Chris Blatmann and Stefan Dercon have published a great NY Times piece about their new field experiment set in Ethiopia. Throughout the developing world, urbanization is taking place. People who were farmers are moving to cities to take industrial jobs. In a world of uncertainty and risk, some people might regret making these moves away from the "sure thing" and social network of the countryside. Yes, cities pay more and offer more excitement but there are urban risks. Blatmann and Dercon run a field experiment in which five factories use a lottery to decide which job applicants to hire. This lottery system means that they have a "treatment group" (those who receive an urban job) and a "control group" (those who do not receive this specific job). This is a "funky control group" because they are highly likely to find some other urban job and this means that the treatment is hard to define but let's ignore this. Among the treatment group, many of those who were randomized in to receive the urban factory job left the firm after just a couple of months. This surprised the researchers.
There are two different labor economics theories for why new workers at a new firm might not stay long. First is the old job matching theory. A worker doesn't know his productivity at a specific firm/job and this can only be learned by trying it. A bad match quickly is terminated.
A second theory relates to Sherwin Rosen's famous work on compensating differentials and non-wage attributes of the job. Companies such as Google have great workplace amenities to keep skilled workers with firm specific human capital. Such non-wage benefits (think of health insurance) are not taxed and this provides an even greater incentive for firms to provide these. But, in Ethiopia, the authors argue that these jobs are risky and nasty with bad work conditions. In this case, if the wage is low and the job attributes and hours are horrible, people have an incentive to quit and go back to their agricultural life. This is the theory of compensating differentials. Think about the words "combat pay". You have to pay a wage premium to keep workers at a firm or industry (think of coal mining) that is nasty.
University of Chicago Price Theory predicts that urban firms in LDC nations with nasty work conditions will have to pay more to keep talent. Or, they will have an incentive to improve the work conditions at their sweat shops. So labor market competition will induce the capitalist to provide a nicer, air conditioned, safer work location. In such a case, capitalist competition is necessary and sufficient for improving worker quality of life. Note that this argument hinges on labor demand being high. If firms know that there is an "infinite" amount of rural labor who can be hired at a low wage then the "sweatshops" will continue. The key here (from a poor person's point of view) is for the supply of labor to be upward sloping and for these workers to have some firm specific training that will be lost if these workers leave the firm.