Heterogeneity in Returns to Scale: A Random Coefficient Analysis with Unbalanced Panel Data

Abstract:\nThis paper analyses the importance of scale economies by means of unbalanced plant-level panel data from three Norwegian manufacturing industries. Focus is on heterogeneous technologies, and unlike most previous work on micro data, the model description includes heterogeneity in both the scale properties (the slope coefficients) and the intercept term, represented by random coefficients in the production function. Three (nested) functional forms are investigated: the Translog, an extended Cobb-Douglas, and the strict Cobb-Douglas. Although constant or weakly increasing returns to scale is found for the average plant, the results reveal considerable variation across plants. Variations in both input and scale elasticities are to a larger extent due to randomness of the production function parameters than to systematic differences in the input mix.