Productivity, Capital and Labor in Labor-Managed and Conventional Firms

Despite a continuing interest in the compared efficiency of labor-managed and conventional firms, only a handful of comparative empirical studies exist. These studies suggest that labor-managed firms have the same productivity levels as conventional ones, but organize production differently. However, the data used in these studies cover a single industry, or firms matched by industry and size in manufacturing, and concern a few dozen firms. In addition, the use of constant-elasticity production functions in past studies has made it difficult to distinguish the effects of incentives embodied in the factors of production from those of scale differences that could be caused by the differences in factor demand behavior between conventional and labor-managed firms hypothesized by economic theory. The paper compares the productivity of labor-managed and conventional firms using two new panel data sets covering several thousand firms from France, including representative samples of conventional firms and all worker cooperatives with 20 employees or more in manufacturing and services. We present Generalized Least Squares (GLS) and Generalized Moments Method (GMM) estimations of translog production functions industry by industry for cooperative and conventional firms and test for the equality of their total factor productivities. We also allow systematic differences in scale and technology to be determined by the ownership form. The translog specification, which allows returns to scale to vary with input levels, makes it possible to disentangle embodied incentive effects from systematic differences in scale due to underinvestment in labor-managed firms. In the process, we also propose updated "stylized facts" about labor-managed firms in comparison with conventional firms. Our production function estimates suggest that cooperatives are at least as productive as conventional firms. However, the two types of firms organize production differently. Cooperatives are more X-efficient, i.e., they use their capital and labour more effectively, than conventional firms. With their current levels of inputs, cooperatives produce at least as much with the technology they have chosen as they would if they were using conventional firms' technology. In contrast, in several industries conventional firms would produce more with their current inputs if they were organizing production as cooperatives do. In all industries and in both data sets, both types of firms would produce at constant or decreasing returns to scale if they were using the same technology at their current input levels, and we find no evidence that returns to scale are systematically higher in cooperatives. Contrary to received wisdom, descriptive statistics indicate that workers' cooperatives are not always smaller or less capitalized than conventional firms, and grow at least as fast as conventional firms in all the industries studied.