Is the Response of Analysts to Information Consistent with Fundamental Valuation? The Case of Intel.
This paper examines the market reaction to a press release issued by Intel on Thursday, September 21, 2000. In response to that release, Intel's stock price dropped 30%, erasing over $120 billion of shareholder wealth. By analyzing the press release in conjunction with analyst reports, and by using a discounted cash flow valuation model, I argue that the information conveyed by the announcement was not sufficient to explain the stock price drop. Surprisingly, analysts were more strongly recommending purchase of the stock in August 2000 at $75 than in September 2000 at $40. This suggests a positive feedback between stock price movements and analyst recommendations that may increase the volatility of prices.
Intel is, without question, one of the world's premier corporations. The company, which is far and away the largest manufacturer of microprocessors, is also a diversified maker of semiconductor chips that are integral parts of boards, systems, and software employed in the production of computers, servers, and networking and communications products. Intel's success has been reflected in its stock price. At various times during the year 2000, it was the largest market capitalization company in the world. In addition, Intel is one of the most actively followed and widely held companies. Major institutions hold well over half of its stock. Dozens of analysts scrutinize every statement the company makes almost as carefully as they examine the pronouncements of Alan Greenspan. If there is one company that the semi-strong form of the efficient market hypothesis should apply to, it would be Intel.
This paper focuses on the market reaction to a press release issued by Intel on Thursday, September 21, 2000 at 4:16 pm Eastern time. The announcement, which is quoted in detail in the next section, stated, among other things, that the company expected revenue for the third quarter to be 3 to 5% higher than second quarter revenue of $8.3 billion. This fell short of the company's previous forecast of 7 to 9% growth and of analysts' projections of 8 to 12%.
The market's response was astonishing. Although trading in Intel was halted for the remainder of the day after the announcement at 4:16 pm, the stock dropped in after-market trading from the "close" of $61.48 to $48.25--erasing over $91 billion in market value. Over the next two days, the price continued to fall. By the close on September 26, 2000, the stock was down almost 30% to $43.31 and $122 billion in shareholder value had evaporated.
In response to the dramatic sell-off, some Intel executives were perplexed. A senior executive, in private conversation, expressed amazement that what appeared to him as a relatively minor announcement had led to the destruction of over $120 billion in shareholder wealth. The announcement, in his view, reflected short-run developments in Europe, in particular, the decline of the Euro, which made it more difficult for Intel to sell products there in the short term. However, it did not indicate any change in Intel's long-run strategy, product mix, competitive position, or even, in his opinion, the long-run demand for Intel's products. Consequently, he argued that while the stock should have dropped in response to the surprise, the decline should have been modest. In his view, the market was irrationally overreacting to a small dose of bad news. Several weeks following the announcement, Intel's chairman, Craig Barrett, adopted this position in a public question and answer session saying that, "I don't know what you call it but an overreaction and the market feeding on itself."
The Intel event is by no means unique. In the weeks that followed, Kodak dropped over 25% in response to an earnings warning and Apple plummeted a remarkable 51% after ii issued a warning. What makes these events particularly surprising is that Kodak and Apple, like Intel, are established companies with long track records that are widely followed by dozens of analysts. What kind of beliefs must analysts and investors have had about the value of the company for information regarding the current quarter's earnings to cause the company's stock price to fall by as much as one-half?
In response to market reactions such as those experienced by Intel, Kodak, and Apple, the view expressed by Craig Barrett--that the market overreacts to news, particularly bad news--has become common among practitioners. Academic researchers have also actively studied the overreaction hypothesis for years without a clear consensus emerging. Uncertainty about the issue remains, because overreaction is so hard to measure. Without a metric for determining the "proper reaction," there is no direct way to say whether or not the market has overreacted. For this reason, the overreaction research in the academic literature has concentrated on the behavior of large samples of companies over the long run. Empirical tests are based on the idea that if the market overreacts, then investors should be able to earn superior risk-adjusted returns over the long-term by taking positions in the direction opposite to the overreaction. This is the approach taken, for example, by DeBondt and Thaler (1985), Zarowin (1989), and Cho pra, Lakonishok, and Ritter (1992). However, as Fama (1998) observes, the statistical power of the tests are weak and results are open to differing interpretation. Consequently, after more than 15 years of research, the overreaction hypothesis remains controversial.
The huge reaction to Intel's stock price raises another question even more fundamental than that of overreaction. Specifically, if the quanta of valuation information contained in the announcement is small and the market reaction is large, then the stock could not possibly be rationally priced both before and after the announcement. The most famous example of this problem is the crash of October 1987 when the overall market dropped more than 21% on what appeared to
be minor news.  In the case of Intel, company spokesman Tom Beerman adopted such a position by arguing that the information in the press release was not sufficient to explain the drop. The problem with this view, of course, is that there is no unambiguous way to measure the quanta of information in an announcement other than by observing stock price movement in response to the announcement.
In an attempt to shed new light on these issues, this paper takes a clinical approach. Rather than drawing statistical conclusions from the behavior of large samples, it examines one company's response to one announcement in great detail. As part of the examination, a complete discounted cash flow (DCF) valuation model of Intel is constructed to serve as a benchmark in evaluating the market reaction to the company's press release. In addition, analysts' reports, both before and after the announcement, are studied in order to understand how analysts responded to the press release and what role their reports and recommendations might have played in determining the magnitude of the market's reaction. Of course, there is a limit to the conclusions that can be drawn from an individual case study. Nonetheless, the results are sufficiently provocative that they raise some challenging questions.
The remainder of the paper is organized as follows. The first section presents more detailed information on Intel's press release and the market's reaction to it. In section II, a detailed discounted cash flow valuation model is developed as a tool to determine whether the market reaction was a reasonable response to the information conveyed by the press release. The third section analyzes the reaction of analysts to the announcement. It is argued that the failure of analysts to develop long-term valuation models and the procyclical nature of analysts' recommendations can potentially exacerbate market reactions to announcements such as Intel's. The conclusions are presented in the final section.
I. Intel's Announcement and the Market Response
In the nine months preceding September 2000, Intel had been one of the best performing large capitalization stocks in the market. As Figure 1 illustrates, between December 31, 1999 and August 31, 2000 Intel rose 82.3%, markedly out-performing the S&P 500 which increased 4.2% and the NASDAQ index which increased 4.3%. Figure 2 shows that the run-up in the stock price also pushed the market capitalization of Intel to record levels. In August 2000, Intel was the only company in the world with a market value exceeding $500 billion.
During the time the stock was running up, Intel was also a darling of the analysts. By the end of August 2000, Bloomberg's summary index of analyst recommendations stood at 4.85 out of 5.0 compared to an average of 4.24 for the S&P 500.  A Bloomberg index of that level indicates that virtually every analyst who followed Intel was highly recommending purchase of the stock. These bullish recommendations were maintained in spite of the fact that the run-up had made Intel's stock appear relatively expensive. The P/E ratio had risen from under 30 in October 1999 to over 55 by August 2000.
In September 2000, Intel's stock price began to falter. It fell from a high of $74.88 on August 31, 2000, to a low of $55.75 on September 18, 2000, before recovering to $61.48 on September 21, 2000. The drop of 17.9% from August 31, 2000 to September 21, 2000 markedly exceeded the 4.5% decline in the S&P 500 and the 8.9% decline in the NASDAQ index. The financial press attributed the drop in Intel shares, at least in part, to statements by Piper Jaffray analyst, Ashok Kumar, who downgraded the stock on September 5, 2000. Kumar based the downgrade on his belief that PC sales growth in the third quarter would be, at best, about half of the 12% estimated by most analysts and that Intel's growth would be similarly affected.
After the market closed on September 21, 2000, Intel issued an earthshaking press release regarding the financial results for the third quarter due to be released at the end of October 2000. Because of its importance, at least in the eyes of the market, the financial information included in the press release is quoted in detail. The company stated:
The following statements are based on current expectations. These statements are forward-looking, and actual results may differ materially. These statements do not reflect the potential impact of any mergers or acquisitions that may be completed after the date of this release.
* The company expects revenue for the third quarter of 2000 to be approximately 3 to 5% higher than second quarter revenue of $8.3 billion.
* The company expects gross margin percentage for the third quarter to be 62%, plus or minus a point. Gross margin percentage for 2000 is expected to be 63%, plus or minus a few points. In the short term, Intel's gross margin percentage varies primarily with revenue levels and product mix as well as changes in unit costs.
* Expenses (R&D, excluding in-process R&D, plus MG&A) in the third quarter of 2000 are expected to be up 7 to 9% from second quarter expense of $2.2 billion, primarily due to higher spending on marketing programs and R&D initiatives in new business areas. Expenses are dependent in part on the level of revenue.
* R&D spending, excluding in-process R&D, is expected to be approximately $4.0 billion for 2000.
* The company expects interest and other income for the third quarter of 2000 to be approximately $900 million. Interest and other income is dependent, in part, on interest rates, cash balances, equity market levels and volatility the realization of expected gains on investments, including gains on investments acquired by third parties, and assuming no unanticipated items.
* The tax rate for 2000 is expected to be approximately 31.8%, excluding the impact of the previously announced agreement with the Internal Revenue Service and acquisition-related costs.
* Capital spending for 2000 is expected to be approximately $6 billion.
* Depreciation is expected to be approximately $790 million in the third quarter and $3.4 billion for the full year 2000.
* Amortization of goodwill and other acquisition-related intangibles is expected to be approximately $400 million in the third quarter and $1.5 billion for the full year 2000.
The reaction of analysts provides a benchmark for interpreting the information conveyed by the announcement. Intel is routinely followed by approximately 50 analysts. These include the employees of every major investment bank and brokerage firm. Larger firms, like Merrill Lynch and Solomon Smith Barney, have more than one analyst tracking Intel. As of September 2000, there were 28 current analyst reports available on the company. An examination of those reports reveals that, in the eyes of analysts, most of the information in the press release was innocuous. Statements regarding capital spending, amortization of goodwill, interest income, R&D expenses, and depreciation were seen as in line with expectations. The critical exception was the statement regarding future revenues. The warning issued by Intel predicted revenue growth for the third quarter at 3 to 5%. This was well below the range of 8 to 12% forecast by Wall Street analysts and even below the number of 6% projected by Ashok Kumar. Furthermore, the lower revenue number implied that margins would decline slightly because of the fixed nature of some of Intel's short-run costs.
What is perhaps most striking about the press release is what it does not contain. There is no discussion of the company's long-run business outlook, the quality of its products, actions by competitors, basic changes in technology or the demand for Intel's products, or new government sanctions or regulations. In short, there is nothing to suggest that the fundamental long-run business conditions for Intel were much different on September 22, 2000 than they had been on September 21, 2000. Intel spokesman, Tom Beerman, stressed this point by arguing that the negative implications should not be exaggerated. According to Beerman, the problems Intel faced were limited both chronologically and geographically. More specifically, he stated that, "It (the slowdown in revenue growth) is demand related, and it is focused exclusively in Europe. The other geographies are coming in as expected."
Despite the ameliorative assessment presented by Beerman, the reaction of the market was immediate and dramatic. In after-hours trading, Intel's stock price plummeted over 21% from $61.48 to $48.25. The next day's trading revealed that the drop was not a transitory event. On September 22, 2000, Intel broke the NASDAQ volume record by over 100 million shares as 309 million shares changed hands (See Figure 2). As the market further assessed the news, the stock price continued to drop. Intel fell 5.4% to $45.38 on Monday, September 25, 2000 and dropped another 4.6% to $43.31 on September 26, 2000. In total, the company's share price fell 29.6% over the three trading day window--erasing $122 billion in market value. Furthermore, in the weeks following the drop, the stock price did not recover.
Figure 3 shows the three-day decline in Intel's stock price in comparison with the S&P 500, NASDAQ, and major chip and computer companies.  The figure shows that most of the decline in Intel's price was unique to the company. The next largest drop in price, for Dell Computer, was 6.4%. Compaq Computer actually rose in price over the interval. This demonstrates that the news the market was reacting to was not primarily a general slowdown in the market for computers and semiconductors, signaled by Intel's warning, but was predominantly Intel-specific.
The Intel-specific nature of the drop, in a sense, deepens the mystery. Because competitive issues and governmental sanctions or regulations were not involved, the remaining long-run story is that the press release presaged a fundamental drop in the demand for the company's products. If such were the case, however, related firms, particularly computer manufacturers, should have been equally affected, because Intel makes the chips that go into all their machines. Furthermore, there was no evidence in the release to suggest that the slowdown was specifically related to Intel's products.
Finally, it should be noted that the information in the press release was probably partially anticipated by the market. By September 21, 2000, Intel's price was already down 15% from the high, largely on Kumar's warning about third quarter demand for the company's products. To the extent that the announcement was anticipated, the market response to the press release should be a downward biased estimate of the amount of information in the release.
II. The Market Response and the Valuation of Intel
To investigate further how the September 21, 2000 press release could have led to the destruction of $122 billion in Intel shareholder value, a discounted cash flow (DCF) valuation model for Intel is employed.  Based on analysts' forecasts developed prior to the announcement, the model is calibrated so that it yields a market value of equity equal to Intel's pre-announcement capitalization of $413 billion. Once calibrated, the model can then be used to gain insight into how expectations of future cash flows for the company must have changed to cause Intel's value to fall to $291 billion over the course of three trading days.
A. The Valuation Model
It should be stressed at the outset that the goal of the model is not to attempt to provide an estimate of the true value of Intel, either before or after the announcement. As a professor of finance, I do not have access to information that would allow the production of unusually accurate cash flow projections. Nonetheless, by calibrating the model using pre-announcement cash flow projections based to the greatest extent possible on pre-announcement analyst reports, it is possible to calculate how much those forecasts must have changed in order to explain the movement in the stock price.
I assume that the discount rate does not change as a result of Intel's press release. Consequently, any movement in the stock price must be attributed to changes in cash flow expectations. This assumption is strongly supported by the data. First, examination of the Treasury yield curve reveals that it was unaffected by Intel's announcement. Second, the minor drop in the S&P 500 demonstrates that the equity market risk premium (ERP) was not significantly affected by Intel's warning. This means that any change in the discount rate would have to be caused by a change in the systematic risk of Intel. Because there is no clear reason to assume that the systematic risk would rise or fall in response to the announcement, it is assumed to remain unchanged.
A standard DCF model is used to value Intel. Because Intel has almost no debt, there is virtually no difference between the WACC approach described by Cornell (1993), the adjusted present value approach described by Kaplan and Ruback (1995), and the capital flows approach advocated by Ruback (2000). Using Intel's lowest stock price in the year preceding September 22, 2000, debt still represents less than 1% of the company's capital structure. For simplicity, therefore, the valuation analysis proceeds as if the company were all-equity financed.
Although the DCF model depends exclusively on expected future cash flows, and not historical data, the past provides a benchmark to assess what can reasonably be expected in the future. For that purpose, Table I presents five years of historical financial data drawn from Intel's 10Ks. The data in Table I differ from the standard income statement presentation in two ways designed to make them more comparable to the organization of data in DCF valuation models. First, Intel's income statement includes amortization of goodwill created by acquisitions. Unlike depreciation, this amortization is not tax deductible. From a valuation standpoint, therefore, the amortization can be ignored--it need not be deducted from operating income, nor added back when calculating free cash flow. Second, the income earned on Intel's financial investments is excluded. Instead, the financial assets are treated as separate from the firm's operating assets, and their value is added to the estimate of the firm's operating value.
The historical data show that Intel's revenue grew at a geometric average rate of 16.1% during the years from 1995 to 1999. As revenues grew, moreover, gross margins improved, rising from approximately 50% to 60%. Because of rising research and development costs, operating margins fluctuated around 35% despite the improvement in gross margins. During the early part of the period, capital expenditures, as a percent of revenues, were relatively high. They fell to an average of about 12.5% of revenues in the last two years. Depreciation averaged about 9.5% of revenues during the period. It was somewhat higher toward the end of the period, reflecting the relatively greater capital expenditures in the first three years. Cash flow as a percent of revenues rose throughout the period to a high of 22% in 1999.
The financial data used to calibrate pre-announcement valuation model are presented in Table II. The cash flow projections are derived using a three-step procedure. For the first two years, the projections are based on detailed pro forma financial statements that are contained in the more complete analyst reports. The numbers selected are chosen to be representative, generally equal to the median, unless Intel provided specific information. For instance, in the September 21, 2000 release, Intel predicted that capital spending for fiscal year 2000 would be approximately $6 billion. Therefore, $6 billion is used in the DCF model.
Beyond two years, only a few sporadic projections are available in the analyst reports. Because of the incomplete data in the reports, projections for the next three years are based on a combination of the long-run I/B/E/S forecasts and an analysis of the operating ratios during both the five-year historical period, and the first two forecast years. The I/B/E/S data are the median five-year earnings growth forecasts from the universe of analysts that report to I/B/E/S. In the month before the announcement, the median forecast growth rate for Intel was 20%. Margins are assumed to remain constant in percentage terms during this period so that revenues also grow at 20%. The gross margin is assumed to be 62.5% and the operating margin is assumed to be 38.5%. These figures are greater than those achieved by Intel during the five-year historical period, but less than analysts' projections for the first two years of the forecast period. Capital expenditures, including investment in working capital, are assumed to be 13.5% of revenue. This is an estimate in line both with recent historical experience and with analyst forecasts for the first two years. Finally, depreciation is assumed to be 9% of revenues.
There are no external forecast data available up to the terminal value at year ten in the final five years. The task, therefore, is to develop reasonable projections that, when combined with an estimate of the discount rate, equate the DCF model value with Intel's market value. To achieve this objective, it is assumed that beginning in year five, Intel's growth rate in revenue slows at a rate of one percentage point per year. In addition, it is assumed that capital spending will drop to 13% of revenues, but depreciation will remain unchanged at 9% of revenues through the terminal horizon. All the other items remain the same percentage of revenues that prevailed during the preceding three years of the forecast period.
At the terminal horizon (year ten), a constant growth model is employed to estimate the remaining value. Notice that by the terminal date, Intel is an immense company. It is assumed, therefore, that beyond that point Intel cannot grow much faster than the aggregate US economy. Accordingly, a 6% nominal growth rate is chosen representing inflation of 3.5% and long-run real growth of 2.5%. Combining these assumptions gives the cash flow projections presented in Table II. In some sense, the terminal growth rate, like the equity risk premium described below, can be thought of as a "plug." If a higher terminal growth rate is assumed, then the equity risk premium can be raised from the low level chosen below.
The discount rate for Intel is composed of two elements: the risk-free rate and the risk premium appropriate for the company. As has become common in valuation applications, a long-term Treasury rate is chosen as the risk-free rate.5 Following Kaplan and Ruback (1995), among others, the 20-year Treasury rate is used here.
The determination of the risk premium is more important, but also more difficult and controversial. At the start, there is the issue of whether to use the CAPM or a more complex three factor model like the one suggested by Fama and French (1992). As reported by Brunner, Eades, Harris, and Higgins (1998) and Graham and Harvey (2000), the CAPM remains the model of choice in valuation practice. However, the evidence presented by Fama and French (1996), among others, indicates that the model does not adequately explain the cross-sectional distribution of equity returns. Despite these concerns, the standard practice of using the CAPM is followed here. That decision does not have a major impact on the results, because the discount rate is assumed to remain unchanged by the Intel press release. To an extent, therefore, the discount rate becomes a plug that is adjusted to equate the projected cash flows with the market value of Intel's equity.
Application of the CAPM requires estimates both of Intel's beta and of the equity risk premium (ERP). Starting with beta, Intel is not immune from the measurement error problems that typically arise when applying the CAPM. Depending on the sample period, observation interval, estimation procedure, and adjustment algorithm, estimates of beta range from a high of 1.65 (estimated using one year of daily data ending one week prior to the press release) to a low of 1.05 (estimated by BARRA using the company's proprietary technology). For the purposes of this paper, a beta of 1.25 is selected. This is close to Bloomberg's adjusted beta of 1.26 using two-years of weekly data for the 104 weeks preceding the press release.
The final piece of the discount rate is the equity risk premium. As noted in Cornell (1999), the values of technology companies, including Intel, are highly sensitive to the choice of the ERP for two reasons. First, they have betas greater than one so that different choices of the ERP produce magnified changes in the discount rate. Second, successful technology companies are characterized by cash flows that are expected to grow rapidly. The large expected cash flows in the distant future make the present value of the stream very sensitive to changes in the discount rate.
To calibrate the model, an ERP of 3.9% is selected. Using a 3.9% ERP produces a discount rate that equates the present value of the pre-announcement cash flow projections in Table II to the market value of Intel on September 21, 2000, as shown at the bottom of the table. It should be noted that an ERP of 3.9% is well below the long-run average difference between the returns on the S&P 500 and 20-year Treasury bond returns. As reported by Ibbotson Associates (2000), the average difference from 1926 to 1999 was 7.8%. However, an ERP of 3.9% is consistent with recent research. Using a forward looking, clean accounting approach, Claus and Thomas (1999) estimate the ERP to be approximately 3%. Similarly, Fama and French (2000) present evidence that the risk premium in recent years has been about 3.4%. Finally, Cornell (1999) argues that an ERP in the neighborhood of 3% is required to rationalize the level of the stock market at the end of 1999. As noted previously, a slightly higher ERP could be used if the termi nal growth rate were raised.
With all the pieces in place, it is worthwhile to take a step back and examine the broad characteristics of the assumptions necessary to calibrate the model to Intel's stock price on September 21, 2000. The most striking fact that emerges from Table II is how bullish the assumptions have to be. Even using a relatively low discount rate (due to choosing a beta toward the bottom of the range and an ERP of 3.9), Intel has to maintain a growth rate of nearly 20% in revenues for the next ten years to justify the stock price on September 21, 2000. This is higher than the growth rate averaged over the previous five years when the company was much smaller. It implies an increase in sales of more than five times from $34 billion to $162 billion in just nine years. (In terms of constant dollars, the increase is only to $119 billion, assuming an inflation rate of 3.5%.) Furthermore, this growth must be achieved while maintaining gross margins of 62.5%, a level higher than that achieved in the past. Operating margins al so must be higher, on average, over the next ten years than they were during the previous five years. Given Intel's massive size as of September 2000, achieving such rapid and profitable growth is a tall order. Furthermore, these bullish assumptions are required to calibrate the model to a stock price of $61.50. If the model were calibrated to the $74 price that prevailed at the end of August 2000, then the assumptions would have to be more optimistic.
B. Cash Flow Forecast Revisions and Stock Price Changes
If Intel's spokesman is correct in stating that the press release should not be interpreted as signaling a fundamental change in the company's business outlook, then it is reasonable to conclude that the path of future revenues would respond to the shortfall in one of three ways. First, the revenue shortfall could be purely temporary in the sense that buyers were postponing, rather than permanently canceling, orders. In that case, revenues should converge back toward the pre-shock growth path as postponed orders are placed. Second, the shock could lead to a permanent loss of short-term unit sales. In that case, the post-shock growth path would lie below the pre-shock path by an amount equal to the revenue associated with the lost sales. Finally, the shock could lead to a permanent reduction in the size of the business in percentage terms. Under this scenario, the difference between pre- and post-shock revenue grows over time in terms of dollars.
By applying the DCF model and using analysts' forecasts, the three adjustment scenarios can be translated into implied stock price changes. The calculations are presented in Table III. The projections for the first two years in Table III are median values taken from post-announcement analyst reports. Beyond the two-year horizon, all the revenue ratios used earlier in Table II are assumed to be applicable despite the lower projected revenues. This implies that, by the end of two years, Intel could have adjusted to a lower sales growth path in a fashion that allows the company to maintain its margins. The table shows the specific calculations for one future revenue path. For that path, post-announcement revenues in all the years after 2001 are assumed to fall short of the pre-announcement growth path by the same percentage they fell short of in 2001.
Results for all three scenarios are presented in Table IV. Each of the scenarios starts with analyst forecasts for 2000 and 2001. For the temporary revenue drop, it is assumed that the revenue shortfall is reduced linearly in each year until it returns to the pre-announcement growth path in 2009. The permanent revenue drop assumes that sales lost in the first two years are never recovered so that dollar revenues in all years after 2001 are below the preannouncement projections by the analyst revision for 2001. The permanent drop in percent (the calculations shown in Table III) assumes that the ratio of pre-announcement to postannouncement revenues remains constant at 2001 level.
Table IV demonstrates that none of the scenarios produces anything close to the observed 30% drop in Intel's stock price. If the decline in revenues is temporary, the implied drop is less than 1%. Even if the decline in revenues is assumed to be permanent in percentage terms, the implied stock price drop is only 4.5%.
Table IV makes it clear that, to explain the decline in Intel's price, it is necessary to assume that investors significantly changed their views regarding the long-run growth potential of the company. The final column of Table IV and the multi-year projections of Table V present one hypothetical revenue growth path consistent with the drop in the stock price. That path implies a dramatic change in Intel's long-run business outlook. Average growth over the next ten years falls from nearly 20% to well under 15%. As a result, by the final year of the forecast, 2009, revenues are over $50 billion below the projections necessary to calibrate the model prior to the press release. This is the type of major reassessment of a company's prospects that may result from significant product problems, changes in government regulation, major innovations by competitors, or a shift in the market away from the company's products to a different technology. However, it is virtually impossible to see how such a reassessment coul d have been precipitated by the September 21, 2000 press release. An announcement of a cyclical slowing of revenue growth in Europe does not amount to a major revision in the company's long-run outlook for its fundamental business.
The possibility of partial anticipation simply compounds the mystery. The stock price response should be attenuated to the extent that the information in the press release was anticipated because of Kumar's warnings. This means that, without partial anticipation, the drop in response to the press release would have been even larger than 30%.
In short, the analysis indicates that Intel, despite being one of the most carefully followed and widely analyzed companies, could not have been efficiently priced, both before, and after the press release. This finding supports Summers' (1986) view that stock prices can deviate markedly from fundamental value over prolonged periods of time. It also has clear implications for corporate managers. If stock prices can deviate widely from informed estimates of fundamental value, then it is incumbent upon senior managers of the firm to develop and be familiar with a fundamental valuation model of their company. Such knowledge would prove useful not only in dealing with the investment community, but also in making important financial decisions, such as repurchasing the stock, issuing new stock, or using the company's stock for acquisitions. Rather than bemoaning market overreaction or misvaluation, managers should strive to exploit it. In fact, the literature on the response of stock prices to new issues is consis tent with the view that astute companies are already doing so. 
For obvious reasons, Intel's financial managers were unwilling to comment on what steps the company might take if management concluded that the stock was under- or over-valued. They did note that, at times in the past, the company had repurchased shares in conjunction with the public statements that management felt that the stock was too low. Nonetheless, with respect to exercise of options and sales of shares by executives, the company follows a strict sunshine policy by which transactions occur on a pre-announced and generally routine basis. Any attempt by insiders to "time" purchases and sales, depending on the level of the stock price, is highly discouraged.
The conclusion that the quanta of information contained in the press release of September 21, 2000 is insufficient to explain the 30% drop in Intel's stock price does not necessarily imply that the stock was undervalued after the drop. It is equally likely--indeed it is more likely--that the company was overpriced before the release. Recall that Table II reveals that, to support its pre-announcement stock price, Intel's revenues had to grow to a massive $170 billion by 2009 assuming an equity risk premium of 3.9%. Had an ERP closer to the historical average been employed, revenues in 2009 would have to be expected to grow well over $250 billion. Similarly, sales would have to rise over $200 billion if the model were calibrated to the end of August 2000 prices. In fact, it appears as if the tech boom that pushed the dot.com stocks to astronomical levels also affected more mundane technology companies like Intel that had long histories of positive earnings and cash flow.
The behavior of the stock price following the announcement is consistent with the overvaluation interpretation. As Figures 1 and 2 show, Intel never recovered from the drop. In fact, from late September 2000 until the end of the year, the company's stock price drifted lower. This added decline, however, should not be interpreted as some sort of delayed reaction to the announcement. In the Fall of 2000, significant new information arrived indicating that demand for PCs and technology products, in general, was weakening. In response, the entire NASDAQ market dropped. From October 1, 2000 until December 31, 2000, Intel fell about 25% and the index dropped about 30%. It also turned out that the technology slump was sufficiently steep that Intel's warning turned out to be optimistic. Third quarter revenues were flat, rather than growing 3 to 5% as previously warned.
If the stock were overvalued prior to the announcement, as the foregoing seems to imply, two obvious questions arise. Why did "arbitrage" fail to reduce the mispricing, and why were analysts so strongly recommending the stock? With respect to the lack of arbitrage, the critical point is that the stock was mispriced relative to theoretical fundamentals, not other financial assets.  Consequently, taking a short position in the stock is subject to all the risks and costs eloquently elucidated by Shleifer and Vishny (1997). As evidence of how great these costs and risks can be, Schill and Zhou (2000), Cornell and Liu (2001), and Lamont and Thaler (2001) report that, during this time period, there were numerous instances when the market values of parent companies were less than the market values of their holdings of publicly traded subsidiaries, particularly in situations where the subsidiaries were technology companies. Furthermore, as D'Avolio (2000) describes, the market for borrowing stock to sell short is far from perfect. Finally, a potential short seller of Intel would be shorting a stock that had been running up sharply for the last nine months and was one of the most highly recommended stocks on the street. Such positive information would presumably tend to undermine the confidence of an investor who felt the stock was overvalued.
That brings the focus back to analysts. The most basic mystery is why analysts were more enthusiastically recommending purchase of Intel at the end of August 2000 at a price of nearly $75, rather than after the price had dropped to $40 in September 2000. This situation is reminiscent of the results of Shiller's (1989) survey of institutional investors at the time of the crash of October 1987. Shiller found that, prior to the crash, a majority of the investors felt that the market was overvalued, but they were not selling because the recent price action had been favorable. Perhaps analysts had been taking a similar stance with regard to Intel. Even though the late summer price was too high to justify with reasonable fundamentals, analysts were still highly recommending purchase of Intel on the basis of the company's reputation and the huge run-up experienced from January through September 2000. This suggests that it would be informative to investigate the role played by analysts and their recommendations in t he response of Intel's stock price to the September 21, 2000 press release.
III. The Role of Analysts
Stock market analysts play a critical role in collecting, analyzing, and transmitting corporate information for investors. For the most part, academic research supports the view that they perform these functions quite well. For instance, there is a large amount of literature which indicates that analysts' earnings forecasts are at least as accurate as statistical models in predicting future earnings.  That literature, however, focuses on short-term forecasts because most analyst reports do not contain point estimates of earnings beyond two years. Furthermore, more recent research, including DeBondt and Thaler (1990), Easterwood and Nutt (1999), and Lim (2000), finds evidence of bias in analyst forecasts.
There are some additional shortcomings in the work of analysts that are highlighted by the Intel experience. The most notable shortcoming is that virtually none of the 28 analyst reports on Intel examined for this study contained a DCF valuation analysis. Furthermore, the few reports that did refer to a DCF value did not present enough information to understand the basis for the calculations. What makes this surprising is that virtually every report contained a recommendation regarding potential purchase or sale of the stock. The mystery is how a purchase recommendation could be offered without an explicit comparison between price and estimated value. From a valuation perspective, attractive securities are those whose price is less than the present value of the expected future cash flows discounted at the appropriate risk-adjusted rate. Unlike bond ratings, which attempt to assess the credit quality of a company and, as a result, can be prepared independent of valuation considerations, equity recommendations are precisely a weighing of price versus fundamental value. One would presume, therefore, that analyst reports designed to help investors make investment decisions would focus on estimation of fundamental value. That is not the case. Short-run financial performance, not fundamental value, is the focus of the reports on Intel. Furthermore, discussions of fundamental value in the reports are typically vague and nebulous, and rarely involve the presentation of a precise, quantitative model that can be dissected and critiqued. In addition, in virtually every case, the financial projections presented are limited to forecast horizons of two years or less. Such short horizons are not sufficient to construct a reasonable DCF valuation model. Because of the lack of explicit valuation models, it is difficult to understand how the analysts arrive at their estimates of fundamental value and to discern how and why those estimates might change in response to events such as Intel's press release.
Failure to present explicit valuation models also makes it difficult to determine how analysts make a distinction between permanent and temporary shocks. For instance, following Intel's press release, virtually every analyst report focused intensely on how forecasts of revenues and earnings should be revised for the forthcoming quarter and fiscal year. This short-run focus does not help an investor answer the critical investment question--does the press release represent a fundamental change in Intel's business outlook and, therefore, its value? More specifically, was the shock to revenue temporary, permanent in units, permanent in percent, or indicative of a basic shift in the long-run demand for the company's products? As the previous section made clear, the valuation impacts are dramatically different depending upon which interpretation is accepted.
When analysts' purchase recommendations are considered, a more puzzling valuation issue arises. In the previous two sections of this paper, evidence is presented supporting the view that the drop in Intel's stock price was too large to be explained by the information conveyed by the September 21, 2000 press release. If the stock price drop in response to the announcement was indeed "too large" in light of the underlying fundamentals, then one would expect that, on average, analysts would have upgraded their investment recommendations following the announcement because Intel was now more attractively priced relative to its fundamental value. Even if the market was efficient, so that the price drop was justified by the announcement, it should still be the case that as many analysts would upgrade their recommendations as would downgrade them because price and value would fall by equal amounts.
The actual recommendation revisions for Intel are markedly at odds with both of the foregoing predictions. In the week following the Intel's press release, Bloomberg reported that 26 of the analysts who follow the company revised their recommendations. Of those 26, 12 decided to revise their previous recommendation. All 12 lowered their recommendations. Four of the analysts lowered their recommendation by more than one category. One dropped Intel from a strong-buy to under-perform. This is a striking finding. Prior to the announcement, investors had to pay over $60 dollars to buy a share of Intel. Following the announcement, the same company, in the same competitive environment, with the same technology and management, could be acquired for just over $40 dollars per share. Nonetheless, analysts, on average, found the company to be less attractive at the lower price. The results are more dramatic if one goes back to the end of August 2000. At that time, prior to Ashok Kumar's statements, analysts were giving Intel even more positive buy recommendations despite the fact that the stock price was over $74.
This finding strongly suggests that analysts' recommendations are based on something other than a comparison of market price with fundamental value. One possibility is that analysts are in some sense rating the company, rather than the investment. When bad news is announced, they "downgrade" the firm much in the same fashion that bond-rating agencies do. The problem with this hypothesis is that it makes no economic sense. Bond rating agencies are trying to assess credit quality (i.e., the probability of repayment.) Such an assessment can be made independent of price. Equity investment decisions, however, do not depend on a company's "quality," but depend on its quality in relation to price instead.
A political element may come into play as well. When a company's stock price is rising sharply and the firm is being touted in the financial press as a star, it seems difficult for an analyst to downgrade the company. Conversely, when the company's stock price falls sharply following a negative press release, such as Intel's, upgrading a recommendation on the company may appear foolish to those who do not understand the underlying valuation issues. This unfortunately includes much of the financial media. Furthermore, investors who are not following the stock closely may become confused about timing and think that the analyst upgraded the stock prior to the drop. Such a false belief would sully an analyst's reputation. Indeed, it is possible analysts fear that recommending purchase of a stock that has just dropped 30% results in a kind of guilt by association. Another political complication arises from the fact that analysts do more than advise investors. Because most analysts are employed by investment banks , numerous commentators and researchers, including Dugar and Nathan (1995) and Lin and McNichols (1998), suggest that analyst recommendations are tainted because investment banks compete for the business of the firms that the analysts are following.
Whatever the explanation, the Intel experience reveals that, in the case of at least one company, analyst recommendations are highly procyclical. As bad news is released and the price of a company's stock declines, analysts downgrade the company. The reverse effect occurs on the release of good news. As a result, a positive feedback loop of the type described by Shiller (1989) develops. If analyst recommendations affect investor behavior, then such a feedback loop could exacerbate price movements in both directions. This issue is of sufficient import that it invites future research to determine whether the Intel experience is common.
The failure to develop a DCF model also causes analysts to overlook, or at least downplay, the role of expected returns in investor decision-making. In the case of Intel, it was necessary to assume an ERP of 3.9% and an associated expected return of 11.03% in order to calibrate the model prior to the announcement. Accordingly, analysts should have been telling investors, to whom they recommended purchase of the stock, that if the company meets lofty growth projections, such as those presented in Table II, then investors could expect to earn returns of only 11%. For returns to exceed that amount, or, in other words, for Intel to outperform the market and justify a buy recommendation, future cash flows would have to be greater than the levels projected in Table II.
As a postscript, it should be noted that as time passed, and more information became available, the analysts did get it right. By the end of 2000, with Intel warning that third quarter revenues would be flat, every report had revised downward future projections of earnings. These lower projections, furthermore, when put into DCF valuation models produced prices in the vicinity of the year-end market price of $35. This does not explain, however, how a largely short-term warning led to such an immense change in Intel's value. The September 21, 2000 announcement did not contain the information that was to come later. It is almost as if the market for Intel's stock was primed for a drop and that the announcement provided the catalyst. On this point, Cornell, Conrad, and Landsman (2000) present evidence that, as the level of the market rises, individual stocks become more sensitive to bad news. However, developing an understanding of how such a catalyst could work, or how prices could rise to the point that would allow such a catalyst to operate, are problems for future research.
IV. Summary and Conclusions
This paper has investigated the stock market response to Intel's September 21, 2000 press release in which the company announced that its revenue growth for the third quarter would be lower than analysts' expectations. By examining the release, in conjunction with analyst reports and by employing a DCF valuation model, it is argued that the press release did not contain sufficient information about the long-run business outlook for Intel to justify the market's 30% drop. This does not mean, however, that the market overreacted in the sense of unrealistically depressing Intel's stock price. On the contrary, the analysis presented here suggests that the misvaluation existed prior to the announcement. The DCF valuation model shows that by even using a relatively low beta and a small equity market risk premium, Intel had to grow profitably at an implausibly rapid pace, from an already high base, to justify the pre-announcement stock price. The conclusion that emerges, consequently, is that the press release acte d as a kind of catalyst which caused movement toward a more rational price, even though the release itself did not contain sufficient long-run valuation information to justify that movement.
This explanation is buttressed by the puzzling procyclical nature of analysts' purchase recommendations. At the end of August 2000, when the stock price was $75, Intel was one of the most highly recommended stocks in the Bloomberg analyst database. By the end of September 2000, with the price less than $40, analysts had significantly downgraded their purchase recommendations. Such a reaction, on the part of analysts, would make sense only if the information that arrived in the interim was sufficient to cause the estimated fundamental value of the company to fall by more than the $35 drop in the price. That conclusion is not supported by the analysis of the press release. The information provided by the release was short-term in nature and limited to a certain geographic region.
What this suggests is that historical price performance itself plays a critical role in influencing analyst recommendations. Strong buy recommendations were maintained into September 2000, despite the fact that the DCF model shows that the high prices then were hard to justify because the stock price had run up 85% since January 2000. Similarly, it appears that many analysts reduced their recommendations following the press release precisely because the price had fallen 30%. Such behavior on the part of analysts, if it affects investors, would tend to exacerbate stock price volatility. This is clearly an exciting area for future research.
More generally, the paper points to a significant shortcoming in analyst reports. The most difficult task for investors is assessing the long-run implications of new information for a company's fundamental business. For instance, how does the Intel press release affect the long-run revenue growth stream and, therefore, the value of the company? By failing to focus on fundamental value, and by not presenting explicit DCF valuation models, analysts short change investors. Furthermore, if analysts develop explicit valuation models, then the procyclical nature of recommendations discussed above might well disappear because the recommendations would be based on a quantitative comparison of estimated value to price. Of course, such an approach would carry added risk for analysts because specific models, while more useful and precise than vague discussions, are also more open to criticism if the projections on which they are based fail to materialize.
Finally, the results reported here have important implications for financial managers. The huge price movement on relatively minor information implies that Intel could not have been efficiently priced both before, and after, the press release. This supports Summers' (1986) contention that stock prices can deviate markedly from fundamental value over prolonged periods of time, even for the most widely followed and carefully analyzed companies. If that is so, then it is incumbent upon senior managers of the firm to develop fundamental valuation models for their companies. Such models would prove useful, not only in dealing with the investment community, but also in making important financial decisions, such as issuing or repurchasing the company's stock.
I would like to thank Elizabeth James and John Haut of Charles River Associates for research support. Andy Bryant and Doug Lusk of Intel, as well as, Michael Brennan, John Cochrane, Wayne Landsman, Eduardo Schwartz, the Editors, and an anonymous referee offered helpful comments. Of course, the remaining errors are mine.
(*.) Bradford Cornell is a Professor of Finance at the University of California, Los Angeles.
(1.) See, for instance, the discussion in Shiller (2000).
(2.) The Bloomberg index averages analyst recommendations on a scale of 1 through 5. For a company to receive a rank of 5, every analyst must give the stock the strongest possible purchase recommendation.
(3.) Given Intel's size, a significant fraction of the drop in the S&P 500 and the NASDAQ Index on September 22, 2000 was due to the company's inclusion in those indices.
(4.) An alternative would be to use a real options model like the type developed by Schwartz and Moon (2000).
(5.) As Cornell (1999) reports, in corporate valuation contexts, long-term Treasuries are chosen because their duration more closely matches that of the assets being valued, despite the fact that they are not risk-free securities over short horizons.
(6.) See, for instance, Masulis and Korwar (1986) or Mikkelson and Partch (1986).
(7.) Of course, Intel is valued relative to other stocks, but there is no straightforward arbitrage relation of the type that exists between underlying securities and derivatives. To say that Intel is cheap or expensive relative to other common stocks requires a fundamental valuation model.
(8.) See, for instance, Brown and Rozeff (1978), O'Brien (1988), and Givoly and Lakonishok (1984).
Brown, L.D. and M.S. Rozeff, 1978, "The Superiority of Analyst Forecasts As Measures of Expectations: Evidence from Earnings," Journal of Finance 33, 1-16.
Brunner, R.F., K.M. Eades, R.S. Harris, and R.C. Higgins, 1998, "Best Practices in Estimating the Cost of Capital: Survey and Synthesis," Financial Practice and Education (Spring/Summer), 13-28.
Chopra, N., J. Lakonishok, and J.R. Ritter, 1992, "Measuring Abnormal Performance: Do Stocks Overreact?" Journal of Financial Economics 31, 235-268.
Claus, J. and J. Thomas, 1999, "The Equity Risk Premium is Much Lower than You Think: Empirical Estimates from a New Approach," Columbia University Working Paper.
Cornell, B., 1993, Corporate Valuation, McGraw Hill, New York, NY.
Cornell, B., 1999, The Equity Risk Premium, John Wiley & Sons, New York. NY.
Cornell, B. and Q. Liu, 2001, "The Parent Company Puzzle: When is the Whole Worth Less than One of the Parts," University of California, Los Angeles Working Paper.
Cornell, B., J. Conrad, and W. Landsman, 2000, "When is Bad News Really Bad News?" University of North Carolina Working Paper.
D'Avolio, G., 2000, "The Market for Borrowing Stock," Harvard University Working Paper.
DeBondt, W.F.M. and R. Thaler, 1985, "Does the Stock Market Overreact?" Journal of Finance 40, 793-805.
DeBondt, W.F.M. and R.M. Thaler, 1990, "Do Security Analysts Overreact?" American Economic Review 80, 52-57.
Dugar, A. and S. Nathan, 1995, "The Effect of Investment Banking Relationships on Financial Analysts' Earnings Forecasts and Recommendations," Contemporary Accounting Research (Fall), 131-160.
Easterwood, J. and S. Nutt, 1999, "Inefficiency in Analysts' Earnings Forecasts: Systematic Misreaction or Systematic Optimism?" Journal of Finance 54, 1777-1797.
Fama, E.F., 1998, "Market Efficiency, Long-Term Returns and Behavioral Finance," Journal of Financial Economics 49, 283-306.
Fama, E.F. and K.R. French, 1992, "The Cross-Section of Expected Returns," Journal of Finance 47, 427-465.
Fama, E.F. and K.R. French, 1996, "The CAPM Is Wanted, Dead or Alive," Journal of Finance 51, 1947-1958.
Fama, E.F, and K.R. French, 2000, "The Equity Risk Premium," University of Chicago Working Paper.
Givoly, D. and J. Lakonishok, 1984, "The Quality Of Analysts' Forecasts Of Earnings," Financial Analyst Journal 40, 40-47.
Graham, J.R. and C.R. Harvey, 2001, "The Theory and Practice of Corporate Finance: Evidence from the Field," Journal of Financial Economics 60, (forthcoming).
Ibbotson Associates, 2000, Stocks, Bonds, Bills and Inflation, Ibbotson Associates, Chicago, IL.
Kaplan, S.N. and R.S. Ruback, 1995, "The Valuation of Cash Flow Forecasts: An Empirical Analysis," Journal of Finance 50, 1059-1093.
Lamont, O.A., and R.H. Thaler, 2001, "Can the Market Add and Subtract? Mispricing in Tech Stock Carve Outs," University of Chicago Working Paper.
Lim, T., 2001, "Rationality and Analysts' Forecast Bias," Journal of Finance 56, 369-385.
Lin, H. and M. McNichols, 1998, "Underwriting Relationships, Analysts' Forecasts and Investment Recommendations," Journal of Accounting and Economics 25, 101-127.
Masulis, R.W. and A.N. Korwar, 1986, "Seasoned Equity Offerings: An Empirical Investigation," Journal of Financial Economics 15, 91-118.
Mikkelson, W.H. and M.M. Partch, 1986, "Valuation Effects of Security Offerings and the Issuance Process," Journal of Financial Economics 15, 31-60.
O'Brien, P., 1988, "Analysts' Forecasts of Earnings Expectation," Journal of Accounting and Economics 10, 53-83.
Ruback, R., 2000, "Capital Cash Flows: A Simple Approach to Valuing Risky Cash Flows," Harvard University Working Paper.
Schill, M.J. and C. Zhou, 2000, "Pricing an Emerging Industry: Evidence from Internet Subsidiary Carve-Outs," University of California Riverside Working Paper.
Schwartz, E.S. and M. Moon, 2000, "Rational Pricing of Internet Companies Revisited," University of California, Los Angeles Working Paper.
Shiller, R.J., 1989, Market Volatility, MIT Press, Cambridge, MA.
Shiller, R.J., 2000, Irrational Exuberance, Princeton University Press, Princeton, NJ.
Shleifer, A. and R.W. Vishny, 1997, "The Limits of Arbitrage," Journal of Finance 52, 25-55.
Summers, L.H., 1986," Does the Stock Market Rationally Reflect Fundamental Values?" Journal of Finance 41, 591-600.
Zarowin, P., 1989, "Does the Stock Market Overreact to Corporate Earnings Information?" Journal of Finance 44, 1385-1400.
Price Changes for Selected Companies and Indices September 22, 2000 to September 26, 2000
This figure shows the percentage drop in the stock price of Intel during the four day period from September 22 through September 26, 2000 compared to the performance of the S&P 500 index, the NASDAQ index and a group of actively traded computer and chip companies.
Intel -29.56% NASDAQ -3.65% S&P 500 -1.51% IBM -3.83% Dell -6.43% Microsoft -0.59% AMD -6.13% Compaq 1.52% Hewlett- Packard -5.82% Gateway -3.41% Note: Table made from bar graph Historical Financial Data for Intel This table presents historical financial data for Intel taken from the company's 10K reports 1995 1996 1997 1998 1999 In Terms of Dollars Net revenues (millions) 16,202 20,847 25,070 26,273 29,389 Cost of sales 7,811 9,614 9,945 12,088 11,836 Research and development 1,296 1,808 2,347 2,509 4,264 SG&A 1,843 2,322 2,891 3,076 3,872 Income from operations 5,252 7,103 9,887 8,600 9,417 Marginal tax rate 33.0% 33.0% 33.0% 33.0% 33.0% Tax on operating income 1,733 2,344 3,263 2,838 3,108 NOPAT 3,519 4,759 6,624 5,762 6,309 Depreciation 1,371 1,888 2,192 2,807 3,186 Increase in working capital 1,700 (200) (320) 40 (380) Capital expenditures 3,550 3,024 4,501 3,557 3,403 Cash flow from operations (360) 3,823 4,635 4,972 6,472 As a Percent of Revenues Net revenues (millions) 100.0% 100.0% 100.0% 100.0% 100.0% Cost of sales 48.2% 46.1% 39.7% 46.0% 40.3% Gross Margin 51.8% 53.9% 60.3% 54.0% 59.7% Research and development 8.0% 8.7% 9.4% 9.5% 14.5% SG&A 11.4% 11.1% 11.5% 11.7% 13.2% Income from operations 32.4% 34.1% 39.4% 32.7% 32.0% NOPAT 21.7% 22.8% 26.4% 21.9% 21.5% Depreciation 8.5% 9.1% 8.7% 10.7% 10.8% Change in working capital 10.5% -1.0% -1.3% 0.2% -1.3% Capital expenditures 21.9% 14.5% 18.0% 13.5% 11.6% Cash flow from operations -2.2% 18.3% 18.5% 18.9% 22.0% Growth Rates Net revenues (millions) 29% 20% 5% 12% Cost of sales 23% 3% 22% -2% Research and development 40% 30% 7% 70% SG&A 26% 25% 6% 26% Income from operations 35% 39% -13% 10% NOPAT 35% 39% -13% 9% Depreciation 38% 16% 28% 14% Change in working capital -112% 60% -113% -1050% Capital expenditures -15% 49% -21% -4% Cash flow from operations NA 21% 7% 30%
Discounted Cash Flow Valuation Model Calibrated to Pre-Announcement Intel Stock Price
This table presents a discounted cash flow model calibrated to Intel's stock price on September 21, 2000. The financial projections are based on analysts' reports for the first five years. Beyond five years, the projections for all individual line items are based on the revenue forecast. Revenues, in turn, are assumed to grow at a declining rate up to the terminal horizon as shown in the table. The terminal growth rate and the equity risk premium are used as "plugs" to equate the discounted cash flow value with Intel's stock price on September 21.
2000 [*] 2001 [*] Net revenues (millions) 35,771 43,000 Cost of sales 12,895 15,000 Research and development 4,002 4,640 SG&A 5,287 6,525 Income from operations 13,587 16,835 Marginal tax rate 33.0% 33.0% Tax on operating income 4,484 5,556 NOPAT 9,103 11,279 Depreciation 3,219 3,870 Change in working capital 358 430 Capital expenditures 5,992 5,375 Cash flow from operations 5,973 9,344 Present value 5,819 8,199 Terminal growth rate 6.00% PV of cash flow (billions) Risk-free rate 6.15% PV of terminal value Beta 1.25 Excess cash and short-term Equity risk premium 3.90% securities WACC 11.03% 2002 2003 2004 2005 Net revenues (millions) 51,600 61,920 74,304 89,165 Cost of sales 19,350 23,220 27,864 33,437 Research and development 5,418 6,502 7,802 9,362 SG&A 6,966 8,359 10,031 12,037 Income from operations 19,866 23,839 28,607 34,328 Marginal tax rate 33.0% 33.0% 33.0% 33.0% Tax on operating income 6,556 7,867 9,440 11,328 NOPAT 13,310 15,972 19,167 23,000 Depreciation 4,644 5,573 6,687 8,025 Change in working capital 516 619 743 892 Capital expenditures 6,450 7,740 9,288 11,146 Cash flow from operations 10,988 13,186 15,823 18,988 Present value 8,684 9,386 10,145 10,965 Terminal growth rate 106 Risk-free rate 297 Beta Equity risk premium 13 WACC 2006 2007 2008 Net revenues (millions) 106,106 125,205 146,490 Cost of sales 39,790 46,952 54,934 Research and development 11,141 13,147 15,381 SG&A 14,324 16,903 19,776 Income from operations 40,851 48,204 56,399 Marginal tax rate 33.0% 33.0% 33.0% Tax on operating income 13,481 15,907 18,612 NOPAT 27,370 32,297 37,787 Depreciation 9,550 11,268 13,184 Change in working capital 1,061 1,252 1,465 Capital expenditures 12,733 15,025 17,579 Cash flow from operations 23,126 27,288 31,928 Present value 12,029 12,784 13,472 Terminal growth rate Total firm value Risk-free rate Value of debt Beta Value of Equity Equity risk premium Shares outstanding WACC Implied share price 2009 Terminal Net revenues (millions) 169,929 Cost of sales 63,723 Research and development 17,842 SG&A 22,940 Income from operations 65,422 Marginal tax rate 33.0% Tax on operating income 21,589 NOPAT 43,833 Depreciation 15,294 Change in working capital 1,699 Capital expenditures 20,391 Cash flow from operations 37,036 39,258 Present value 14,076 Terminal growth rate 415 Risk-free rate 1 Beta 414 Equity risk premium 6.71 WACC $ 61.77 Operating Assumptions 2000 [*] 2001 [*] 2002 2003 Net revenues (growth rate) 21.7% 20.2% 20.0% 20.0% Cost of sales (% of revenue) 36.0% 34.9% 37.5% 37.5% Research and development 11.2% 10.8% 10.5% 10.5% SG&A 14.8% 15.2% 13.5% 13.5% Income from operations 38.0% 39.2% 38.5% 38.5% Marginal tax rate Tax on operating income NOPAT 25.4% 26.2% 25.8% 25.8% Depreciation 9.0% 9.0% 9.0% 9.0% Change in working capital 1.0% 1.0% 1.0% 1.0% Capital expenditures 16.8% 12.5% 12.5% 12.5% Cash flow from operations 16.7% 21.7% 21.3% 21.3% Operating Assumptions 2004 2005 2006 2007 2008 Net revenues (growth rate) 20.0% 20.0% 19.0% 18.0% 17.0% Cost of sales (% of revenue) 37.5% 37.5% 37.5% 37.5% 37.5% Research and development 10.5% 10.5% 10.5% 10.5% 10.5% SG&A 13.5% 13.5% 13.5% 13.5% 13.5% Income from operations 38.5% 38.5% 38.5% 38.5% 38.5% Marginal tax rate Tax on operating income NOPAT 25.8% 25.8% 25.8% 25.8% 25.8% Depreciation 9.0% 9.0% 9.0% 9.0% 9.0% Change in working capital 1.0% 1.0% 1.0% 1.0% 1.0% Capital expenditures 12.5% 12.5% 12.0% 12.0% 12.0% Cash flow from operations 21.3% 21.3% 21.8% 21.8% 21.8% Operating Assumptions 2009 Terminal Net revenues (growth rate) 16.0% Cost of sales (% of revenue) 37.5% Research and development 10.5% SG&A 13.5% Income from operations 38.5% Marginal tax rate Tax on operating income NOPAT 25.8% Depreciation 9.0% Change in working capital 1.0% Capital expenditures 12.0% Cash flow from operations 21.8% Post-Announcement Revenue Path Consistent with the Stock Price Decline This table uses the discounted cash flow model to calculate the decline in Intel's stock price implied by a permanent drop in revenues in percentage terms. 2000 [*] 2001 [*] 2002 Net revenues (millions) 34,621 41,000 49,200 Cost of sales 12,857 15,643 18,450 Research and development 4,002 4,560 5,166 SG&A 5,287 6,450 6,642 Income From operations 12,475 14,347 18,942 Marginal tax rate 33.0% 33.0% 33.0% Tax on operating income 4,117 4,735 6,251 NOPAT 8,358 9,612 12,691 Depreciation 3,116 3,690 4,428 Change in working capital 346 410 492 Capital expenditures 5,799 5,125 6,150 Cash flow from operations 5,329 7,767 10,477 Present value 5,191 6,816 8,280 Terminal growth rate 6.00% PV of cash flow (billions) Risk-free rate 6.15% PV of terminal value Beta 1.25 Excess cash and short- term securities Equity risk premium 3.90% WACC 11.03% 2003 2004 2005 Net revenues (millions) 59,040 70,848 85,018 Cost of sales 22,140 26,568 31,882 Research and development 6,199 7,439 8,927 SG&A 7,970 9,564 11,477 Income From operations 22,730 27,276 32,732 Marginal tax rate 33.0% 33.0% 33.0% Tax on operating income 7,501 9,001 10,801 NOPAT 15,229 18,275 21,930 Depreciation 5,314 6,376 7,652 Change in working capital 590 708 850 Capital expenditures 7,380 8,856 10,627 Cash flow from operations 12,573 15,087 18,104 Present value 8,950 9,673 10,455 Terminal growth rate 99 Risk-free rate 283 Beta 13 Equity risk premium WACC 2006 2007 2008 Net revenues (millions) 101,171 119,382 139,677 Cost of sales 37,939 44,768 52,379 Research and development 10,623 12,535 14,666 SG&A 13,658 16,117 18,856 Income From operations 38,951 45,962 53,775 Marginal tax rate 33.0% 33.0% 33.0% Tax on operating income 12,854 15,167 17,746 NOPAT 26,097 30,795 36,030 Depreciation 9,105 10,744 12,571 Change in working capital 1,012 1,194 1,397 Capital expenditures 12,141 14,326 16,761 Cash flow from operations 22,050 26,019 30,443 Present value 11,469 12,190 12,846 Terminal growth rate Total firm value Risk-free rate Value of debt Beta Value of equity Equity risk premium Shares outstanding WACC Implied share price 2009 Terminal Net revenues (millions) 162,025 Cost of sales 60,759 Research and development 17,013 SG&A 21,873 Income From operations 62,380 Marginal tax rate 33.0% Tax on operating income 20,585 NOPAT 41,794 Depreciation 14,582 Change in working capital 1,620 Capital expenditures 19,443 Cash flow from operations 35,313 37,432 Present value 13,421 Terminal growth rate 395 Risk-free rate 1 Beta 394 Equity risk premium 6.71 WACC $ 58.78 2000 [*] 2001 [*] 2002 2003 2004 Operating Asumptions Net revenues (growth rate) 17.8% 18.4% 20.0% 20.0% 20.0% Cost of sales (% of revenue) 37.1% 38.2% 37.5% 37.5% 37.5% Research and development 11.6% 11.1% 10.5% 10.5% 10.5% SG&A 15.3% 15.7% 13.5% 13.5% 13.5% Income from operations 36.0% 35.0% 38.5% 38.5% 38.5% Marginal tax rate Tax on operating income NOPAT 24.1% 23.4% 25.8% 25.8% 25.8% Depreciation 9.0% 9.0% 9.0% 9.0% 9.0% Change in working capital 1.0% 1.0% 1.0% 1.0% 1.0% Capital expenditures 16.8% 12.5% 12.5% 12.5% 12.5% Cash flow from operations 15.4% 18.9% 21.3% 21.3% 21.3% 2005 2006 2007 2008 2009 Operating Asumptions Net revenues (growth rate) 20.0% 19.0% 18.0% 17.0% 16.0% Cost of sales (% of revenue) 37.5% 37.5% 37.5% 37.5% 37.5% Research and development 10.5% 10.5% 10.5% 10.5% 10.5% SG&A 13.5% 13.5% 13.5% 13.5% 13.5% Income from operations 38.5% 38.5% 38.5% 38.5% 38.5% Marginal tax rate Tax on operating income NOPAT 25.8% 25.8% 25.8% 25.8% 25.8% Depreciation 9.0% 9.0% 9.0% 9.0% 9.0% Change in working capital 1.0% 1.0% 1.0% 1.0% 1.0% Capital expenditures 12.5% 12.0% 12.0% 12.0% 12.0% Cash flow from operations 21.3% 21.8% 21.8% 21.8% 21.8% Terminal Operating Asumptions Net revenues (growth rate) Cost of sales (% of revenue) Research and development SG&A Income from operations Marginal tax rate Tax on operating income NOPAT Depreciation Change in working capital Capital expenditures Cash flow from operations
Stock Price Drops Associated with Various Cash Flow Forecast Revisions
This table presents the results of using the discounted cash flow model to calculate the valuation impact of three responses to the September 22, 2000 demand shock: 1) a temporary drop in revenues, 2) a permanent drop in revenues in dollars, and 3) a permanent drop in revenues in percentage terms. The table also shows future revenue forecasts that are consistent with the valuation decline that actually occurred following the announcement.
Pre- Temporary Permanent Permanent Announcement Revenue Revenue Revenue Year Projections Drop Drop in Units Drop in Percent 2000 35,771 34,621 34,621 34,21 2001 43,000 41,000 41,000 41,000 2002 51,600 49,600 49,200 49,200 2003 61,920 60,206 59,920 59,040 2004 74,304 72,875 72,304 70,848 2005 89,165 88,022 87,165 85,018 2006 106,106 105,249 104,106 101,171 2007 125,205 124,634 123,205 119,382 2008 146,490 146,204 144,490 139,677 2009 169,929 169,929 167,929 162,025 Stock Price $ 61.77 $ 61.31 $ 60.66 $ 58.78 Percent Drop 0.74% 1.80% 4.84% Revenue Drop Consistent with Year Stock Price 2000 34,621 2001 41,000 2002 47,767 2003 55,172 2004 63,454 2005 72,344 2006 81,756 2007 91,574 2008 101,657 2009 111,833 Stock Price $ 43.31 Percent Drop 29.89%
Post-Announcement Revenue Path Consistent with the Stock Price Decline
This table presents a complete discounted cash flow valuation for forecasts that are consistent with the post-announcement valuation of Intel.
2000 [*] Net revenues (millions) 34,621 Cost of sales 12,857 Research and development 4,002 SG&A 5,287 Income front operations 12,475 Marginal tax rate 33.0% Tax on operating income 4,117 NOPAT 8,358 Depreciation 3,116 Change in working capital 346 Capital expenditures 5,799 Cash flow from operations 5,329 Present value 5,191 Terminal growth rate 6.00% Risk-free rate 6.15% Beta 1.25 Equity risk premium 3.90% WACC 11.03% 2001 [*] Net revenues (millions) 41,000 Cost of sales 15,643 Research and development 4,560 SG&A 6,450 Income front operations 14,347 Marginal tax rate 33.0% Tax on operating income 4,735 NOPAT 9,612 Depreciation 3,690 Change in working capital 410 Capital expenditures 5,125 Cash flow from operations 7,767 Present value 6,816 Terminal growth rate PV of cash flow (billions) Risk-free rate PV of terminal value Beta Excess cash and short-term securities Equity risk premium WACC 2002 2003 2004 2005 Net revenues (millions) 47,767 55,172 63,454 72,344 Cost of sales 17,912 20,690 23,795 27,129 Research and development 5,015 5,793 6,663 7,596 SG&A 6,448 7,448 8,566 9,766 Income front operations 18,390 21,241 24,430 27,852 Marginal tax rate 33.0% 33.0% 33.0% 33.0% Tax on operating income 6,069 7,010 8,062 9,191 NOPAT 12,321 14,232 16,368 18,661 Depreciation 4,299 4,966 5,711 6,511 Change in working capital 478 552 635 723 Capital expenditures 5,971 6,897 7,932 9,043 Cash flow from operations 10,172 11,749 13,512 15,406 Present value 8,039 8,363 8,664 8,896 Terminal growth rate 83 Risk-free rate 195 Beta 13 Equity risk premium WACC 2006 2007 2008 Net revenues (millions) 81,756 91,574 101,657 Cost of sales 30,658 34,340 38,121 Research and development 8,584 9,615 10,674 SG&A 11,037 12,363 13,724 Income front operations 31,476 35,256 39,138 Marginal tax rate 33.0% 33.0% 33.0% Tax on operating income 10,387 11,635 12,915 NOPAT 21,089 23,622 26,222 Depreciation 7,358 8,242 9,149 Change in working capital 818 916 1,017 Capital expenditures 9,811 10,989 12,199 Cash flow from operations 17,819 19,959 22,156 Present value 9,268 9,350 9,349 Terminal growth rate Total firm value Risk-free rate Value of debt Beta Value of equity Equity risk premium Shares outstanding WACC Implied share price 2009 Terminal Net revenues (millions) 111,833 Cost of sales 41,937 Research and development 11,742 SG&A 15,097 Income front operations 43,056 Marginal tax rate 33.0% Tax on operating income 14,208 NOPAT 28,847 Depreciation 10,065 Change in working capital 1,118 Capital expenditures 13,420 Cash flow from operations 24,374 25,836 Present value 9,264 Terminal growth rate 91 Risk-free rate 1 Beta 291 Equity risk premium 6.71 WACC $ 43.31 Operating Assumptions 2000 [*] 2001 [*] 2002 2003 2004 Net revenues (growth rate) 17.8% 18.4% 16.5% 15.5% 15.0% Cost of sales (% of revenue) 37.1% 38.2% 37.5% 37.5% 37.5% Research and development 11.6% 11.1% 10.5% 10.5% 10.5% SG&A 15.3% 15.7% 13.5% 13.5% 13.5% Income from operations 36.0% 35.0% 38.5% 38.5% 38.5% Marginal tax rate Tax on operating income NOPAT 24.1% 23.4% 25.8% 25.8% 25.8% Depreciation 9.0% 9.0% 9.0% 9.0% 9.0% Change in working capital 1.0% 1.0% 1.0% 1.0% 1.0% Capital expenditures 16.8% 12.5% 12.5% 12.5% 12.5% Cash flow from operations 15.4% 18.9% 21.3% 21.3% 21.3% Operating Assumptions 2005 2006 2007 2008 2009 Net revenues (growth rate) 14.0% 13.0% 12.0% 11.0% 10.0% Cost of sales (% of revenue) 37.5% 37.5% 37.5% 37.5% 37.5% Research and development 10.5% 10.5% 10.5% 10.5% 10.5% SG&A 13.5% 13.5% 13.5% 13.5% 13.5% Income from operations 38.5% 38.5% 38.5% 38.5% 38.5% Marginal tax rate Tax on operating income NOPAT 25.8% 25.8% 25.8% 25.8% 25.8% Depreciation 9.0% 9.0% 9.0% 9.0% 9.0% Change in working capital 1.0% 1.0% 1.0% 1.0% 1.0% Capital expenditures 12.5% 12.0% 12.0% 12.0% 12.0% Cash flow from operations 21.3% 21.8% 21.8% 21.8% 21.8% Operating Assumptions Terminal Net revenues (growth rate) Cost of sales (% of revenue) Research and development SG&A Income from operations Marginal tax rate Tax on operating income NOPAT Depreciation Change in working capital Capital expenditures Cash flow from operations
|Printer friendly Cite/link Email Feedback|
|Date:||Mar 22, 2001|
|Previous Article:||The Debt Maturity Structure of Small Firms.|
|Next Article:||The Efficiency of Equity-Linked Compensation: Understanding the Full Cost of Awarding Executive Stock Options.|