Abstract
The paper examines the theoretical and empirical validity of task-based segmentation of rural labour contracts in seasonal agriculture. Regular labour is hired to perform tasks that are difficult to supervise for which casual labour is not incentive-compatible, and a regular wage above the reservation wage ensures no shirking in these tasks. It is argued that there is a hoarding cost of regular labour which is the cost when productivity is low during the slack season. This implies that minimization of supervision costs necessitates the employment of regular labour in certain tasks, but enhances hoarding cost. Results from the ICRISAT villages in India suggests that daily regular wages are lower than daily casual wages, adjusted by the probability of unemployment. Estimates of a tobit selection model suggest that (a) task characteristics are not significant even among the farms hiring regular labour, (b) larger employment-intensive farms tend to hire more regular labour irrespective of the choice of crop, and (c) there is a significant substitutability between regular and family labour. These results seem to question the very basis of task-based segmentation and strengthen the hoarding cost argument: farms hiring regular labour use it indiscriminately in both non-monitorable and monitorable tasks and, if possible, they may substitute family labour for regular labour with a view to minimizing hoarding cost.