320  Estimating the Cost of Capital operating assets, and the beta of the tax shields (βtxa) will equal the beta of the unlevered company (βu). Setting βtxa equal to βu eliminates the final term:20 β β β β e u u d D E = + − ( ) Some people further simplify by assuming that the beta of debt is zero. Others use a beta of 0.15 for the debt of investment-grade companies, which is the implied beta based on the spread between investment-grade corporate debt and government debt. Thus, a company’s equity beta equals the company’s operating beta (also known as the unlevered beta) times a leverage factor. As leverage rises, so will the company’s equity beta. Using this relationship, we can convert equity betas into unlevered betas. Since unlevered betas focus solely on operating risk, they can be averaged across an industry, assuming industry competitors have similar operating characteristics. To calculate an industry beta, follow these steps. First, calculate the beta for each company in your peer set and unlever each beta at each company’s debt-to- equity ratio. Remove any outliers, that is, companies where the beta is unusually far away from those of the other companies; these are typically driven by anoma- lous events and are unlikely to recur. Calculate a median beta and an average beta of the sample set. Statistically speaking, the sample average will have the smallest estimation error. However, because small-sample averages are heavily influenced by outliers, we prefer the median beta. The final step is to plot the median indus- try beta over a long period. Look to see if the beta is changing in a predictable way and whether the current beta is the best predictor of future beta for the industry. Examining the Long-Term Trend  To determine the cost of equity for Costco, we create an industry peer beta from a set of discount retailers. We start by estimating the beta for each company using regression analysis (as shown in Exhibit 15.5) and then unlever the results using each company’s respective debt-to-equity ratio. Rather than using beta from a single point in time, we look for trends. Unless there is a discernible trend or dramatic change in the industry, we believe the long-run unlevered beta provides a better estimate of future beta than a single point estimate. Therefore, use the long-run mean when relevering the industry beta to the company’s target capital structure. Exhibit 15.6 presents estimates of levered betas for a selection of industries, including retailers. For Costco, we use an unlevered beta of 0.8, at the low end of the historical range. We use this value because discount retailers have been trading recently at a beta well below 1. To estimate the cost of equity for Costco, we relever the unlevered beta to a peer group debt-to-equity ratio. To lever beta, we use the same capital structure that was used to weight debt and equity in the WACC. The levered beta for Costco equals 0.88 (in practice, we often round 20 See Appendix C for a comprehensive set of equations with different assumptions for the proportion of debt to equity, the beta of debt, and the beta of the tax shields. Estimating the Cost of Equity  321 to one decimal to avoid misleading precision). Using a 4.1 percent risk-free rate and a 5 percent market risk premium, this leads to a cost of equity of 8.5 percent. In some cases, examining the long-term trend will reveal important insight about beta and market prices. During the dot-com boom of the late 1990s, equity markets rose dramatically, but this increase was confined primarily to extremely-large-capitalization stocks and stocks in the telecommunications, media, and technology (commonly known as TMT) sectors. Historically, TMT stocks contribute approximately 15 percent of the market value of the S&P 500. Between 1998 and 2000, this percentage rose to 40 percent. And as the market portfolio changed, so too did industry betas. Exhibit 15.7 presents the median beta over time for stocks outside TMT, such as food companies, air- lines, and pharmaceuticals.21 The median beta dropped from 1.0 to 0.6 as TMT became a dominant part of the overall market portfolio. EXHIBIT 15.6  Unlevered Beta Estimates by Industry Industry Beta range Electric utilities 0.5–0.7 Healthcare providers 0.7–0.8 Integrated oil and gas 0.7–0.8 Airlines 0.7–0.9 Consumer packaged goods 0.8–0.9 Pharmaceuticals 0.8–1.0 Retail 0.8–1.0 Telecom 0.8–1.0 Mining 0.9–1.0 Automotive and assemblers 0.9–1.1 Chemicals 0.9–1.1 IT services, hardware 0.9–1.1 Software 0.9–1.1 Banking 1.0–1.1 Insurance 1.0–1.1 Semiconductors 1.0–1.3 EXHIBIT 15.7  Effect of the Dot-Com Bubble on Beta Beta of non-TMT sectors1 TMT share of aggregate market value1 TMT sector’s share,1 % Beta 1.2 1.0 0.8 0.6 0.4 0.2 0 50 40 30 20 10 0 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 1 TMT = telecommunications, media, and technology.