Risk is not volatility: telling the two apart

Two ideas, one word, a lot of damage

Ask most people what investment risk is and they describe volatility. The value goes up and down a lot, so it is risky. Ask a lender, an insurer or a financial stability economist and you get something different. Risk is the probability and severity of a bad outcome, weighted by whether you can recover from it.

These are not the same thing, and the confusion runs in both directions. It makes people avoid things that wobble but are survivable. It also makes them accept things that are quiet but can destroy them, because nothing in the recent price history looked alarming.

A useful working definition for a household is this. Volatility is how much the price moves. Risk is the chance that you end up unable to do the thing the money was for. Notice that the second definition contains you in it. Volatility is a property of an asset. Risk is a property of the relationship between an asset, a person, and a deadline. The same investment can be low risk for one person and high risk for another on the same day.

What volatility actually measures

Volatility is a statistic describing the dispersion of returns over some past window, usually expressed as an annualised standard deviation.

Standard deviation in plain terms

Take an investment's returns over many periods. Find the average. Then measure how far the typical period sits from that average. That typical distance, scaled to a year, is the volatility number. Higher means the returns are more spread out around their average. Lower means they cluster.

It is a measure of scatter. That is all it is. It does not know whether the scatter came from good surprises or bad ones, and in its standard form it treats a large gain and a large loss as equally volatile, which is not how anyone actually experiences them.

What the number assumes

Three assumptions are baked in, and each can fail.

• It assumes the past window is informative about the future. A quiet three years produces a low number, whether or not the quiet period was representative.
• It assumes prices exist and are meaningful. An asset that is priced rarely, or valued by a model rather than by transactions, will show low measured volatility because the measurement is sparse, not because the underlying value is stable.
• It assumes the distribution of outcomes is reasonably well behaved. Real markets produce occasional moves far larger than a normal distribution predicts, and those rare moves are precisely the ones that matter for whether you survive.

That third point is why volatility is a reasonable description of ordinary weather and a poor description of storms.

What risk means when you are the one holding it

Try this test instead. For any potential loss, ask two questions.

1. How likely is it, roughly.
2. If it happens, can I recover, and how long would recovery take.

Call it the recoverability test. It replaces a single statistic with a judgement, which is less precise and considerably more honest.

A portfolio that falls thirty per cent and recovers over four years is a serious event for someone who needs the money in two years and an ordinary event for someone who needs it in twenty. Nothing about the asset changed between those two people. The deadline changed.

This is why risk cannot be printed on a product. A fund can publish its volatility, because volatility is a property of the fund. It cannot publish your risk, because your risk depends on your time horizon, your income stability, your other commitments, and whether you would be forced to sell during a fall. Securities regulators require risk factors to be disclosed in offering material precisely because these are facts about the investment that a reader must combine with facts about themselves.Source^source

Four risks volatility cannot see

Permanent impairment

Volatility measures round trips. It has nothing to say about a fall that never comes back.

A diversified index falling thirty per cent and a single company falling thirty per cent may register similarly as volatility. They are different events. The index is a shifting group of businesses and has a mechanism for recovery as new companies enter and weak ones leave. The single company may have lost a licence, a patent, a founder, or a market, and there is no rule that says its price must return.

Permanent impairment is the difference between a drawdown and a loss. Volatility cannot distinguish them, because both look like a large negative move.

Forced selling and liquidity

The most dangerous property of an investment is often not how much it moves, but whether you can convert it to cash when you need to, and at what discount.

Market functioning is analysed by central banking institutions in terms of liquidity and funding conditions, which are separate from the observed variability of prices.Source^source The practical version for a household is simple. If you must sell during a fall, a temporary decline becomes a permanent loss. Whether you must sell depends on your cash buffer, your income, your debt, and whether the asset can even be sold quickly.

An illiquid asset with a stable quoted value can be far riskier than a liquid one that moves daily, because the stable value is not a price at which anyone is currently obliged to buy.

Sequence of returns

Two portfolios can produce the same average return over ten years and leave you in completely different positions, purely because of the order in which the returns arrived. This matters enormously if you are adding or withdrawing money along the way.

Suppose you have 500,000 and withdraw 40,000 a year. In scenario A the first three years return minus fifteen per cent, minus ten per cent, then a long stretch of positive years. In scenario B the positive years come first and the two bad years arrive at the end. The average annual return across the decade is identical. The balance at the end is not, and the gap can be large, because the early withdrawals in scenario A are taken from a shrunken pot and those units never participate in the recovery.

Sequence risk is invisible to volatility. The two scenarios have the same dispersion of returns. They do not have the same consequences.

Correlation that appears under stress

Diversification is measured in calm conditions and tested in bad ones. Assets that behave independently most of the time can move together at exactly the moment you were relying on them not to, usually because the same investors are selling all of them to raise cash.

Official financial stability analysis treats vulnerabilities such as leverage and liquidity mismatch as distinct from measured market volatility, and notes that periods of low measured volatility can coexist with building vulnerability.Source^source That is a formal way of saying that calm is not safety, and that the correlations you measured last year are not a promise.

Why the industry uses volatility anyway

It is worth being fair to the statistic, because it is not a scam and it is not useless.

• It is computable. You can produce it from a price series with no judgement, which means it can be compared across thousands of products consistently.
• It is comparable. Two funds with volatilities of six and eighteen are genuinely telling you something different about their typical experience.
• It is the input to a large body of theory that supports portfolio construction, option pricing and capital requirements, all of which need a number rather than an opinion.
• It correlates loosely with risk in ordinary conditions. Most of the time, things that move more do carry more chance of loss.

The problem is not that volatility is used. It is that it gets renamed. When a document labels a volatility figure as a risk rating, the reader reasonably assumes the number reflects the chance of a bad outcome, and it does not. It reflects the chance of a bumpy ride, which is a different question and one that mostly matters through its effect on behaviour.

A worked example, identical volatility, opposite consequences

Suppose two portfolios each show an annualised volatility of twelve per cent over the past five years.

Portfolio A is a broad, diversified holding across many companies and several economies, held by someone with a stable income, a twelve-month cash buffer, no debt against the portfolio, and no need to touch the money for fifteen years.

Portfolio B is concentrated in three companies in one sector, held by someone with irregular income, no cash buffer, a loan secured against the holding, and a plan to use the money for a property deposit in eighteen months.

The statistic is the same. The risk is not remotely the same.

Run a thirty per cent fall through both. Portfolio A's owner does nothing, keeps contributing, and the event becomes a story about a difficult year. Portfolio B's owner faces three separate problems at once. The secured loan may require additional collateral or trigger a forced sale at the worst price. The absent cash buffer means any household emergency during the fall forces further selling. And the eighteen-month deadline means there may be no recovery period at all, so a temporary fall is crystallised into a permanent shortfall on the deposit.

Everything that made Portfolio B dangerous was invisible to the volatility number. Concentration, leverage, buffer, deadline. Three of the four are facts about the person, not the asset.

The most important risk controls available to an ordinary household are not asset choices at all. They are the cash buffer that prevents forced selling, the absence of borrowing against volatile assets, and an honest deadline for the money. Get those right and a great deal of volatility becomes survivable. Get them wrong and even modest volatility can end a plan.

Low volatility that is genuinely dangerous

Because the confusion runs both ways, it is worth naming the pattern where quiet things carry real risk.

• An asset valued monthly by a model rather than daily by a market will report low volatility almost by construction. The smoothness is a measurement artefact.
• A holding that produces a steady income stream can look stable right up to the point the payer stops paying. Steady until it stops is a common risk shape and it has almost no volatility until the end.
• An arrangement promising fixed returns regardless of conditions has zero measured volatility and concentrates every risk into the single question of whether the promiser can pay. Anything offering guaranteed high returns with no variability deserves more scepticism than a volatile investment, not less.
• Cash held for decades has very low volatility and a well understood long-run risk, which is the erosion of purchasing power. It moves slowly enough that it never registers as scary.

Each of these can be a perfectly reasonable thing to hold. The point is only that the low number did not tell you they were safe.

What to look at instead

You do not need statistics to make progress here. Five questions get most of the way.

1. What would have to be true for this to lose most of its value permanently, and how plausible is that.
2. If it fell by half tomorrow, would anything force me to sell. Debt, a deadline, a margin call, a household emergency with no buffer.
3. When do I actually need this money, and what happens if it is worth twenty per cent less on that date.
4. What am I depending on that I have not examined. A single employer, a single sector, a single currency, a single counterparty holding the asset.
5. How would I find out if something went wrong. Do I receive real reporting, and can I sell if I want to.

Question two is the one people skip and it is the one that most often converts a bad year into a disaster. Question five is the one that separates regulated products from arrangements that only look like them.

Where this distinction breaks down

An honest treatment should include the counterargument, because the risk-is-not-volatility line can be taken too far.

Volatility is not irrelevant. It has two real effects on outcomes. It affects behaviour, and behaviour is where most household investment damage originates. A person who sells during a fall converts volatility into permanent loss, which means for that person volatility was risk. It also matters mechanically when you are withdrawing money, as the sequence example showed, because variability plus withdrawals interact.

There is also a practical problem with rejecting the statistic. Volatility is measurable and risk is not, so a demand to measure risk instead often produces nothing measured at all, which is worse. The honest position is that volatility is a partial, backward-looking proxy that is useful when you know what it omits.

So use it as one input. Read the number, then ask what it cannot see. Concentration, leverage, liquidity, deadlines, counterparties, and whether you would be forced to act at the worst moment. Those are the variables that decide whether a bad year is an inconvenience or the end of a plan, and none of them appear anywhere in a standard deviation.

Source^source: Bank for International Settlements.

Source^source: Global Financial Stability Report, International Monetary Fund.

Source^source: UAE Securities and Commodities Authority.