The elasticity of substitution σ measures how easily a firm can substitute between inputs when their price ratio changes. It quantifies the percentage change in input ratio relative to percentage change in price ratio. σ = 0 means inputs are complements (no substitution), σ = 1 for Cobb-Douglas, σ = ∞ for perfect substitutes. Higher elasticity means firms have more flexibility in factor adjustment.
From your study of isoquants, you know that the slope of an isoquant at any point — the marginal rate of technical substitution (MRTS) — tells you how many units of one input a firm can trade for one unit of another while keeping output constant. A steeply curved isoquant means the MRTS changes quickly as you move along it: inputs are hard to substitute. A gently curved isoquant (nearly straight) means the MRTS stays roughly constant: inputs are nearly perfect substitutes. The elasticity of substitution σ turns that geometric intuition into a number.
Formally, σ = % change in (K/L) ÷ % change in (w/r), where w is the wage and r is the rental rate of capital. It asks: if the relative price of labor rises by 1%, how much do firms shift their input mix away from labor and toward capital? A high σ means firms respond aggressively — they substitute readily. A low σ means firms are stuck using roughly the same mix no matter what prices do. The formula connects isoquant curvature to observed factor demand behavior: flatter isoquants → less curvature → higher σ → more substitution.
The three benchmark cases illuminate the range. When σ = 0, the isoquants are right-angled (Leontief): inputs must be used in fixed proportions like two tires per bicycle axle. No amount of price change induces substitution because the inputs are perfect complements. When σ = ∞, the isoquants are straight lines: inputs are perfect substitutes and the firm uses entirely whichever is cheaper, switching completely if the price ratio crosses a threshold. The Cobb-Douglas case at σ = 1 sits in between: factor shares of income (wL/pY and rK/pY) remain constant as input prices change, a prediction that has been historically useful for modeling aggregate production.
Why does σ matter economically? When σ is high, labor and capital markets are tightly linked: a wage increase can be substantially offset by substituting toward capital (automation). When σ is low, wage increases translate more directly into higher costs, because the firm has little flexibility to rebalance. For policy analysis, σ governs how income is distributed between workers and capital owners as factor prices change: with Cobb-Douglas (σ = 1), factor shares are fixed; with σ < 1, the relatively more expensive factor sees its share *rise* (because quantity can't adjust much); with σ > 1, the more expensive factor sees its share *fall* as firms successfully substitute away from it.