UC Berkeley

Economic development rarely happens in the absence of large-scale job creation. The scarcity of research on formal employment in Africa in the field of development economics is thus noteworthy. Part of the explanation is that, although steady employment represents an overarching aspiration for many Africans - often preferred, for example, over self-employment or small-scale farming - formal jobs were until recently relatively uncommon on the continent. Variation that can be exploited in statistical analysis is thus hard to come by. Another reason is that few African countries systematically record detailed employment data for large samples of workers. Researchers are therefore typically compelled to collect their own data.

Rapid urbanization and sustained economic growth - including in more labor-intensive sectors - has, however, begun to increase the availability of formal jobs in some parts of Africa, simultaneously enhancing the importance of employment research and the ability of researchers to carry out such research. Focusing on both causes and consequences of formal employment in East Africa, this dissertation examines the effect of ethnic diversity - a characteristic of many African societies - on worker productivity in the Kenyan context, as well as the impact within the household of a parent gaining employment in the Ethiopian context. Knowledge about the factors that constrain labor productivity and the consequences for households once jobs appear is necessary for effective policymaking and a goal for researchers.

I explore both issues in the context of a sector that has been particularly successful in Africa in recent decades: floriculture. A rapid expansion of the sector began in the 1980s; Kenya, for example, is now the third-largest exporter of flowers in the world and supplies approximately 31 percent of flowers imported into Europe (African Business, 2011). Neighboring Ethiopia, with its lower labor costs and abundant land, has more recently been taking market share from other African countries. Agribusiness as a whole is expected to see significant growth in Africa in the coming decades and flower farms account for a notable proportion of formal jobs in Kenya and Ethiopia - such farms are of interest to researchers in their own right. Because the workforce on flower farms often resembles a microcosm of the labor force as a whole, they also represent a meaningful case study from which broader lessons can be learned.

Two types of data are used in this dissertation: surveys of flower farm workers and applicants (ethnicity, time use, etc) and the output records of a flower farm in Kenya. The farm recorded individual and team output for pay purposes.

The first chapter of this dissertation explores the influence of ethnic diversity on labor productivity in a team production setting. Ethnic diversity has long been known to constrain economic development, but the direct effect on output remains largely unexplored. In Kenya, the land- and water-abundant areas where flower farms are located have experienced in-migration from other parts of the country, yielding ethnic diversity in the farms' workforces. I study teams of "packing plant" workers at a large flower farm. Working in teams of three, the workers pack flowers and prepare them for shipping.

I show that ethnically diverse teams are less productive than homogeneous teams. Although an inability to socially sanction non-coethnics may also play a role (see Miguel and Gugerty, 2004 and Habyarimana, Humphreys, Posner and Weinstein, 2007), the primary reason appears to be preference-driven: workers upstream in the triangular production chain lower total output and their own pay by skewing their supply of intermediate flowers toward coethnic downstream workers.

I then go on to analyze the firm's response and the change in the magnitude of the ethnic diversity effect during a period of increased ethnic conflict in Kenya, illuminating how the response of output to diversity is likely to vary across time and space. I find that the productivity loss from ethnic diversity in teams varies with the political environment (see also Posner, 2004). It appears that, in high-cost environments firms are forced to adopt second-best policies to limit discrimination distortions.

Overall chapter 1 shows that inter-ethnic rivalries lower allocative efficiency and productivity in Kenyan floriculture, and highlights the likely consequences for firm behavior and employment growth in the private sector in Africa. The implications for policy and future research are potentially wide-ranging. Most African countries are ethnically diverse and cross-ethnic joint production will increase as urbanization brings together larger groups of workers in cities. Modernization of the economy typically entails greater specialization which also increases the scope for distortions due to ethnic discrimination in production chains.

In the second part of my dissertation, which consists of two separate articles, I focus on the consequences (rather than the causes) of employment. I analyze the effects within the household of a parent gaining employment in rural Ethiopia. Taking advantage of a unique situation in the labor market for farm-workers in Ethiopia at the time, I worked with five flower farms that agreed to randomize fall 2008 hiring due to significant excess demand for jobs and a perceived inability to screen applicants.

In chapter 2, I analyze the impact on children's lives, focusing primarily on time use. Mother's employment has been argued to especially benefit children, but there is little existing evidence to back up such claims. I therefore analyze the effect of mother's and father's employment separately.

The results show that mother's and father's employment affects sons and daughters very differently. Daughters spend significantly less time in school when mothers work because they are expected to take over house-work tasks. Daughters' time use is unaffected by father's employment, while sons spend significantly more time in school when either parent works. It appears that both the reconfiguration of a parent's time use implied by employment and the associated increase in income affect children's time use. Daughters' human capital accumulation suffers from the greater time requirements of "female" house-work in Ethiopia.

In chapter 3, I analyze the impact of female employment on domestic violence, which is believed to respond to large shifts in spouses' relative incomes in poor countries. Contrary to the predictions of standard economic models of the household, I find a significant increase in domestic violence when women get employed. The reason appears to be that men in rural Ethiopia attempt to restore their dominance in the household through violence when their relative economic standing is weakened.

In combination chapters 2 and 3 give a rather bleak picture of the influence of female employment on the position of women and girls in poor countries. It is important to recognize that this dissertation focuses on the effects of employment in the short-term, however. In the longer term gender norms may respond to employment, in which case the longer term impact could differ from the deleterious effects observed here. Rather than suggesting that female employment should not be encouraged, the evidence presented thus highlights that theory and employment policy should take traditional gender roles seriously.

In combination, the three chapters of this dissertation highlight that features of society that particularly characterize Africa - such as ethnic diversity in the workforce and time-consuming house-work - interact in first-order order ways with the causes and consequences of employment. We must thus study Africa directly rather than rely on evidence from rich countries when shaping policy.

Beyond seeking to address the substantive issues raised, it is my hope that this dissertation illustrates how direct, micro-level output data can be used to advance research on the determinants of productivity in poor countries, and how a labor market situation often found in developing countries with small formal sectors allows randomized evaluations of an otherwise hard-to-analyze "treatment" - employment itself.