Refugee inflows, surplus farm labor, and crop marketization in rural Africa

Shunsuke Tsuda

Journal of Development Economics, Volume 155 (2022), Article number 102805


This article investigates the long-term effects of refugee inflows on host farmers in Tanzania, focusing on effects in labor and crop markets. The Kagera region in the northwest of Tanzania received large-scale inflows of refugee from Burundi and Rwanda in the early 1990s. The Kagera region is remote and impoverished, and most local households engage in subsistence agriculture. 

Refugee inflows can potentially affect host farmers through: (1) the labor market, as refugees expand labor supply; (2) the crop market, as refugee inflows increase local demand for foods that are not included in externally-sourced food assistance; and (3) market transaction costs, which may decrease (due to infrastructure development around refugee camps) or increase (due to a mix of different ethnicities in the labor market and security concerns).  

The author uses a difference-in-difference empirical method to estimate the effect of refugee flows on labor market outcomes, crop supply to markets, and agricultural labor productivity. The author estimates agricultural labor productivity (shadow wages) for family farm labor. In a perfectly efficient labor market, market and shadow wages should be equal. So, disparities between market and shadow wages indicate the degree of labor market inefficiencies. Labor market inefficiencies can arise from a proportional transaction cost (such as a commuting cost) and/or an off-farm labor market participation constraint (a restriction on how much a household member can work outside their household). The analysis of effects in the crop market focuses on two food aid crops (maize and beans) and two non-food aid crops (cooking bananas and cassava).  

The analysis draws on longitudinal household data from the Kagera Health and Development Survey (KHDS) from 1993 (prior to the refugee inflows) and 2004 (after the refugee inflows). Descriptive statistics reveal that: 

  • Most households engage in subsistence agriculture using family labor. It is rare for households to have both off-farm wage employment and hired farm labor, implying the skills of family workers and hired workers are similar. 
  • Off-farm labor market participation is low. Women have lower off-farm labor market participation than men.  
  • Off-farm market wages are substantially higher than the shadow wages of family farms. Men have higher market wages compared to women in 2004. 
  • Women are more likely to engage in cultivation compared to men. The female agricultural labor productivity is on average higher than the male one in both 1993 and 2004. 
  • Market transaction rates are low for the main food crops (maize, beans, cooking bananas, and cassava). 

Empirical results: 

  • Refugee inflows increased surplus farm labor implying an efficiency loss in the labor market. This effect is demonstrated by: (1) a weak correlation between market and shadow wages for both men and women (significant only for female wages), indicating labor market inefficiencies, which is not significantly altered by refugee inflows; (2) a widening gap between market and shadow wages for both male and female workers due to refugee inflows; and (3) a decrease in male off-farm labor market participation due to refugee inflows.  
  • Refugee inflows positively affected the transition from subsistence farming to sellers of maize and beans. This is due to a decrease in fixed market transaction costs (due to improved infrastructure) and not due to a consumption demand shift by refugees, as demonstrated by the following results: (1) crop marketization is concentrated around Rwandan refugee camps after most refugees had already left; (2) crop marketization is only observed for food aid crops for which refugee demand would be low; (3) the crop supply response around Rwandan refugee camps is detected only for initial subsistence households and not initial sellers, suggesting fixed transaction costs play a more important role than costs proportional to farm-gate prices; (4) marketization of coffee, a major export crop that would not be responsive to local demand, becomes concentrated around Rwandan refugee camps; (5) there isn’t any evidence that marketization is explained by alternative mechanisms, such as a price effect, technological change, or proximity to neighboring countries; and (6) there is an increase in supply of crops that were not included in food aid (cooking bananas and cassava) only around Burundian refugee camps where many refugees remained in 2004. These results suggest that investment in infrastructure around refugee camps also creates new opportunities for host populations and its impact lasts long after refugees have left camps. 
  • Overall, refugee inflows have a negative impact on agricultural labor productivity. The “surplus farm labor effect” and the “crop marketization effect” act in opposite directions. Refugee inflows caused losses in the labor market and gains in the crop market for host agricultural households.  

The author concludes that refugee inflows cause market-specific gains and losses for agricultural households. The results imply that, in the long run, refugee inflows increased labor market transaction costs and decreased crop market transaction costs. In both markets, fixed transaction costs play a dominant role. The author claims that the paper demonstrates that looking only at consumption or wage levels is insufficient to uncover important underlying mechanisms behind the impact of refugee inflows in rural developing areas where factor and output market imperfections are prevalent.