Under team pay, biased upstream workers are unable to increase the relative pay of favored downstream workers by distorting relative supply. As a result, horizontal misallocation of flowers was eliminated. Total output in teams in which the two processors were of different ethnic groups therefore increased, the introduction of team pay returning the difference in output between such teams and homogeneous teams to pre-conflict levels. Overall output also increased, even though the results indicate that team pay led processors to freeride on each others’ effort. This paper’s results indicate that, if taste for discrimination is high enough, firms are forced to adopt “second best” policies to limit the distortions caused by such discrimination. But entirely removing workers’ incentives for discrimination is difficult. At the plant, team pay had little effect on the degree of discrimination in teams that were ethnically differentiated vertically rather than horizontally, as also predicted by the model. The obvious “solution” to discrimination – segregating workers – may be undesirable for reasons unrelated to productivity in the short term. The extent and multiplier effects of taste-based misallocation also depend on a number of other factors, such as pay systems, the structure of production, and the “geographical” distribution of ethnic groups in the productive system, however. More speculatively, cannabis grow equipment it is possible that such factors respond endogenously to ethnic diversity. Social segregation is commonly observed in diverse societies but likely becomes harder to achieve as urbanization brings larger groups of workers together. The linkages and specialization required in industrialized production are rarely observed in the most ethnically diverse countries.
My findings also suggest that the economic costs of ethnic diversity vary with the political environment. Relatively brief episodes of ethnic conflict can have a long-lasting impact on economically distortionary attitudes: I find no decay in discrimination in the nine months after conflict ended. Multiple equilibria may thus exist if the occurence of conflict itself depends on attitudes towards non-coethnics, some diverse societies being characterized by tolerance and little conflict and others by ethnic biases and frequent conflict.The quotes above illustrate a prevailing view among policymakers which sees the creation of job opportunities for parents – especially mothers – as a quintessential tool for improving the lives of children in poor countries. The view appears to be based in large part on extrapolation of findings from studies of the effects in the household of increases in unearned income. Relying on such extrapolation may be adequate if the dominant household models – in which children typically appear only as an expenditure category for the decision-making parents 1 – provide an accurate picture of a poor country household. If instead there is substitution between parents’ and childrens’ time use, then employment may be a fundamentally different “treatment” than pure income transfers due to its implications for the employed parent’s time use. In that case the lack of causal evidence on the consequences for children of parent’s employment is a problematic gap in the literature on poor countries. Taking advantage of a field experiment that randomized long-term job offers this paper presents direct evidence on the impact of a parent’s employment on children’s lives. Five Ethiopian flower farms agreed to allocate fall 2008 job offers through a lottery system.
The experiment was “natural” in the sense that parents sought employment in the exact same way they would have done in the absence of the research team. Because households thus themselves determined if the mother or the father applied, I analyze the two sub-samples separately. The farms were willing to randomize job offers because open positions attracted large numbers of mostly inexperienced applicants and screening was difficult. Before the lottery took place, enumerators surveyed acceptable applicants. Winners and losers were re-surveyed five to seven months after employment commenced. The randomization was effectively stratified on gender. The main results are as follows. As daughters take over house-work left undone when a mother gets employed, their school-time falls by 24 percent per week. Daughters’ time use is unaffected by father’s employment. An increase in sons’ school time of about ten percent when a mother or a father gets employed appears to be due to higher household income; sons’ house-work time is unaffected by parents’ employment. After documenting the impact of parents’ employment on childrens’ time use, I present a simple collective framework in which each parent attaches weight to daughters’ well-being and daughters derive utility from going to school, but only females can do house-work . The framework highlights the variables upon which heterogeneity in the response to mother’s employment is likely to depend if the primary underlying force is time use substitution between mothers and daughters.
Testing the framework’s predictions, I find that, the higher the proportion of daughters – a variable that is shown to be exogenous in the sample studied – the less negative the impact of mother’s employment on a given daughter’s school-time, the greater the weight attached to daughters’ well-being, the less negative the impact of mother’s employment on a daughter’s school-time, and the greater the initial bargaining power of the mother, the greater the reduction in daughters’ school-time when mothers get employed. Daughters themselves appear to have little influence over the change in their time use when mothers get employed. Interestingly, selection into mother’s versus father’s employment appears to depend on the same covariates, providing further evidence of the importance of female house-work substitution. These results have important implications for the design of employment programs and for how selection into parent’s employment and its effects in the household should be modeled. If full-time school enrollment is not universal, explicitly accounting for children’s time use is important. In situations where the house-work necessary to run a household is time consuming, the substitutability between parents’ and children’s effort introduces a potential trade-off between parents’ and children’s preferences. If house-work is effectively gender specific, then the conventional wisdom – that economically empowering mothers is of greater benefit to daughters than empowering fathers – is not necessarily the full story when it comes to parent’s employment, even if mothers weigh daughters’ well-being more than fathers do. The reason is that mothers may face a trade-off between own and daughters’ time use that fathers do not. If female participation in the market economy over time influences the norms governing the division of labor in the household, then the longer-term effects of mother’s employment may differ from those observed here, but such norms are likely slow to change. This paper builds on and extends the overlapping literatures on adult employment, vertical grow rack child labor and schooling, and intra-household decision-making in poor countries. Causal evidence on the effects in the household of long-term parental employment in poor countries is to my knowledge largely absent, credible exogenous variation in employment rarely being available. Indirect inference – for example on the basis of findings from studies of unearned income – has been attempted, but there are good reasons to study parent’s employment directly. Beyond the implied time use reconfiguration, employment may for example affect the two parents’ relative bargaining power differently than government transfers or income from other sources do. This paper presents the first experimental evidence on the effects in the household of a parent’s long-term employment. Children’s time use is one of the primary determinants of human capital accumulation and child well-being. The degree of substitutability between parents’ and children’s time use is therefore important. Several existing studies find correlations between a mother’s employment status and children’s time use in poor countries . Doran convincingly shows that adults in Mexico work more when children work less due to an exogenous increase in time spent in school.
But his focus is on paid child labor; though understudied in the literature due to a lack of data child house-work is much more common than paid work in most of the developing world, and the effect of parents’ time use on children’s time use is typically of greater relevance for policy than the converse. Gender specificity of house-work in combination with the typically greater time requirements of “female” responsibilities may be a particularly important though often overlooked form of son favoritism, especially because child labor and schooling are negatively related . I take advantage of an exogenous increase in mother’s and father’s work hours to provide causal evidence on time use substitution between mothers, fathers, daughters and sons. Bar and Basu argue that an inverted-U relationship can arise because of missing labor markets for children: the results in this paper suggest that missing labor markets for adults can also lead to a range in which child labor may appear to be increasing in parents’ income. As formal employment opportunities arise for mothers, daughters may be forced to take over house-work. The preferences of children and parents are not perfectly aligned, even if parents are partially altruistic. An important question is how much influence children have over their own lives: the review in Edmonds argues that our almost complete lack of knowledge about parent-child agency and who makes child time use decisions is the most pressing issue in the literature on child labor. This paper’s results indicate that the reconfiguration of a daughter’s time that occurs when a mother gets employed in rural Ethiopia is decided by parents, primarily mothers, while daughters themselves have little influence over the change in their time use. The paper is organized as follows. In section 2, I present the setting and the experiment. The reduced form time use estimates are in section 3. In section 4, I present a simple theoretical framework of household work and schooling decisions that illustrates the forces that underlie the results in section 3, and derive auxiliary predictions. The predictions are tested in sections 5. Section 6 provides further evidence on how time use decisions are made and section 7 analyzes selection into employment.Growth in the commercial floriculture sector in Ethiopia has been explosive in recent years, fueled in part by government incentives and in part by the abundant availability of cheap land and labor in rural areas. In 2008, 81 flower farms employed around 50,000 unskilled workers. Most flower farm workers work in greenhouses, growing and harvesting flowers, or in “pack houses”, packaging flowers and preparing them for shipping. Over 70 percent of flower farm workers are women . Hiring on Ethiopian flower farms typically takes place in October and November, before the main growing and harvesting season. The supervisors on five flower farms agreed to randomize job offers during fall 2008 because of an unusual situation in the labor market for flower farm workers at the time. Because comparable jobs were seldom available in the areas suitable for flower growing, applicants almost always outnumbered the positions to be filled by large margins. Ethiopian flower farms – still getting to grips with cost components significantly larger than labor and with little ability to predict the productivity of the mostly uneducated, illiterate and inexperienced applicants – did not prioritize optimization of the unskilled workforce . Because supervisors were already allocating job offers relatively arbitrarily when approached by the research team, explicit randomization was a modest procedural change. When Ethiopian flower farms hire, word is typically spread in nearby villages. Job-seekers arrive at the farm on announced “hiring days”. At the participating farms, supervisors first excluded any unacceptable applicants. A team of enumerators then carried out the baseline survey with the remaining applicants. Finally, the names of the number of female and male workers to be hired were drawn randomly from a hat. The full sample thus consists of 527 households in which at least one spouse applied to a flower farm job and was deemed acceptable for hiring. There are 346 women in the sample and 188 men: in almost all cases one of two spouses applied. We attempted to re-interview everyone in the treatment and control groups 5 – 7 months after employment commenced. Because few farms were hiring workers in the season that followed the randomization, only 6 re-interviewed individuals in the control group had managed to obtain employment. Careful tracking procedures led to a re-interview rate of 88 percent and no statistically significant differential attrition. Almost all the job-seekers are parents: the focus here is on the effects of a parent’s employment for children in the household.