Risk Premium

A risk premium is the additional expected return an investor requires for holding a risky asset instead of a risk-free one. Its two ingredients are a baseline, the risk-free rate, which is the return on an asset whose payment is treated as certain (in practice a short-term U.S. Treasury security), and the extra return layered on top to compensate for the chance that the risky asset disappoints. Written out, the expected return on a risky asset equals the risk-free rate plus its risk premium.

The word "expected" is doing the heavy lifting. A risk premium is a forward-looking demand, agreed to before the outcome is known. It is the reason a rational investor is willing to hold something that might fall in value at all: without the prospect of a higher average payoff, there would be no reason to accept the uncertainty. It does not mean the risky asset will in fact beat the safe one over any particular stretch. If it always did, there would be no risk to be paid for.

The general concept splits into named premia that apply to particular risks. The premium of stocks over the risk-free rate is the equity risk premium, the single most studied example and large enough that it drives most long-horizon return estimates. Within bonds, a term premium compensates for lending over a longer period, and a credit premium compensates for the chance that a corporate or lower-rated issuer fails to pay. Each is a risk premium for a specific, identifiable risk, and a single investment can carry several at once: a long-dated junk bond is paid a term premium and a credit premium together.

Two properties are worth holding onto. The first is that a risk premium compensates only for risk that cannot be diversified away. An investor who can spread a danger across many holdings until it nearly vanishes will not be paid to bear it, because someone else is willing to hold it for less. This is the idea that separates market risk, which is compensated, from the company-specific risk that diversification removes. The second is that the premium is not stable. It widens when investors are frightened and demand more to hold anything uncertain, which is the same event, seen from the other side, as prices falling. It narrows when confidence returns. Because it is an expectation about the future rather than a recorded fact, it can only be estimated, and the estimates rest on assumptions that honest analysts dispute.

Start from the return you could earn with near-certainty. Suppose a short-term Treasury yields 4%. That is the risk-free rate, the baseline against which everything riskier is measured.

Now consider a stock fund an investor expects to return 9% a year over the long run. The risk premium built into that expectation is 9% minus 4%, or 5 percentage points. The investor is, in effect, demanding five extra points a year as compensation for the fund's swings and for the real possibility of a bad decade.

The number is an expectation, not a promise, and that is the whole point. In a given year the fund might return 25% or lose 20%. Over that single year the realized premium could be strongly positive or sharply negative. The 5-point figure is the average extra return the investor believes is fair pay for the uncertainty, and it is only visible in hindsight as an average across many years. All figures are illustrative.

Why the concept is useful

• It explains, in one idea, why riskier assets tend to earn more over long periods and why safe assets earn little.
• It gives a disciplined way to build a long-run return estimate: start with the risk-free rate and add a premium for each risk the asset carries.
• It clarifies that only undiversifiable risk is rewarded, which is an argument for diversifying away the rest rather than expecting to be paid for it.

What it does not do

• It is an expectation, not a guarantee. Realized returns over any real holding period can and do fall short of the risk-free rate.
• It cannot be measured directly, so every stated figure is an estimate resting on contestable assumptions.
• It moves over time, widening in fear and narrowing in calm, so a premium estimated in one climate can mislead in another.