Unsystematic Risk

The mechanics rest on the algebra of portfolio variance. For an equally weighted portfolio of n assets, total variance equals the average variance of the individual assets divided by n, plus the average covariance between pairs multiplied by (n minus one) divided by n. As n grows, the first term collapses toward zero while the second converges to the average covariance. That limit is the systematic floor; everything eliminated on the way down was unsystematic.

The intuition is that idiosyncratic shocks arrive with different signs at different times. A product recall at one issuer coincides with a patent win at another, and in a portfolio of many positions the surprises partly offset. Nothing offsets a shift in real rates, growth expectations or global risk appetite, because those move all holdings in the same direction at once.

Empirically, the split is estimated by regressing an asset's returns on a market proxy. The slope coefficient is beta; the intercept is alpha; the unexplained residual captures firm-specific news. Total variance of the asset then equals beta squared multiplied by the variance of the market, plus the variance of the residual. The first component is systematic; the second is unsystematic, and it is what disappears under diversification.

The regression R-squared expresses the systematic share of total variance directly. A large integrated bank with an R-squared near 0.7 derives most of its return variability from market-wide factors; a small biotechnology firm awaiting a trial readout may show an R-squared in the low single digits, meaning almost all of its volatility is idiosyncratic. Note that high total volatility does not imply high beta: the two can diverge sharply.

Concentration is the usual channel. Executives and employees paid in restricted stock hold portfolios dominated by a single issuer's fate, compounded by human capital tied to the same firm. Founder wealth, family holding companies and single-country pension mandates carry the same structural flaw. Sector funds and thematic ETFs reduce single-name exposure but retain industry-specific risk, which behaves as unsystematic risk relative to a broad global benchmark.

On the institutional side, active equity managers deliberately accept idiosyncratic risk: tracking error against a benchmark is largely residual risk taken in pursuit of alpha. In fixed income, default of an individual issuer is the classic diversifiable event, which is why credit portfolios hold many small positions rather than a few large ones. Correlations rising toward one in a systemic crisis is precisely the moment when that protection weakens.

Three capabilities matter. First, classify a given event: an accounting scandal, a fire at a plant or a CEO's resignation is firm-specific, whereas a central bank policy surprise, an oil shock or a recession is systematic. Second, work the algebra: decompose total variance into beta-driven and residual components, or show that portfolio variance converges to average covariance as holdings increase. Third, connect the split to asset pricing.

That last link is the substantive point. Because a diversified investor can eliminate idiosyncratic variance at negligible cost, the market offers no compensation for bearing it. Equilibrium models price only non-diversifiable exposure, which is why beta rather than standard deviation appears in the expected-return relation. An undiversified investor bears the residual risk fully but is still paid only for the systematic part.