CFA Level 1 · Portfolio Management

A negative beta asset moves against the market. When the index rises the asset falls, and when the index falls the asset rises. That much is easy to state and it is where most explanations stop.
The part worth understanding is what a negative beta does to expected return. Under the capital asset pricing model, an asset with a negative beta has an expected return below the risk free rate, and for a deeply negative beta it can be negative outright. An investor who buys it is knowingly accepting a poor expected return, and is rational to do so. Once that is clear, everything else about negative beta follows.
Beta is often described as a measure of volatility. That description is wrong, and correcting it is the first step to understanding a negative one.
where Ri is the return on the asset, Rm is the return on the market, Cov is the covariance between them, and Var is the variance of the market return.
Read the numerator. Beta is built on covariance, which is a measure of co-movement, not of size. An asset’s own volatility appears nowhere on its own. A wildly volatile stock whose swings are unrelated to the market has a beta near zero, and a placid stock that reliably tracks the index has a beta near one. Beta measures the share of an asset’s movement that the market explains, and the direction of that relationship.
The sign therefore comes entirely from the covariance, because variance in the denominator can never be negative. A negative beta means one thing precisely: the asset and the market move in opposite directions more often than not.
There is a second reading of the same formula that is worth carrying. Beta is also the slope of a regression of the asset’s returns on the market’s returns. That is why it answers the question of how much the asset moves for a given market move, and why a beta of two implies roughly twice the response rather than twice the risk. Written that way, a negative beta is simply a downward sloping regression line, which is an ordinary thing for a line to do and stops the concept feeling exotic.
Treating beta as a volatility measure. Beta is a measure of systematic risk, the part of an asset’s risk that cannot be diversified away because it is shared with the market. Total volatility includes idiosyncratic risk as well, which diversification removes and which beta ignores entirely. An asset can be highly volatile and have a beta of zero.
The formula is easier to trust once it has produced a negative number from real figures.
A market index and a hedging strategy report the following returns over six periods, in percent. Market: 4, −3, 2, −5, 6, −1. Strategy: −2, 2, −1, 4, −3, 1. Compute the strategy’s beta.
Answer: beta is −0.62. When the market rises by 1%, this strategy is expected to fall by about 0.62%, and when the market falls by 1% it is expected to rise by about 0.62%.
Notice step 3, because it is where the sign is decided. Every paired product came out negative, which is what a consistent opposing relationship looks like in the arithmetic. A single period moving the other way would have reduced the magnitude of the beta, and enough of them would have flipped its sign.
Put that beta into the capital asset pricing model and something surprising happens.
where Rf is the risk free rate and E(Rm) − Rf is the market risk premium, which is positive.
The market risk premium is positive. Multiplying a positive premium by a negative beta gives a negative number, which is then added to the risk free rate. The expected return therefore falls below the risk free rate.
Take the strategy above, with a beta of −0.62. The risk free rate is 4% and the expected market return is 10%. What return should be expected?
Answer: an expected return of 0.28%, well below the 4% available risk free. A beta below about −0.67 would push the expected return negative outright.
The line does not stop at zero on the left. It continues into negative beta territory, and everything below the risk free rate on that chart is a position an investor holds in full knowledge that lending the money risk free would be expected to pay more.
An investor who accepts an expected return below the risk free rate is not making a mistake. They are buying insurance. The asset is expected to pay off precisely when the rest of the portfolio is losing money, and that timing is worth paying for. The negative expected excess return is the premium on the policy.
That also explains why reliable negative beta assets are scarce. If something dependably rose when markets fell, demand for it would push its price up until its expected return dropped to the point where holding it hurt in normal times. The cost is the mechanism, not a flaw in it.
Assets divide fairly cleanly into those whose negative beta is structural and those where it is an observation that may not repeat.
| Position | Beta is negative because | How dependable |
|---|---|---|
| Long put options on an index | The payoff is contractually defined to rise as the index falls | Structural, holds by construction |
| Short index futures or forwards | The position is the market with the sign reversed | Structural, beta is close to minus one |
| Inverse exchange traded funds | Built to deliver the opposite of a daily index move | Structural daily, but compounding causes drift over longer holding periods |
| Dedicated short bias funds | The strategy is net short by mandate | Usually negative, varies with the manager’s positioning |
| Long dated government bonds | Rates often fall in a growth shock, lifting bond prices | Conditional, and it reverses in an inflation shock |
| Gold | Demand rises in some stress episodes | Episodic, and a beta near zero is the more usual reading |
Long dated government bonds deserve a note of their own, because the row above compresses a conditional relationship into one line. Bonds have protected portfolios in growth shocks, where central banks cut rates and bond prices rise while equities fall. In an inflation shock the two fall together, since rising rates hurt bonds and equities at the same time. And in a broad liquidity crisis both are sold regardless of merit. A bond beta estimated across a decade dominated by one of those regimes says very little about the next.
The first three rows are different in kind from the last three. A short futures position has negative beta because of what the contract is, and no amount of market history can change that. Gold and government bonds have shown negative beta in particular episodes, which is a statement about a sample rather than about a mechanism.
Gold is routinely offered as the standard example of a negative beta asset. The claim deserves more care than it usually gets.
Measured over long periods against a broad equity index, gold’s beta typically sits near zero rather than clearly below it. That is a genuinely useful property, because an asset uncorrelated with the market still improves a portfolio, but it is a different property from negative beta and it carries a different expected return. An asset with a beta of zero should be expected to earn the risk free rate, not less than it.
What is fair to say is that gold has behaved defensively in particular episodes, especially those combining falling real interest rates with financial stress. What is not fair to say is that it reliably moves opposite to equities. In some sell-offs it has fallen alongside them, for the reason most assets do in a liquidity crisis: holders sell what can be sold rather than what they would prefer to sell.
Before calling any asset negative beta, name the period the estimate came from and the index it was measured against. A negative beta measured through one crisis is a description of that crisis. A short futures position needs no such qualification, which is exactly what separates the two categories.
Negative beta is not low beta. A beta of 0.2 is a muted version of the market, still moving with it. A beta of −0.2 moves against it. Under the capital asset pricing model the first earns more than the risk free rate and the second earns less, so the sign changes the conclusion rather than the degree.
Negative beta is not negative correlation alone. Beta combines the correlation with the ratio of the two volatilities, so two assets with identical correlation to the market can have very different betas. Take two assets both correlated −0.5 with the market. If the first is half as volatile as the index its beta is −0.25, and if the second is twice as volatile its beta is −1.0, a fourfold difference in response from the same correlation. Correlation tells you the direction and the consistency of the relationship. Beta tells you the magnitude of the response as well, which is why it and not correlation is what enters the capital asset pricing model.
Negative beta is not a negative expected return in every case. The expected return falls below the risk free rate, but whether it goes below zero depends on how negative the beta is and how large the market risk premium is. In the worked example a beta of −0.62 still gave a small positive expected return of 0.28%. The threshold is the point where the beta multiplied by the premium exceeds the risk free rate, which with a 4% rate and a 6% premium arrives at a beta of about −0.67. Below that, the model says an investor should expect to lose money on the position and hold it anyway.
Beta is estimated from a window of past returns, which makes every negative beta a statement about a period rather than a permanent property.
Three things move it. The length of the window matters, because a beta measured over one year of a particular regime can differ sharply from one measured over five. The choice of index matters, because a beta against a broad domestic index and a beta against a global one are different numbers for the same asset. And the return frequency matters, since daily, weekly and monthly data give different estimates for the same asset over the same period.
Estimation error compounds the problem. A beta is a point estimate with a confidence interval around it, and for an asset whose true beta is close to zero that interval routinely straddles zero. A reported beta of −0.1 from a short window is often statistically indistinguishable from no relationship at all, which means the negative sign carries far less information than it appears to.
The deeper problem is that relationships change under stress. Correlations that hold in calm markets often converge in a crisis, which is when a hedge is most needed. An asset selected for its historical negative beta can find that beta drifting toward zero, or turning positive, in exactly the episode it was bought to protect against.
The most common question gives a beta and asks for the expected return, testing whether a negative beta is handled correctly in the capital asset pricing model rather than treated as an error. The second most common asks why a rational investor would hold an asset whose expected return is below the risk free rate, where the answer is the hedging payoff and its timing. A third form asks you to distinguish beta from volatility, where the answer is that beta measures only the systematic part.
It means the asset tends to move in the opposite direction to the market. A beta of minus one implies that a 1% rise in the market is associated with a 1% fall in the asset, and the reverse. The sign comes from the covariance between the asset and the market, since the variance in the denominator of the beta formula can never be negative.
Yes, and under the capital asset pricing model a negative beta asset must. The market risk premium is positive, so multiplying it by a negative beta produces a negative number that is added to the risk free rate. Investors accept this because the asset is expected to pay off when the rest of the portfolio is losing money, and that timing is worth paying for.
Not reliably. Measured over long periods against a broad equity index, gold’s beta typically sits near zero rather than clearly below it, which is a valuable property but a different one. Gold has behaved defensively in particular episodes, especially where falling real rates coincided with financial stress, but in some sell-offs it has fallen alongside equities because holders sell what they can rather than what they would prefer to.
Volatility measures the total size of an asset’s price movements. Beta measures only the part of that movement explained by the market, which is the systematic component that diversification cannot remove. A highly volatile asset whose swings are unrelated to the market has a beta near zero, so the two measures can point in completely different directions.
Short index futures or forwards, long put options on an index, and inverse exchange traded funds. In each case the payoff is contractually defined to move opposite to the index, so the negative beta follows from the structure of the instrument rather than from a historical estimate. Inverse funds carry a caveat, because daily rebalancing causes their longer horizon returns to drift from the simple inverse.
Not necessarily. Beta is estimated over a chosen window against a chosen index at a chosen return frequency, and changing any of those changes the number. Relationships also tend to converge under stress, so an asset selected for its historical negative beta may find that beta moving toward zero in precisely the crisis it was bought to protect against.
Loading comments...
Add your Thoughts: