Measuring postharvest losses at the farm level in Malawi
Measuring postharvest losses at the farm level in Malawi
Reducing food loss and waste are important policy objectives prominently featured in the United Nation’s Sustainable Development Goals. To optimally design interventions targeted at reducing losses, it is important to know where losses are concentrated between the farm and fork. This paper measures farmlevel postharvest losses for three main crops—maize, soy, and groundnuts—among 1,200 households in Malawi. Farmers answered a detailed questionnaire designed to learn about losses during harvest and transport, processing, and storage and which measures both total losses and reductions in crop quality. The findings indicate that fewer than half of households report suffering losses conditional on growing each crop. In addition, conditional on losses occurring, the loss averages between 5 and 12 percent of the farmer’s total harvest. Compared to nationally representative data that measure losses using a single survey question, this study documents a far greater percentage of farmers experiencing losses, though the unconditional proportion lost is similar. We find that losses are concentrated in harvest and processing activities for groundnuts and maize; for soy, they are highest during processing. Existing interventions have primarily targeted storage activities; however, these results suggest that targeting other activities may be worthwhile.
The survey modules used in this study are adapted from farmer surveys being used as part of the larger effort to examine postharvest losses among producers and processors (see, for example, Schuster and Torero 2016). The survey specifically asks farmers to self-report whether losses occurred during three particular activities between harvest and sale into the value chain: harvest and transport from the field to home, processing, and postprocessing storage. Moreover, the instrument carefully elicits both quantity and quality losses to better understand how losses are occurring.
This study measures postharvest losses and their determinants among smallholder farms in farmer groups in Malawi for three crops: maize, soy, and groundnuts. The postharvest loss survey modules were added to the second follow-up survey for a cluster randomized controlled trial studying the impact of individualized extension services and capital transfers. Doing so allows us to relate losses during the follow-up to household and farm characteristics collected over three rounds of surveys. The detail available in the survey modules helps pinpoint when postharvest losses occur. Moreover, the existence of recently collected, nationally representative data allows us to compare the farmers in our sample with a representative sample of farmers both nationwide and in the districts in which the intervention took place. Finally, because the nationally representative survey includes a single question on postharvest losses, we can study the differences between asking a single question and the more systematic body of questions included in our survey.
Our results contribute to the existing literature in several ways.
First, across crops, fewer than half of households report losses (conditional on growing that crop). Conditional on experiencing a loss, losses vary across crops: 5 percent for maize, 8 percent for soy, and 12 percent for groundnuts. Although the maize estimates are small relative to FAO (2011) estimates, they are comparable to other recent studies examining staple crops in Africa south of the Sahara (Kaminski and Christiaensen 2014; Minten, Engida, and Tamru 2016; Abdoulaye et al. 2016). Losses in our sample are concentrated in harvest and processing activities; for soy, they are highest during processing. Most interventions have targeted storage activities; however, our results suggest that targeting other activities might also be effective. Furthermore, the findings indicate that large losses tend to be concentrated on a few farmers.2
Second, we compare our results to those from the most recent available data of the nationally representative Integrated Household Survey (IHS), which relies on a single question to measure postharvest losses. Although we find considerably higher incidence of postharvest losses using our detailed approach, we find fairly similar unconditional proportions of lost production estimates.
Third, we move beyond measuring the incidence of losses to consider the determinants of losses across activities and crops. We find broadly consistent patterns across the three different crops examined and across the different activities, as well as for quantity and quality losses separately. Wealthier households, smaller households, and households with a female farmer group member are better equipped to avoid crop loss. Further, individuals who are less risk averse are less likely to experience losses. This last finding is robust to using either a self-reported scale measure or an incentivized experimental measure of risk aversion. Fourth, after examining the impact of our individualized extension support program (relative to group-based extension), we find suggestive evidence that two years of exposure reduces postharvest losses.
Finally, we measure postharvest losses for maize, soybean, and groundnuts at the farm level. Although the literature shows other estimates of maize losses, there are only two estimates of groundnut losses from Ghana (Affognon et al. 2015), and there appear to be no estimates of soy losses anywhere in Africa south of the Sahara. Therefore, this paper both serves as a complement to Kaminski and Christiaensen (2014) in maize and provides new estimates for other crops underrepresented in the literature.