A vast literature has developed on the theory and measurement of surplus labour in overpopulated developing countries, of which India is often taken to be the paradigm. (See Kao et al., 1964 and Sen, 1966 for reviews.) The subject continues to be one of practical importance, particularly for the estimation of shadow wage rates for project analysis. (See Lal, 1973, 1974, for a review of this literature.) It was in the context of estimating shadow wage rates for industrial public sector projects in India that the research reported in this paper was undertaken. For this purpose the important question is: what wilI be the effects on agricultural output when an agricultural worker is withdrawn either directly or indirectly through the process of rural-urban migration induced by increased industrial employment? (See Harberger, 197 1; Harris and Todaro, 1970, for models of the rural-urban migration process.) In the early literature on surplus labour it was commonly assumed that, in overpopulated developing countries, labour in agriculture was in surplus in the sense that its marginal product was zero, so that agricultural output would not fall if some agricultural labour were withdrawn. Tests for the existence of surplus labour were based in part on estimates of the marginal product of agricultural labour and partly on measurements of the surplus labour time available in agriculture. The latter measures also yielded estimates of the extent of surplus labour and disguised unemployment, as follows. First, the total requirements of labour in agriculture (in a given region) with given technology were estimated. Next, the total availability of labour was estimated by multiplying the number of workers available by some postulated number of ‘normal’ working hours in a day (or by some estimated ‘normal work-load). The number of labour days required were then subtracted from those available to provide an estimate of the number of surplus labour days (and hence labourers) available in the area. Mehra (1966) provides the most notable attempt to measure surplus labour in India on these lines. These ‘tests’ and estimates of surplus labour are implicitly based on a ‘stock’ definition of the surplus of labour time per worker which is available. They fail, however, to take account of the economic notion of a supply price of labour which is given by the possibilities of substitution of income and leisure (open to any utilitymaximizing labourer), at levels of labour input per worker below that postulated in the ‘norm’ of the full ‘work-load’ per worker. That is, these estimates implicitly assume that the disutility of work does not alter until the normal number of working hours postulated in making estimates of the full work-load have been worked.’ For, as the recent theoretical literature on surplus labour has shown (Sen, 1966; Berry and Soligo, 1968; Stiglitz, 1969; Zarembaka, 1972), the necessary and sufficient conditions for surplus labour are given by a constant disutility of effort, which implies a constant marginal rate of substitution between income and leisure, over the relevant range of hours worked per man in the traditional sector. ’ Hence zero marginal productivity (tests for which were one route toward testing the hypothesis of surplus labour) is neither a necessary nor a sufficient condition for the existence of surplus labour. Nor is the evidence
Read full abstract