How do search frictions affect firm hiring decisions? We conduct a randomized control trial among 799 private firms with an active job vacancy in Addis Ababa, Ethiopia. A random subset of these firms are provided subsidized access to a new type of employment agency, which provides additional applicants with college diplomas or degrees. In our first main finding, we show that treated firms are 17.5% more likely to fill the vacancy within one month, but the effect is not driven by hiring workers provided by the agency. Instead, having had more interactions with college educated applicants, treated firms become less optimistic about the average productivity of college graduates. Among those firms requesting a college graduate at baseline, treated firms are significantly less likely to hire a college graduate and more likely to hire a non-college educated worker. There are no significant treatment effects on worker turnover, performance, or effort for the worker hired for that vacancy. These findings demonstrate that search frictions can distort firm hiring behavior by affecting learning and belief formation about the labor market, a potentially important but understudied barrier to firm growth in low- and middle-income countries. 

Many developing countries are undergoing a rapid process of industrialization, yet high worker turnover rates constitute a barrier for manufacturing firms to sustain their operation. This paper studies how misperceptions about career ladder can affect turnover rates in manufacturing sector in developing countries. We conducted a field experiment in one of the main industrial parks in Ethiopia, where we document significant misperceptions about the salary trajectory and the likelihood of being promoted to higher positions. We then conduct an information treatment, where we provide accurate information on career ladder estimated using records from the industrial park, and examine how misperceptions about career ladder causally affects workers' turnover decisions. We find that optimistic updates about upper-level salary significantly increase the probability of remaining employed within the industrial park, while pessimistic updates reduce it. We find no evidence of spillover effects to control workers, suggesting informal network may not be able to fully address the information frictions. For workers with higher educational attainment and previous garment experience, providing accurate information does not drive away over-optimistic workers and instead, it decreases overall turnover rates, suggesting skilled workers may be incentivized to exert more effort and climb up the career ladder when their over-optimistic perceptions are corrected. Our findings call for firms to provide more transparency of the career ladder to address information frictions and retain skilled workers.

We propose a new perspective on the challenges of building a state. We show in the context of a weak state that asserting the state can give existing armed actors   incentives to plunder. We combine data from 239 villages of eastern Congo with quasi-experimental variation induced by the completion of a  large military campaign which asserted the state's exclusive right to tax in specific villages. After the campaign was complete, and during three years, members of the  armed factions previously taxing those villages regularly attacked them. The rise is driven by violent theft operations targeting  wealthy households, and is muted for retaliatory attacks,  conquest operations, as well as for attacks by other perpetrators. The targeting is consistent with a destruction of  their incentive to refrain from violent theft. Asserting the state's exclusive right to tax increased household material welfare, by decreasing overall tax burdens, but it increased sexual violence and abductions as a byproduct of the theft operations. An alternative state building strategy based on bargaining with armed actors rather than attempting to diminish their ability to tax does not create this incentive. Our findings suggest that asserting the state by removing armed actors who have established themselves, tax, and protect, can induce a temporary trade-off between growth and safety and challenge conceptions of state power based only on monopoly of violence.

Why does violent collective action succeed? We conduct a series of surveys reconstructing the history of  violent collective action covering twenty years (1994-2014) of the eastern Democratic Republic of the Congo's civil conflict. The surveys include one of the largest violent social movements of the region, which formed at the time of our survey. For a sample of 6,809 individuals, we construct a comprehensive history of participation in violent organizations and of their communities. We document several facts about participation in violent collective action: (i) militia---armed organizations that emerge from rural communities with community-oriented missions,  predominate all other violent organizations; (ii) while militias offer little material rewards, militias' combatants joined them with the purported motivation to protect (or exact revenge for) their communities against threats of violence by actors perceived to be foreign; (iii) using quasi-experimental variation induced by a military policy which drastically reduced the state's ability to protect entire communities, we show that violent threats against the community lead to the formation of violent social movement chapters, fueled by spikes in participation from people who purportedly joined  voluntarily, and by people who reported to be socially coerced, sparked by community leader efforts to mobilize participation; (iv) attacks by armed groups perceived to be foreign against an individual's community are associated with a subsequent increase in the probability that the individual joins a militia chapter purportedly motivated by the desire for revenge, to protect the community; this effect is particularly strong if their kin was also attacked, in which case they join with the purported motive to protect the family;  (v) using shocks to the demand for gold, a valuable mineral that cannot easily be taxed by armed groups, we document that one such attack in the kin induces motivations to join a militia whose opportunity cost equals 8 times the yearly p.c. income. Our findings suggest that, alongside selfish strategic motivations, parochial altruism and community coercion are crucial in explaining the origins, and success, of violent collective action.