82 Afiriciilliiral Research and Prodiictivily 



falls from a plausible .24 in regression 1 to practically zero in 

 regressions 2 and 4. This may in part be due to what amounts to a 

 distinction between short-run labor productivity and long-run 

 labor productivity. That is, the estimates reported in regressions 2 

 and 4 are based on ''within country'' variance, which in these 

 data is based on relatively few years. The estimates based on 

 regressions 1 and 3 allow "between country" variance to identify 

 the coefficient. These can be viewed as long-run production 

 elasticities, in that the countries being compared are at different 

 historical stages and the coefficients thus reflect long-run adjust- 

 ment. 



The problems of labor productivity in agriculture have been in- 

 tensively debated in the economic literature (see, for example, Fei 

 and Ranis 1964; Lewis 1954; Schultz 1964). Boserup described 

 the technological requirements of "movement along the produc- 

 tion function'" by changes in labor force as a reaction to changing 

 population density in primitive agriculture. Consider, for exam- 

 ple, the shift from forest fallow to plough cultivation (Boserup 

 1965, p. 35): "root crops [grown under forest fallow] are unsuit- 

 able under the system of short fallow [ploughing] because under 

 that system the plants will need to be protected against grassy 

 weeds by hard weeding. When long fallow is replaced by short 

 fallow, food consumption usually becomes concentrated on 

 cereals." 



Boserup's example is only one case of the kinds of technologi- 

 cal changes which are needed to absorb the increased labor force 

 and which require changes in production that also entail changes 

 in consumption and habits. To adapt and adopt new methods, to 

 change farming habits, and to modify institutions, even in mod- 

 ern agriculture, requires costly and time-consuming adjustment 

 process. In the short run, labor absorption — given existing culti- 

 vation systems, land tenure arrangements, and institutional struc- 

 tures—can result in relatively low marginal labor productivity. 



The short-run marginal productivity of labor can be expected 

 to be different for labor increases and for labor decreases. An 

 abrupt reduction in labor supply, as in the Indian influenza 

 epidemic (Schultz 1964) will leave many farms unattended, create 

 pockets of extreme labor shortage, and sharply reduce production. 



