Beta And What It Does Not Tell You

Beta is among the most quoted risk statistics and among the most frequently misread. It answers one specific question and is silent on most others.

The measure is relative movement

Beta describes how much a holding has historically moved for a given movement in a chosen market index, estimated from past data.

A figure above one indicates the holding has amplified the market's movements, and below one that it has moved less. Both directions are included.

It is therefore a statement about the relationship with the index rather than about the holding's absolute variation.

It depends entirely on the index chosen

Beta is measured against a specific benchmark, and the same holding can show quite different figures against different indices.

A technology company measured against a broad market and against a sector index will produce different numbers, neither of which is wrong.

A quoted figure is therefore incomplete without the benchmark it was calculated against, which is often omitted.

Specific risk is excluded by construction

Beta captures only the portion of movement explained by the index. Everything specific to the holding falls outside it.

A company with a low beta can still fail entirely for reasons unrelated to market conditions, and the statistic would not have indicated it.

This is why beta is a measure of market sensitivity rather than of risk in any complete sense. Two holdings sharing a beta can face entirely different specific circumstances.

The proportion of movement the index explains varies enormously between holdings, and where that proportion is small the beta figure describes very little of what actually happens.

The estimate is unstable

Beta is calculated from a historical window, and the figure changes as that window moves or as its length changes.

Companies also change over time through acquisitions, disposals and shifts in financing, so a historical relationship may no longer describe the current business.

Treating the number as a fixed property of a holding rather than as an estimate from a particular period is the most common error in its use.

Relationships break down when it matters

Beta describes an average relationship over the measured period. During severe stress, holdings frequently move together more than their historical figures imply.

A portfolio constructed to have low sensitivity to the market can therefore exhibit higher sensitivity precisely during the periods it was designed to soften.

The statistic remains useful for describing typical behaviour, provided it is not relied upon as a description of behaviour in unusual conditions.