Chapter 6 Risk Return And The Capital Asset Pricing Model

Why Beta Matters More Than You Think

When I first started working with portfolio managers back in the early 2000s, almost nobody could explain what beta actually was without sounding like they were reading from a textbook. They'd recite the formula—covariance of the asset with the market divided by the variance of the market—and then stare at me like the explanation should just land on its own. It doesn't. The reality is that beta measures sensitivity to market movements, nothing more, and most people confuse it with total risk. Total risk is standard deviation. Beta only captures systematic risk, the kind you can't diversify away. If you're building a well-diversified portfolio, non-systematic risk should already be nearly gone, so using standard deviation to compare assets in that context is just wrong. The CAPM equation itself is straightforward: Expected Return equals the Risk-Free Rate plus Beta times the Market Risk Premium. That's it. Rf plus Beta multiplied by (Rm minus Rf). But the assumptions behind it are where things fall apart in practice. The model assumes you can borrow and lend at the risk-free rate, which obviously isn't true for most investors. It assumes all investors have the same expectations, which is absurd. It assumes no transaction costs or taxes, which makes every real trade more expensive than the model predicts. And it assumes the market portfolio includes every single risky asset in the world, which we can never actually observe. I ran into a specific problem last year that I still think about. A client came to me with a small-cap value fund that had historically shown a beta of 0.85 against the S&P 500. The CAPM suggested it should return about 6 percent given the risk level. But the fund was consistently delivering 11 or 12 percent annually. When I dug into the numbers, the issue was that the fund's returns were driven heavily by factor exposure to size and value—things the basic CAPM with a single beta doesn't account for. Fama and French spent decades building multi-factor models specifically because of this gap. The workaround I used was running a regression against both the market factor and the SMB and HML factors from the Fama-French three-factor model. Once I isolated the alpha, it turned out the fund manager wasn't generating skill. He was just taking on extra factor risk that the basic CAPM was mislabeling as abnormal return. That distinction matters enormously when you're trying to decide whether to keep a manager or cut them loose.

Another thing people miss is that beta is backward-looking. It's calculated from historical data, usually spanning three to five years, and there's no guarantee that past sensitivity to market moves will repeat. I had a situation where a technology stock showed a beta of 1.4 over a five-year period that included the 2020 pandemic crash. During that crash, the stock dropped twice as hard as the market, which drove the high beta. But when I looked at the rolling 12-month beta, it had swung from 0.9 to 2.1 over that same period. Using the five-year average beta of 1.4 would have given a completely misleading picture of current risk. The lesson here is to check rolling betas and see how stable they actually are before committing to any valuation. The market risk premium is another area where people make embarrassing mistakes. Some textbooks suggest using 6 percent as a standard value, but that's outdated. Current implied market risk premiums based on forward earnings estimates and dividend yields sit somewhere between 3.5 and 5 percent depending on the source. Using 6 percent when the actual premium is closer to 4 percent will systematically overvalue risky assets and make your required rates of return too high. I've seen this happen in valuation reports where a 2 percent difference in the market risk premium assumption changed a discounted cash flow valuation by 30 percent on a mid-cap company. That's not a rounding error. That's a material difference. There's also the question of what risk-free rate to use. A lot of introductory courses tell you to use the yield on a 10-year Treasury bond. But if you're discounting cash flows that are 30 years out, a 10-year rate doesn't match the duration of those cash flows. The Treasury curve isn't flat, and using the wrong maturity creates a mismatch. I prefer matching the risk-free rate to the duration of the cash flows being discounted. For long-lived infrastructure projects, that means using the 20-year or 30-year Treasury yield. For short-term working capital assessments, a 3-month T-bill rate makes more sense. It's a small adjustment but it matters over long time horizons.

The CAPM also breaks down completely in certain environments. During periods of extreme market stress, betas tend to converge toward 1.0 because everything moves together. This happened during the March 2020 crash when even defensive stocks sold off in line with the broader market. If you're using a stable beta of 0.6 for a utility company during a crisis, the model will severely underestimate the actual risk. The model assumes normal market conditions, and markets aren't normal very often. I learned this the hard way when a pension fund I consulted for was using CAPM-derived discount rates throughout 2020. Their required returns dropped sharply because betas compressed, but the actual portfolio risk had increased dramatically. They were under-reserving for liabilities at the worst possible time. For practical portfolio construction, the CAPM gives you a useful framework for thinking about expected returns, but you should never treat it as a crystal ball. It's a starting point, not an endpoint. The security market line it produces is an equilibrium concept that describes what returns should look like if all the assumptions held true. They don't. Markets have anomalies, behavioral biases, and structural frictions that the model ignores. The best use of CAPM is as a benchmark—a way to ask whether an investment's expected return is reasonable given its systematic risk. If the answer is clearly no, that's where you start digging deeper into factor exposures, valuation assumptions, and the specific risks that the model is missing. One more practical tip that most courses don't mention: when you're calculating beta for a private company or a thinly traded stock, you should unlever and relever it. The raw historical beta includes the effect of the company's current capital structure, which may not be sustainable. Unlevering removes the debt effect using the formula Beta unlevered equals Beta levered divided by 1 plus (1 minus Tax Rate) times Debt to Equity. Then you relever it using the target capital structure of the company you're evaluating. This gives you a cleaner measure of business risk that isn't distorted by financing decisions. I've seen analysts skip this step and end up comparing betas from companies with completely different leverage ratios, which makes the whole exercise meaningless.

If you're studying for an exam on Chapter 6 Risk Return And The Capital Asset Pricing Model, focus on understanding the intuition behind the model rather than memorizing formulas. The exam questions will test whether you know when the model works and when it fails, not whether you can reproduce the equation from memory. The real world doesn't care about your ability to write out the CAPM formula. It cares whether you can recognize its limitations and adjust your approach accordingly.