On January 6, Lucent Technologies Inc., makers of high-tech communications equipment for companies that provide wireless access, cable television service, and voice, data, and video services, announced that its fourth-quarter earnings would fall short of analysts' estimates by about 29 percent. The next day, 179 million Lucent shares changed hands - an all-time record for any company - and its stock price plunged by 20 points, erasing a whopping $64 billion or almost 30 percent of its market value.

Lucent attributed the earnings surprise, in part, to manufacturing problems that left the company unable to meet demand when its customers shifted more quickly than expected to its new optical products. Changes in the timing of customer purchases were also responsible for lower network and software revenues, Lucent said. Analysts agreed that the firm's shortfall was not indicative of an industry-wide slowdown, but of faulty "execution.” As a major player in a rapidly changing industry, Lucent faces fierce competition from Cisco Systems, Inc., Nortel Networks Corp., and a number of hot start-ups, all vying to supply the newest, quickest network equipment for sending data over the Internet.

So it is perhaps more remarkable that Lucent had reported relatively few earnings surprises before this. In technologically dynamic and fast-growing industries, one might expect customers and costs to be unpredictable (and for investors to factor this into the stock price). So why aren't there more earnings surprises?

One conjecture is that firms use accounting latitude to "manage” their reported earnings. In some instances, this can involve pumping up earnings; in others, firms can take excessive charges which reduce current earnings (but make future earnings look better) or smooth earnings so they appear less volatile. As far back as October 1998, at least one analyst suggested that Lucent's reported earnings were higher and more predictable because of sizable reserves which had been set aside to pay for restructuring costs and then pulled back into income when not all of those costs materialized.

Lucent's numbers have passed Securities and Exchange Commission (SEC) scrutiny, but others have not. Last June, for example, W.R. Grace settled with the SEC after it was accused of holding off "excess” earnings from a subsidiary that was growing faster than the firm's targeted 24 percent growth rate, and then using this to boost earnings later on. And last July, Microsoft announced that its financial statements were also under SEC investigation, apparently prompted by a former Microsoft employee who alleged he had been fired after reporting earnings smoothing to his boss.

Such financial reporting practices have clearly attracted the SEC's attention. For troubled companies, earnings management is obviously questionable - a way of hiding problems or buying time in hopes that managerial effort or simply good fortune will turn things around. Maintaining the appearance of profitability or earnings growth may even allow struggling firms to continue securing funds.

In instances where the company is, in fact, healthy all along, earnings management is also problematic since this is only known for certain in retrospect. Firms sometimes counter by pointing out that investors dislike volatility. But masking information about the true course of earnings doesn't make the underlying business any less risky. It only makes financial reports more difficult to interpret.

In a high-profile speech at New York University in the fall of 1998, SEC Chairman Arthur Levitt called earnings management "a game that runs counter to the very principles behind our market's strength and success,” namely transparent, timely, and reliable financial statements. If investors are to evaluate firms' prospects, they need an accurate picture of financial performance over time. They must also be able to compare the performance of one firm to another. Without such transparency and comparability, noted Levitt, "investors grow anxious and prices fluctuate for no discernible reasons.”

In time, the game can become self-perpetuating; companies tweak the numbers to meet analysts' expectations and analysts look to firms for guidance in setting those expectations. Over the long haul, unreliable financial reports increase the risk for investors, and may lead to a higher cost of capital - and a less efficient flow of funds - to the very companies we depend on to fuel future economic growth.

DISCRETION VS RULES

Investors want financial statements to provide a precise and accurate picture of the firm's performance and not just reflect wishful thinking. An income statement should aggregate the revenues and expenses for transactions that occurred during the quarter. The cost of an asset should be charged against revenues over the asset's useful life. And firms should be consistent in their accounting measures and procedures, so investors can more easily compare one year to the next, and one firm to another.

In the United States, the rules and standards governing such matters are set by two groups. The SEC, a federal agency established by Congress in the 1930s, enforces U.S. securities laws and is the final authority for establishing accounting and reporting standards for publicly held companies. The SEC generally relies on a private-sector group, the Financial Accounting Standards Board (FASB), to write the standards for financial reporting and accounting that companies and auditors use in preparing financial reports, subject to SEC review.

But setting standards that match up to accounting principles is not always simple and may necessitate giving the firm some flexibility. At any given point in time, some of the firm's future revenues and costs are genuinely uncertain and no set of hard-and-fast rules can nail them down. Put another way True economic income is generally unobservable. Thus, accounting is not an exact science. It allows wiggle room for certain adjustments not considered large enough to be "material.” And, inevitably, there are instances where firms exercise judgment. Such discretion leaves open the opportunity for firms to manage earnings.

Probably the most common example is how quickly or slowly firms book revenue. (It is also the most common example of outright fraud, such as when a company fabricates sales.) This decision can be straightforward, but products such as software often come with long-term service contracts and upgrades that can stretch out for years. Microsoft, for example, must decide how much of its current revenue should be attributed to current software sales and how much should be set aside for future upgrades and service.

The use of reserves is another instance where firms have discretion. Reserves are supposed to dedicate some part of today's income to an expected future cost, over and above ordinary operating costs, such as the anticipated outlays that will arise from a merger or restructuring (for severance, retraining, combining systems, etc.), or to cover the costs of a future lawsuit. Banks, for example, take reserves to cushion against loans that they suspect will go unpaid. Such reserves should be taken for charges when the liability can be estimated. But only hindsight is 20/20, so the rules leave room for judgment about exactly what sorts of circumstances can give rise to such charges and when they can be taken and how large they should be.

For Lucent, the event was its spinoff from AT&T. In 1995, Lucent set up a $2.8 billion reserve to cover the future costs of restructuring - its best estimate of the outlays over the next several years to cover severance for 20,000 employees and the cost of leaving businesses such as AT&T's Phone Center Stores. So earnings in 1995 took a huge hit, with Lucent reporting a net loss of $867 million. But when severance payments and other costs were incurred later on, they did not get subtracted from earnings, and reported earnings were higher than they would have been otherwise.

As it turns out, Lucent greatly overestimated the reserves it would need to pay for the restructuring; a booming economy helped employees find new jobs quickly and kept severance costs down. Over the next four years, more than $500 million of this restructuring reserve was converted back to pretax income, increasing and smoothing out what would have otherwise been more volatile earnings. For example, in 1999, Lucent included in income an additional $141 million from these reserves. Brad Rexroad, an analyst at the Center for Financial Research and Analysis, told The Wall Street Journal  "Lucent has been reversing reserves most every quarter during the past five years.” By September of 1999, however, all restructuring plans had been essentially completed and reserves stood at only $18 million.

Firms also have some discretion in the accounting method used in a merger or acquisition. One factor that managers consider is the impact on the firm's financial statements. In a "purchase” transaction, the acquiring company records the current market value of everything it has bought, including identifiable intangible assets, such as patents and licenses. Any gap between the purchase price and the value of those assets is called "goodwill.” Goodwill represents the value of the target to the acquiring firm over and above the fair market value of all assets that can be identified; for some companies, especially high tech firms where the bulk of assets is in intangibles that can't be identified, almost the entire purchase price can show up in goodwill. That goodwill - representing valuable assets for which the buyer is paying and which have a limited life span - must over time be subtracted from future revenues, thus reducing future earnings.

Lucent was able to reduce the goodwill that came from some of its acquisitions (Octel Communications, Yurie Systems, and others) because of the way accounting currently handles "in-process” R&D. When a firm invests in a tangible asset, the cost is charged against revenue over the asset's useful life. But for many intangible assets, such as R&D spending, whether the investment will yield any revenue is highly uncertain, and firms are required to deduct the entire cost immediately. In an acquisition that uses purchase accounting, the firm can include in the purchase price its own estimate of the future value of R&D in the target company. In this way, Lucent has avoided $2.5 billion in goodwill (and the associated drag on earnings) since 1996.

The alternative to purchase accounting is "pooling,” which was originally intended for fims that were joining forces (as opposed to one firm buying the assets of another). In order to pool, a firm must meet twelve conditions certain financial transactions are prohibited for a period of time; the owners of one company must be paid predominantly in the stock of the other, etc. As an accounting matter, the merging firms simply combine their individual balance sheets by adding together their assets and liabilities.

Pooling results in a more "attractive” balance sheet because assets are not marked up to current market value (they sit on the balance sheet at historical cost), so future depreciation charges will be lower. And there is no goodwill to drag down future earnings. There is also nothing on the balance sheet to indicate that the price paid was greater than the market value of identifiable assets. So pooling makes it hard for investors who want to figure out later whether the acquisition was worth the price paid - perhaps one reason for the recent FASB decision to propose its elimination.

INCENTIVES TO MANAGE

If accounting discretion gives managers the opportunity to manage earnings, what provides the motive? This question has always puzzled those who believe that financial markets are perfectly efficient. In their view, an accounting choice may change reported earnings, but it doesn't increase a firm's assets, reduce its liabilities, or alter anything about its underlying economics or future prospects. Armies of savvy investors and analysts have enormous incentives to "see through” managed accounting numbers. Why should it make any difference to them how two merging firms combine their balance sheets?

In the long run, financial markets may be efficient. But in the short run, perceptions matter and can affect the way firms act. For example, once an accounting change makes an item more "visible,” firms tend to increase their efforts at managing it. They also spend time and money complaining whenever FASB proposes a policy change which - even though it will not affect firms' underlying economics - will reduce reported earnings. Presumably, the firms believe that latitude in accounting confers at least some advantage. One big advantage is the opportunity for managers to protect their jobs and income. Earnings are the single most important explanation of firms' stock market returns over the medium to longer run (one to ten years), and a significant determinant, even in the short run. So investors and the firm's board of directors - who are responsible for senior executive hiring, firing, and compensation - generally link their evaluations to earnings, stock price performance, or both. This can create the incentive for executives to boost reported earnings. They may even try to do so at the expense of future earnings if, for example, they expect their efforts or economic forces to produce a turnaround, or hope that good fortune might intervene. Managers who expect to retire or leave the firm also have an incentive to engage in this practice.

But why might firms use accounting latitude to reduce earnings? Evidence from both behavioral finance and psychology suggests that people tend to focus on the present and "forget” about the past. This gives firms an incentive to bunch restructuring charges or losses all at once, and even pull in normal operating expenses, thus clearing the decks for the future. In 1997, for example, McDonalds tried to lump in a $190 million charge for installing new grills and ovens - presumably normal operating costs for a fast-food company - with other one-time charges of about $160 million, until the SEC questioned the charge. Similarly with in-process R&D charges at the time of an acquisition The costs will be written off immediately, so there is an incentive to pile in other expenses or bad news to get them out of the way.

And why might firms choose to smooth earnings? Prospect theory tells us that when people choose among risky alternatives, they seem to evaluate possible outcomes not in absolute terms, but as changes from a reference point (which can shift over time). Moreover, losses tend to count about twice as heavily as gains in their mental calculus.

Thus, boards and investors tend to retain and reward managers depending on whether the firm hits (or misses) a certain reference point, or threshold, argue economists Francois Degeorge, Jayendu Patel, and Richard Zeckhauser. The three most common targets are earnings greater than zero; earnings greater than last quarter; and earnings that meet analysts' estimates. Miss the mark by a little and a manager may be fired or his or her bonus drastically reduced; exceed it and the bonus may hardly increase at all. This gives managers an incentive to use accounting latitude to stash earnings above the threshold and pull them out later when earnings fall below, making reported earnings smoother than otherwise. Managers may also fear that if they exceed their target by too much, the bar may be higher in the future.

A focus on earnings targets may also help to reduce information costs. Banks may find it convenient to screen loan applicants by eliminating firms unless they report positive earnings, for example. Or customers wanting to invest in a new computer system may find it easier to judge a firm's chances of still being in business when the machine needs to be serviced by using a simple earnings rule of thumb. And managers may find it easier to explain to shareholders that earnings are up for ten quarters in a row than to explain that they have risen nine quarters and fallen by 1 percent in the most recent quarter. Faced with this, managers may not be able to resist using accounting discretion to "make their numbers.” Even if only a small number of bankers or customers focus on earnings targets, shareholders and managers will figure this out and eventually care, too.

Degeorge et al. look at the distribution of firms' earnings and find that "too few” of the firms come in just below each threshold and "too many” at or directly above it. (In their study, nonaccounting methods, such as year-end price cuts to boost revenue or choosing to delay or accelerate spending on repairs or investment in new equipment, also count as earnings management.) This suggests that managers do indeed tend to smooth earnings. There also seems to be a hierarchy involved. Firms seem to care first about reporting positive earnings, then about meeting last period's earnings, and, only when both of those are met, about meeting analysts' estimates.

ACCOUNTING MISMATCH?

Intangibles now make up a significant portion of the assets of many high tech firms. And mergers seem to be an almost daily occurrence as new technology, deregulation, and globalization provide the impetus for one industry after another to restructure. Some have wondered whether this has increased the opportunities to manage earnings and reduced the quality of financial reporting data.

Professors Baruch Lev and Paul Zarowin, of NYU, have tried to assess the value of accounting data to investors by looking at whether it tends to be associated with changes in stock market performance. They found that reported earnings, cash flows, and book values have all deteriorated in usefulness over the last twenty years, despite increasing investor demand for information and persistent efforts by regulators to improve that information. They point to changes in the economy that have resulted in large investments in R&D and restructuring. Since costs are deducted immediately while benefits are recorded later, this tends to understate current earnings and overstate earnings in the future. In their view, this has "seriously distorted” accounting measures and undermined the usefulness of financial information. Not every one agrees; the recent low inflation rate, for example, has probably reduced the difficulty in accurately accounting for depreciation.

Still, Lev and Zarowin claim that some accounting principles are ill matched to our high tech economy. Accounting has traditionally been based on reporting discrete transaction-based events such as sales, purchases, and investments, they argue; yet the impact of such things as R&D are rarely triggered by specific transactions. Product development, for example, often affects the value of the enterprise long after any revenue or expense warrants an accounting record. And, traditional accounting measures provide relatively little useful information about the intangible assets which make up an increasing share of corporate wealth.

There are a number of new efforts to address these issues. The Sloan School of Management at MIT and Arthur Andersen recently formed the New Economy Research Lab in Cambridge, Massachusetts. One of the items on its agenda is devising methods for quantitative valuations of intangible assets. Brookings has also begun a project entitled, "Understanding Intangible Sources of Value.” And Chairman Arthur Levitt announced that the SEC would ask the accounting profession to clarify the rules for in-process R&D, and that it would "formally target” for review companies that announce restructuring reserves, major write-offs, or other practices that "appear to manage earnings.”

In the meantime, investors will have to rely on traditional accounting detective work, such as comparing earnings with cash flow from operations or checking whether receivables are piling up faster than sales or whether insiders are selling their stock. And, rather than rewarding firms that smooth, investors who want to reduce volatility would be better off diversifying so that the surprises in their portfolios tend to cancel each other out.

There will always be changes in the economy that require new accounting rules and practices. This past December, the SEC sought to clarify its rules on revenue recognition, prompted by Internet firms that were reporting revenues based on the total dollar value of web transactions, not just their fees. As the economy evolves, such strains are probably inevitable and accounting will have to continue to adapt. Like earnings, the course of economic change rarely runs smooth.