of time, lose most of their motivational impact. Telling managers that their performance will be evaluated, for example, over the forthcoming ten-year period, with rewards assigned at the end of that evaluation period, will not provide much of an incentive. At the extreme, if the feedback and incentives are provided beyond the time the managers expect to serve in this job, the motivational effect will be zero. This is sometimes referred to as the \"horizon problem\" (e.g. Dechow and Sloan, 1991), and the horizon problem is a real concern at general manager organization levels. The average tenure of a large company CEO is approximately six years, and it is less than that for lower-level general managers (Fredman, 2003; Booz Allen Hamilton/Business Council of Australia, 2004). Thus, so providing incentives based on, say, a ten-year performance period will not affect the behavior of the average general manager. Accuracy Ideally, performance measures should be accurate. The term accuracy includes two concepts - precision and objectivity. Precision is the inverse of the variance in the performance measure or, in other words, the lack of \"noise\" in the measure (Banker and Datar, 1989). Precise measures can be expressed reliably within relatively narrow bounds. Ascertaining that prot for the period was $1.24 million is more precise than is being able to say only that prot was somewhere in the range of $1-3 million. But some performance qualities, such as corporate reputation, employee morale, or board (of directors) effectiveness, cannot be measured very precisely. If multiple measurers are asked to measure these qualities, they will likely arrive at signicantly different answers. Without precision, a measure can lose much of its information value. Imprecise measures can cause evaluators to misevaluate performance or to fail to distinguish among performances that vary materially. Employees will react negatively to the inequities that will inevitably arise when good performances are not recognized and rewarded. Objectivity, the other element of accuracy, means freedom from bias. Holding managers accountable for product quality and then allowing the managers to perform their own quality measurement without any oversight will lead to a high potential for bias. Managers have two main alternatives for increasing measurement objectivity. They can have the actual measuring done by people who are independent of the process, such as a quality control staff or outside consultants, or they can have the measurements veried by independent persons, such as auditors. Understandability Two aspects of understandability are important. Managers must understand what the measure reects; i.e. how the measure is calculated. But in addition, the managers must understand what they must do to inuence the measure, at least in broad terms. So, for example, it may not be sufcient to tell managers that they are being held accountable for, say, risk adjusted rate of return (RAROC) even if they know how that measure is calculated. The managers must also know, or be able to gure out, what can be done to improve RAROC. Cost effectiveness Measures are like all economic goods in that they provide benets but cost money to produce. Some performance measures are quite inexpensive to use for incentive purposes because they are already produced for other purposes. Examples are standard nancial reporting measures, like net income, and measures that are already generated for use in production areas of the business, such as throughput, labor efciency, and quality. Other measures, though, can be quite expensive. Customer satisfaction surveys can be expensive to administer, as are ratings provided by \"mystery shoppers.\" And development and implementation of complex, tailored measurement systems, such as balanced scorecards, can cost many millions of dollars in consultant fees and management time. All measures and systems of measures should only be used if the benets they provide exceed their costs. Evaluating the measurement alternatives for general managers The measurement criteria discussed above can be used for evaluating measurement alternatives for all levels and all roles in an organization. They will be used here to evaluate the measures that can be used to motivate general managers. Market measures Market measures of performance are based on changes in the market value of the entity being managed or, if dividends are also considered, returns to shareholders. Market measures have broad appeal largely because they provide direct indications of the amount of value that has been created or destroyed. Here is a representative expression of that conclusion by some nance academics: \"Any nancial performance measures used in managerial compensation . . . must be correlated highly with changes General managers' performances 897 AAAJ 19,6 in shareholder wealth\" (Bacidore et al., 1997, p. 11). Here is a similar conclusion in a practitioner-oriented publication (Rutledge, 1996, p. 131): Shareholders get paid when managers create equity value, not when managers check off items on to-do lists. To align manager interests with owner interests, pay managers the same way shareholders are paid. 898 This obvious measurement congruence allays political pressure that otherwise might be brought on the company by outsiders. Who is to complain if managers share rewards in direct proportion to those earned by the entity's owners? If the market value changes are measured in terms of recent transaction prices, as is common, the market measures also have other advantages. For publicly traded entities, market values are available on a timely basis. They are available on a daily, or even more frequent, basis. They are accurate. The values can be measured precisely[5], and the values are usually objective, not manipulable by the managers whose performances are being evaluated. They are understandable, at least in terms of what the measures represent. And they are cost effective, as they do not require any company measurement expense. With all these advantages of market measures, what is there not to like? Actually, market measures do have some severe limitations and problems. One is a severe feasibility constraint. Market measures are only readily available for the small minority of business entities that are publicly traded. Except in highly unusual cases, they are not available either for privately-held corporations or wholly owned subsidiaries or divisions, and they are not applicable to not-for-prot organizations. Second, market measures present two types of controllability problems. They can generally be inuenced to a signicant extent by only the top few managers in the entity, those who have the power to make decisions of major importance. They say little about the performances of individuals lower in the organizational hierarchy, even those with signicant general manager responsibilities, except in a collective sense. Thus, market measures provide meaningful indications about the individual performances only of the top management team. And even for the top management team, market measures may be far from being totally controllable. Market prices are affected by many factors that the managers cannot control, such as changes in macroeconomic activity, interest rates, factor prices, exchange rates, and the actions of competitors (e.g. Kim and Suh, 1993; Sloan, 1993). Research has shown that the proportion of stock price changes that can be \"explained\" by macroeconomic and competitive factors is high, in some markets and situations perhaps as high as 98 percent (e.g. Islam and Watanapalachaikul, 2002; Sloan, 1993). When this \"noise-to-performance signal\" ratio is high, stock prices do not provide much information about even top-level managers' performances. A second problem with market measures is that market values are not always reective of realized performance. Instead the values are heavily inuenced by future expectations (Barclay et al., 2003), and it is risky to pay bonuses based on expectations because those expectations might not be realized. Consider three examples: (1) In January 1993, the Eastman Kodak Company hired Christopher Steffen as its chief nancial ofcer. Mr Steffen was a key participant in turnarounds at Chrysler and Honeywell, and he was widely viewed as just what Kodak needed to trim its bloated bureaucracy. As a result, within two days of Steffen's appointment, Kodak's market value rose by about 17 percent ($2.2 billion). The headline in the Wall Street Journal was, \"Kodak obtains services of well known change agent: stock value jumps by two billion dollars.\" At the time, Business Week dubbed Steffen as the \"$2 billion man.\" Unfortunately for Kodak, Steffen resigned just 11 weeks later, claiming that company management had found his ideas \"too revolutionary.\" Kodak's market value quickly dropped by about $2 billion. (2) On July 19, 1996, Al Dunlap was hired as CEO of Sunbeam, a struggling small appliance maker. The hope was that Dunlap, whose nickname was \"Chainsaw Al\" because of his aggressive cost-cutting style, would have the same success in turning around Sunbeam that he had had at Scott Paper and other companies (Schifrin, 1996). On the day Dunlap was hired, Sunbeam stock jumped 49 percent, creating almost $500 million in shareholder value. Dunlap quickly implemented a major restructuring plan that involved, among other things, a 50 percent reduction in the company's work force, the closing of 18 of the company's 26 factories, and a nearly 90 percent reduction in the company product lines. By March 1998, the Sunbeam stock price had reached a peak that was over 300 percent higher than the July 1996 value. But had Sunbeam's true intrinsic value really increased? The stock slid down quickly when Sunbeam announced that earnings for the rst quarter of 1998 would fall below company estimates. In response, Dunlap quickly announced a new round of layoffs, 40 percent of the workforce, but the stock continued to fall. The fall accelerated when, on June 8, 1998, Barrons published an article criticizing Sunbeam's aggressive accounting practices (Laing, 1998). When Dunlap was red on June 15, 1998, Sunbeam stock price had fallen 70 percent from its high, which had occurred just three months earlier. (3) In May 1997, the Boston Celtics Limited Partnership, which owned the Boston Celtics basketball team, announced the hiring of Rick Pitino, a highly regarded new head coach[6]. The Pitino contract was quite lucrative, reportedly $70 million to be paid over a ten-year period. The partnership's shares traded publicly on the New York Stock Exchange. On the day following the Pitino hiring, the stock price rose by 7.4 percent. Trading volume soared to about 70 times the normal daily average, and after a month, the stock price had risen a total of 8.2 percent. Unfortunately for the Celtics, the team's performance did not rise similarly. The team never won consistently under Coach Pitino, who resigned in January 2001. The Celtics' share price languished after the initial burst. In these and other similar cases, what would have happened if the period for which the managers were being evaluated ended just after the hiring of these promising individuals but before the lack of accomplishments was realized? Judging by the market measures alone, the top management of Kodak, Sunbeam, and the Boston Celtics would have been evaluated quite favorably even though no real, sustainable improvements were ever realized from these hiring decisions. These managers might have been rewarded handsomely, and it would be quite difcult and costly for the corporations to recover these awards paid in advance of the real accomplishments (Barclay et al., 2003). Similar problems can occur with other types of unrealized expectations, such as failures of \"promising\" research projects late in the development General managers' performances 899 AAAJ 19,6 900 process, or even when problems are discovered after a product has been introduced to the market as was the case in late 2004 with Merck's VIOXX. A third problem with market measures of performance is actually a potential congruence failure. Market valuations are not perfect; they are not always totally congruent with changes in the true \"intrinsic\" value of the entity. Markets are not always well informed. For condentiality and competitive reasons, not everything that managers know is disclosed to the market. Managers who wish to do so can manipulate stock prices in the short-run through the timing of disclosures (e.g. Aboody and Kasznik, 2000). The market valuations cannot reect information that is not available to it. Even market valuations with well-informed participants might not be correct. Over the years, a number of valuation \"anomalies\" have been documented, including the \"January effect\" (e.g. Rozeff and Kinney, 1976; Bhardwaj and Brooks, 1992), the \"Monday effect\" (e.g. French, 1980; Kamara, 1997), the \"small rm effect\" (e.g. Banz, 1981; Reinganum, 1981), the \"P/E ratio effect\" (e.g. Basu, 1977; Fama and French, 1995), and the \"Value-line enigma\" (e.g., Stickel, 1985). These value anomalies are relatively small, typically only a few percentage points at most and only temporary in duration. And these are anomalies only because they cannot be explained within the existing efcient market hypothesis paradigm, so they may not be anomalies at all. More signicant for incentive purposes are some other, larger market imperfections and lags, which are particularly likely, and more likely to be signicant, in developing countries where stocks are not as actively traded (e.g. Islam and Watanapalachaikul, 2002). Indeed in developing countries, suggestions to reward managers based on stock market valuation changes are met with incredulity. Many locals believe that because regulations in their country are not as well developed and not as well enforced as those in advanced countries, managers can time or slant their disclosures to affect market valuations, and large investors can manipulate the markets. The drivers of market valuations in these markets and the degree of market efciency are the subjects of much current research (Litan et al., 2003). Even large, actively traded markets exhibit what seem to be biases, mood swings, and other imperfections (e.g. Barberis and Thaler, 2003; Shiller, 1981). For example, it is difcult to explain the rapid stock market decline of 22.7 percent in the crash of 1987 within the theoretical framework of the efcient markets hypothesis without considering behavioral factors. And, more generally, markets seem to overreact to earnings developments in both the up and down directions, so it is difcult to explain all market valuations without considering some behavioral (\"non-economic\") factors (e.g. DeBondt and Thaler, 1985, 1990; Bernard, 1992). In summary, market measures of performance have some signicant advantages. Importantly, congruence is generally high. Market measures, where they exist, provide perhaps the best indications that are available as to whether an entity has created or destroyed value in any given period. Further, for publicly traded entities, market measures are available on a timely basis; they are accurate (e.g. precise and objective); they are understandable, at least in the sense that people understand what they represent; and they are cost effective. However, market measures have some severe limitations and problems. They are only available for publicly-traded rms. They are largely uncontrollable by any employees except the top few individuals in the management hierarchy. And even for those few individuals, the measures are buffeted by many uncontrollable inuences, making the market measures \"noisy\" indicators of performance. Regarding this latter problem, however, it is possible to \"improve\" the market measures, to make them more reective of the controllable elements of performance, using standard management accounting techniques such as variance analyses, exible budgeting, or relative performance evaluation procedures. For example, managers can be held accountable for generating market returns greater than those of the overall market or, better yet, greater than those of the closest peer group. Or adjustments can be made to adjust for the effects of specic factors deemed uncontrollable, such as recessionary conditions, foreign currency exchange movements, the loss of a major customer that declared bankruptcy, or unforeseen \"acts of nature\". Accounting-based measures Traditionally, most corporations have based general managers' evaluations and rewards heavily on standard accounting-based, summary nancial measures. Murphy (1999) found that 161 of the 177 rms in his sample included at least one summary accounting measure in their annual bonus plans. Accounting-based, summary, \"bottom-line\" performance measures come in two basic forms: (1) residual measures (e.g. net income, operating prot, EBITDA, residual income); (2) ratio measures (e.g. return on investment, return on equity, return on net assets, RAROC). These measures are typically derived from the rules dened by standard-setters for nancial reporting purposes. Summary nancial measures have some signicant advantages. They satisfy many of the measurement criteria: . They present the illusion of being congruent with market returns. Many managers believe that markets respond vigorously to changes in reported accounting prots and returns, and they dene their organization's objectives in terms of accounting prots or returns (e.g. 8 percent prot growth, 18 percent return on assets). . They can be tailored to match the authority limits of any level of manager, from the CEO down to the lowest level of general manager and, hence, to improve controllability. The lower-level managers are typically held accountable for fewer of the income statement and balance sheet line-items. . They can be measured on a timely basis. Net income and accounting returns are regularly measured on a quarterly and monthly basis, and some rms even measure these quantities daily. . They are relatively accurate. The rules for measuring accounting income statement and balance sheet items are described in great detail, thus reducing much of the measurement variance, and they are largely objective because external auditors provide a periodic objectivity check. . They are understandable. Accounting training is a core element in every business education program. Virtually every manager who rises to a general management level knows through formal education or experience what the accounting measures represent. General managers' performances 901 AAAJ 19,6 902 . And nally, the measures are cost effective. Firms are required to produce the measures for nancial reporting purposes, so there is little or no incremental cost. However, summary nancial measures of performance present two major problems. One is an aspect of the controllability problem: Financial performance measures are affected by many of the same macroeconomic distortions that affect the market performance measures (Oxelheim and Wihlborg, 2001). The measurement lags and magnitudes of effects are likely to be different, but the direction of most of the effects is the same. When the economy goes into recession or when oil prices rise, the nancial measures of performance are affected like the market measures, and for virtually all but the oil producers, the effect is adverse. A second, potentially more serious, problem with the summary nancial performance measures is that in many situations they are not highly congruent with changes in value. The congruence failures can be explained theoretically and illustrated empirically. It is easy to explain theoretically why accounting prot measures do not reect value changes perfectly. Many things affect accounting prots but not economic prots, and vice versa. . Prot measures focus on the past. Economic value is derived from future cash ows, and there is no guarantee that past performance is a reliable indicator of future performance. . Accounting systems are transactions-oriented. Accounting prot is primarily a summation of the effects of the transactions that took place during a given period. Most changes in value that do not result in a transaction are not recognized in income. When a rm receives a patent or regulatory approval for a new drug, huge value is created, but there is no transaction, no accounting entry, and no effect on accounting income. . Accounting prot (and measures derived from it) is highly dependent on the choice of measurement method (e.g. Brown, 2002). Multiple measurement methods are often available to account for identical economic events. Fixed asset depreciation accounting options (assets lives and depreciation methods) and inventory accounting alternatives (e.g. FIFO versus LIFO) are but two examples. . Accounting prot is derived from measurement rules that are often conservatively biased. Accounting rules require slow recognition of gains and revenues but quick recognition of expenses and losses (e.g. Watts, 2003a, b). For example, accounting rules dene strict criteria that must be satised before revenue (and the associated prot) can be recognized, and expenditures on intangible assets are generally expensed immediately and not recognized on a company's nancial statements. Thus, accounting measures do not match revenues and expenses well, and this problem is particularly acute where measurement periods are shorter than the rms' investment payoff horizons. . Prot calculations ignore some economic values and value changes that accountants feel cannot be measured accurately and objectively. Investments in major categories of companies' intangible assets, such as research in progress, human resources, information systems, and customer goodwill, are expensed immediately. These types of assets do not appear on the balance sheet. The . . . omission of intangible assets occurs even though for many companies these types of assets are much more important than the old industrial era-type assets of plant, equipment, and land (e.g. Lev, 2001). For many companies, the tangible assets represent only a small fraction of the company's total market value. Prot ignores the costs of investments in working capital, even though these investments tie up capital and, hence, have real economic costs. Prot reects the cost of borrowed capital but ignores the cost of equity capital. Firms earn real income only when the returns on capital are greater than the cost of that capital, and ignoring the cost of equity capital overstates the difference between returns and costs (that is, prot). This omission is serious because equity capital is typically more expensive than borrowed capital. In the USA, stock returns have historically been approximately six percentage points higher than returns on long-term government bonds, and the cost of equity capital is even higher for companies with risky (volatile) stocks. Failure to reect the cost of equity capital also hinders comparisons of the results of companies with different proportions of debt and equity in their capital structures. Accounting prot ignores risk and changes in risk. Firms, or entities within rms, that have not changed the pattern or timing of their expected future cash ows but have made the cash ows more certain (less risky) have increased their economic value. This value change is not reected in accounting prots. Because of these divergences between market and accounting performance measures, empirical tests of the congruence of accounting performance measures have consistently shown, not surprisingly, that the signals provided by reports of accounting income are not reliable indicators of market value changes. Rappaport (1981) conducted a study of 172 of the largest US rms that had annual growth in earnings per share of 15 percent or more over a six-year period. In 27 (15 percent) of those rms, shareholders realized negative rates of return (dividends less capital losses). In 60 (35 percent) of those rms, shareholders had returns that were negative in a \"real\" sense; i.e. less than ination. More generally, studies have consistently shown that correlations between annual accounting income and annual market value changes correlations are typically quite low, usually in the range of 0.1 to 0.3. The correlations for some companies, such as those with relatively high proportions of tangible assets already in place, are likely to be higher than the correlations for others, such as those that are investing heavily in intangible \"growth options\