Risk-Free for Whom?

I was in a discussion recently where U.S. Treasuries were described as “the” risk-free asset, without qualification. It is a fair shorthand and, for many purposes, a perfectly workable convention. But for a long-horizon asset owner, it is worth pausing to consider what that description assumes.

The real question is not whether U.S. Treasuries are high-quality assets. They are. The real question is whether they are actually risk-free for the institution making the decision.

That depends on what the capital is for.

For an asset owner, “risk-free” cannot be defined only by the absence of default risk. It also has to be understood relative to liabilities, liquidity needs, the payment currency, and the horizon over which obligations must be met. An asset can be free of credit risk and still be the wrong proxy for safety if it introduces the wrong currency exposure, the wrong duration exposure, or the wrong relationship to the institution’s actual solvency framework.

That matters more than people sometimes admit.

If liabilities are paid in U.S. dollars, then U.S. Treasuries may be a sensible reference asset. But if liabilities are paid in another currency, then the answer is no longer so straightforward. From the perspective of a fund with non-USD obligations, a U.S. Treasury position entails foreign-exchange risk. That does not make it a bad asset. It just means it is not automatically “risk-free” in the institutional sense people often assume.

The same issue shows up in real terms. If the institution’s obligations are linked, formally or informally, to inflation, then nominal bonds may be much less safe than they appear in standard portfolio discussions. And for institutions with long and uncertain horizons, reinvestment risk matters too. A short-duration asset can look safe on a mark-to-market basis while still leaving the fund exposed to the need to reinvest under very different future conditions.

So the meaning of safety depends on the problem being solved.

This is not simply a semantic point. It sits very close to the centre of asset allocation.

Once a risk-free asset is chosen, risk premia begin to take shape around it. Expected returns on risky assets are measured, explicitly or implicitly, relative to that reference point. And asset allocation, at bottom, is really the problem of deciding how to maximise exposure to attractive risk premia within a given risk target.

That is why the definition of risk-free matters so much.

If the reference asset is poorly aligned with the institution’s liabilities, the resulting measures of excess return, hedging value and portfolio risk may provide an incomplete picture. Assets that appear attractive against a market convention may look materially different once currency, duration, inflation sensitivity and solvency risk are considered. What looks attractive in one framework may look much less attractive once liabilities, currency, and solvency are factored back in.

This is especially important when thinking about liquidity.

Liquidity is often talked about as though it were a universal property of an asset. In practice, it is more useful to think about liquidity relative to the obligations the fund may need to meet under stress. A liquidity buffer is not just a pool of assets that can be sold quickly. It is a reserve of assets that can be relied on in the relevant currency and time frame without undermining the broader purpose of the portfolio.

That, in turn, brings governance into the room.

The choice of a risk-free rate influences much more than discounting. It shapes how institutions think about hurdle rates, hedge ratios, rebalancing triggers, liquidity reserves, and even what counts as prudent risk-taking. If the wrong reference asset gets embedded too deeply into the investment process, the institution can end up optimising around a market convention rather than around its actual mandate.

That is why I think the risk-free rate is not only a technical input.

It is also a governance choice.

For long-horizon asset owners, the starting point should not be habit. It should be the mandate.

What liabilities have to be paid? In what currency? Over what horizon? Under what solvency framework? With what governance capacity? Only after those questions are answered can the institution define what safety actually means.

That may still lead to U.S. Treasuries.

But it should not automatically lead there.

For a long-horizon asset owner, “risk-free” isn't simply a market convention. It is a mandate decision.