Adoption of modern agricultural inputs in Africa remains low, restraining agricultural productivity and poverty reduction. Low quality agricultural inputs may in part explain low adoption rates, but only if farmers are aware that some inputs are low quality. We report the results of laboratory tests of the quality of glyphosate herbicide in Uganda and investigate whether farmers' beliefs about the prevalence of counterfeiting and adulteration are consistent with the prevalence of low quality in their local market. We find that the average bottle in our sample is missing 15 per cent of the active ingredient and 31 per cent of samples contain less than 75 per cent of the ingredient advertised. Farmers believe 41 per cent of herbicide is counterfeit or adulterated. Beliefs are significantly correlated with quality at the local market level, but beliefs remain inaccurate, adjusting for only a fraction of actual differences in quality. Price is also significantly correlated with quality in local markets, but prices also adjust for only a fraction of quality differences. Although, like fertiliser and hybrid maize seed, herbicide in Uganda is low quality, herbicide use is substantially higher.

To identify which spouses respond to asymmetric information, I play two dictator games with each member of the couple. One decision is kept secret from the respondent's spouse, while another choice is public. Most people give the same amount in public and secret, while a minority behave opportunistically and give more to their spouse in public and less in secret. The types identified in the lab also behave differently at home. For those who behave opportunistically in the lab, greater knowledge about finances at home is associated with worse opportunism in the lab, suggesting that for these couples complete information at home is not due to trust but rather is an endogenous response to non-cooperative behavior. The paper highlights that allowing for heterogeneous types changes the conclusions we draw about appropriate models of the household and suggests that laboratory games can be used to identify household types useful in the interpretation of field data.

No abstract

Decision-making structures may be different across polygynous and monogamous households, leading to different economic outcomes and requiring different targeting of anti-poverty programmes. We study efficiency in semi-nomadic pastoralist households in Northern Senegal with lab-in-the-field games. We find that monogamous and polygynous families are equally productively inefficient overall. However, average contributions at the household level mask differences across dyads. Junior wives receive less but give more to their husbands than senior wives, leaving junior wives worse off than other household members.

No abstract