Unionization affects cost of production through compensation premia, technology shifts, and deviations from the least-cost combination of inputs. The first two are familiar, but the last is not. This article distinguishes the three effects, illustrates the factor-use effect, and suggests that it may resolve several apparent inconsistencies: union-induced cost effects appear larger than those implied by union compensation and productivity differentials; union compensation and productivity differentials suggest a larger effect on labor intensity of output than is observed; and employers complain that union work rules reduce productivity when there is little evidence that this is so.

The traditional model of collective bargaining confines unions to settlements constrained by the employer's labor demand curve, but an alternative model places wage-employment outcomes on a contract curve that extends beyond the labor demand curve. This paper derives a multidimensional (hedonic) contract-curve model in which employment-security provisions are used to maintain efficient bargains outside the employer's demand curve and distinguishes empirically between the contract-curve and demand-constraint models using data for public school teachers in New York State. Estimates clearly support the contract-curve model over the demand-constraint model by linking the gap between compensation and the value of the marginal product to the strength of employment-security provisions.