The stock market bubble might have burst, but the memory lingers. One pernicious aftereffect of the rise and fall in the valuations of telecommunications, media, and technology shares is a continuing distortion of the risk metrics that companies use—for example, to estimate their cost of capital. Unless they correct for this bias, their executives could overvalue investment and acquisition opportunities and overestimate their companies’ historical financial performance.

Most companies estimate their cost of capital using the capital asset-pricing model,¹ in which the nondiversifiable risk of a company is measured by calculating the way its stock price moves, both in speed and volatility, in relation to market indexes. The resulting measure is known as a beta, which is greater than 1.0 if a stock moves, over time, ahead of the market (and is therefore riskier) and less than 1.0 if it tends to move behind the market. A beta of 1.5 foretells a 1.5 percent change in an asset’s return for every 1 percent change in the market’s return. The higher the beta, it is argued, the higher the cost of capital.

Despite volatility in the market during the 20 years before 1998, industry-specific betas—as opposed to the betas of individual companies—were remarkably stable. But during the bubble, betas for many industries appeared to decline significantly: the US utility sector’s beta fell to 0.1, from 0.6, for example, suggesting a 2 percent decline in the industry’s cost of equity.²

Yet these apparent decreases actually reflect the influence of telecom, media, and technology share prices on the indexes during the 1998–2001 bubble and distort the real change in the relative risk borne by companies in other industries. From December 1997 to August 2000, the three sectors together accounted for about 68 percent of the change in the total S&P 500 index, for example. By 2000, the stocks of such companies represented 45 percent of its value and were responsible for as many as 21 percentage points of its 26 percent annual growth (Exhibit 1).

As the three sectors expanded, their influence not only shaped the market as a whole but also skewed calculations of its returns and volatility. The betas of other sectors, such as automotive, chemicals, and utilities, were driven down as the overvalued telecom, media, and technology sectors increasingly dominated the S&P 500 (Exhibit 2). By the end of the bubble, these sectors had such a strong influence on market indexes that theirs were the only rising betas.

Sharp recent declines in telecom, media, and technology valuations suggest that the past three to five years were truly extraordinary. Since July 2002, these sectors’ share of the market’s total capitalization has returned to its average 20-year historical level. Moreover, an analysis carried out in August 2002 shows that as telecom, media, and technology valuations have declined, betas for many other sectors have risen. For a significant number of them, the recent beta estimates are also more closely in line with pre-1998 values.

But in assessing future values for betas, most practitioners look to the equity returns of the recent past—and the most recent three- to five-year averages and correlations of returns to shareholders are of course quite extreme. By excluding the bubble years entirely,³ it is possible to calculate betas that are more consistent with the long-term historical results and indicate more accurately the relative risk borne by companies in other sectors. In the absence of such a correction, data drawn from the bubble may generate artificially low betas for the next couple of years.

With this adjusted approach, we used the most recent 36 months’ returns (as of August 2002) to estimate the betas of a broad cross section of sectors but then eliminated the boom months, from 1998 to 2001. As Exhibit 3 shows, in many cases this approach leads to significantly higher and, we believe, more accurate betas. Eliminating the impact of the boom implies a beta of 0.65 for the food, beverage, and tobacco sector, for example, a figure that is more consistent with its average historical beta from 1990 through 1997—0.85—and that contrasts sharply with the low 0.02 beta that an unadjusted approach produces. Obviously, in some sectors the bubble led first to a decline in betas and then, in recent months, to some degree of recovery.

Since the market has seldom if ever seen the kind of bubble that developed around the telecom, media, and technology sectors in the late 1990s, we are continuing to sort out the impact of that aberration on metrics and economic tools that rely on historical data. Indeed, while betas for other sectors are higher than previously thought, the effect on the cost of capital might have been mitigated by a lower equity risk premium, as earlier research suggests. Some industries in the telecom, media, and technology sectors, however, might find themselves with a fortunate combination of lower betas and a lower equity risk premium.