Nonparametric estimation of returns to scale: Method and application

The paper describes a method of estimating variable returns to scale in production that adaptively fits spline functions, using model selection criteria, to determine the appropriate number and location of break points for a fixed factor of production. Unlike other approaches, the method obtains nonparametric estimates of variable returns to scale for small samples while ensuring global curvature and flexibility properties are maintained. An application of the method is presented using data from the British Columbia sablefish fishery.