One of the by-products of the technology market bubble in the late 1990s and the subsequent correction after 2000 was a sharp distortion in betas, the traditional measure of risk that companies use to estimate their cost of capital.¹ The telecommunications, media,and technology (TMT) sectors led a sharp rise and fall in the market, making other industries appear almost flat by comparison and thus lowering non-TMT betas. The beta for electric utilities, for example, dropped from 0.6 in 1998 to 0.1 in 2001, falsely suggesting that the sector's risk level had become unrealistically low—and implying a two-percentage-point decline in the industry's cost of capital.² Indeed, most non-TMT sectors saw similar significant declines.

Our analysis from 2003 determined that the decline in betas was a distortion caused by the increased weight of TMT shares in broader market indexes, such as the SS&Pamp;P 500, during the bubble. As a result of this distortion, certain sectors (including automotive, chemicals, consumer goods, and utilities) showed a much lower correlation with market indexes, pushing down those sectors' betas. And although TMT valuations had declined significantly by 2003, betas remained skewed because these calculations typically rely on historical performance over the previous three to five years. We anticipated that this bias in estimates of betas would disappear within a couple of years and advised practitioners to adjust their calculations in the meantime.

We are now nearing the end of the three- to five-year period when calculations of betas would include the bubble years. In comparing our earlier analysis with what has since transpired in the markets, we find that betas for non-TMT sectors in 2006 have indeed risen to their prebubble levels (Exhibit 1), but not because the non-TMT sectors have become more volatile. Rather, their correlation with the market index has increased as the weight of TMT shares in the index has declined from 15 to 20 percent (Exhibit 2).³

These findings mean that the impact on betas of the high-tech market bubble has disappeared for most sectors and companies (Exhibit 3). The corrections of beta estimates that we suggested three years ago are no longer needed. Even so, this episode reminds practitioners to be careful when using estimates of historical risk as a proxy for forward-looking risk. Because we cannot directly measure forward-looking betas, we usually rely on historical data and make an implicit assumption that they are reasonable proxies for the future. This assumption clearly didn't hold during and immediately after the tech bubble. Such a disconnect could happen again, possibly under very different market circumstances.