Gender Bias in Bargaining: Lessons from Haggling over the Price of Seed in Rural Uganda
Bjorn Van Campenhout and Leocardia Nabwire
In developing countries, haggling over prices of goods and services is common practice. A range of personal, economic, social and cultural factors influence strategies used during the bargaining process—and ultimately its outcome. For example, bargaining between a government official who pulls up in an expensive SUV and a poor woman selling tomatoes in a roadside stall is likely to be very different from that of two traders of equal standing cutting a deal involving a few tons of maize.
In a recent study conducted in rural Uganda, we built on prior research that hinted at biased perceptions against female-managed agro-input shops, examining the intricate dynamics that govern bargaining over a bag of maize seed of an improved variety. Employing a lab-in-the-field experiment, we focused on unraveling the complexities of negotiations between sellers and buyers, and in particular on how the gender of the seller shapes the bargaining process and outcomes—finding that women face a distinct negotiating disadvantage.
A bilateral bargaining experiment through a gender lens
In the experiment, a representative group of (both male and female) maize farmers was offered the opportunity to buy a bag of hybrid maize seed. A trained enumerator, guided by a script implemented on a tablet computer, acted as a seller. After explaining the virtues of hybrid maize seed to the buyer, the seller asked if the buyer wanted to buy the seed at an initial offer price. If the buyer rejected the offer, he or she was encouraged to make a first counter bid.
The seller’s actions were guided by an algorithm on the tablet, which determined whether to accept the counter bid (if the difference between the two bids was small enough) or if not, to begin a second round of negotiations, with a second offer price lower than the first but higher than the farmer’s initial counter bid. This process continued until the farmer accepted an offer price, or the seller was instructed to accept when the difference between last bid and new offer fell below a set threshold.
To estimate the causal impact of the seller’s gender on the bargaining process and outcomes, we randomly assigned farmers to either a male of a female seller. We found that a discernible gender effect emerged as negotiations unfolded. Buyers facing female sellers were less likely to accept initial offer prices. Counter bids were significantly lower in the group of farmers paired with female sellers than among those negotiating with a man. Negotiations also extended over more rounds.
Most importantly, the final price that emerged after bargaining was almost 9% lower for female vs. male sellers. The probability density plot below shows a clear peak in the distribution of outcomes around 5,000 Ugandan shillings for transactions involving a female seller. If the seller is a man, meanwhile, the probabilities that farmers buy at 10,000 or 5,000 are almost the same. These findings suggest that biases manifest early in the negotiation process, shaping subsequent interactions.
Significance
If gender proves to be an important determinant of bargaining dynamics and outcomes more generally, our study has important implications for efforts to improve women’s empowerment. Indeed, in many patriarchal societies characterized by strong gender norms and customs, operating small informal businesses is one of the few ways in which women can earn money independently from their husbands. In Uganda, we found that a surprisingly large share of women run agro-input shops. Such “outside options” in the from of assets or income that women control have been shown to be key to women’s empowerment and household well-being—thus, addressing this bias is likely to have a range of positive effects.
Our study is limited in the sense that it merely shows the existence of gender bias in bargaining in a particular context and for a particular commodity; it does not provide reasons why this bias exists (e.g., prejudice, stereotyping, or statistical discrimination). More research is needed on the roots of this bias, such that appropriate policies and strategies can be designed to level the bargaining playing field.
This work was carried out under the CGIAR GENDER Impact Platform, Market Intelligence, and SeedEqual Initiatives, which is grateful for the support of CGIAR Trust Fund contributors: www.cgiar.org/funders
The impact of COVID-19 on maize value chains: evidence from Uganda
Leocardia Nabwire and Bjorn Van Campenhout
As the COVID-19 pandemic broke out, governments around the world have been scrambling to contain the virus through various measures. In developing countries, vaccination proceeds slow due to a variety of reasons, and so governments have no other option than to resort to measures to reduce the spread of the virus through restricting movements and limit people’s ability to congregate in places of worship, pubs, restaurants, etc.
While governments are trying to safeguard food security as much as possible, the restrictions will affect food supply chains in many ways, often especially indirectly through labour shortages or reduced demand.
Almost 2 years after the outbreak of the pandemic, we now already have a lot of studies that look at the pandemic on all aspects of life, including global food security.
However, many of the food supply chains that have been studied involve higher value commodities in complex and often global supply chains. Their structure, which is generally characterized by vertical integration, multiple interactions and contracting between actors, may make the chain more or less resilient to shocks than more local informal chains that are characterized by spot transactions and relational contracts that are hard to enforce.
One such value chains is the maize value chain in eastern Uganda. Smallholder maize famers grow predominantly for own consumption, but also often sell part of their harvest to small traders. Maize is consumed in the form of maize flour, and so maize mills are also important value chain actors (Figure 1).
We look at the impact of the pandemic by building on baseline data that was collected in 2019 at three levels of the maize value chain—farmers, traders, and millers—who were re-interviewed twice by phone since the start of the pandemic.
Findings are summarized in the three radar graphs below. We focus on five key outcomes: business closures, decreased in quantities being transacted (or the scale of operations), decrease in revenues, decrease in household income, and decrease in food security. We do this for the three actors (maize producers, maize traders, and maize processors) separately. We also compare outcomes in round 1 and round 2 to differentiate between sort run effects and long run consequences of the pandemic.
Figure 2 shows that among maize producers, business closures were limited and did not change over time. However, we do see that many farmers report that they reduced the scale of operations. Results also seem to suggest that amounts sold within the supply chain recovered somewhat over time. However, the terms on which these amounts were sold seemed to deteriorate over time as more farmers report a decrease in revenue in round 2 than in round 1. We also see that close to 80 percent of the farmers report that overall household income decreased. This suggests that farmers reduce risk by relying on a variety of income sources, some of which may be relatively more affected than maize value chains.
Figure 2: Impact of COVID-19 and associated measures on maize farmers
Figure 3 reports on the impact of COVID-19 and associated measures among maize traders. Also here, closures are limited, but we do see that they are increasing over time. More than 60 percent report a decrease in the scale of operations immediately after the outbreak of the pandemic, but quantities traded seem to recover over time. However, despite the recovery, a large share of traders indicate that revenue keeps decreasing. Household income of traders also suffers, similar to that of farm households. Food security increases over time.
Figure 3: Impact of COVID-19 and associated measures on maize traders
Figure 4 turns to maize millers. Closures are again limited, but quantities processed reduced dramatically in round 1. In round 2, there is also a recovery, in line with what we see among traders. Income of millers seem to be somewhat less affected than that of other actors.
Figure 4: Impact of COVID-19 and associated measures on maize millers
The above shows that limited closures of businesses does not mean that COVID-19 and associated measures did not affect value chain actors. Furthermore, it shows that different actors are also differently affected: for actors upstream, the impact seems to be less but sustained over time, while for actors downstream, initial impacts were substantial, but the situation improved markedly over time. That said, while volumes recovered, this is not always reflected in matched changes in revenue and this is also reflected in household wellbeing and food security.
Does bottom-up monitoring improve public services? What we found in Uganda
Nassul Kabunga, Caroline Miehe, Tewodaj Mogues and Bjorn Van Campenhout
In many developing countries, poor delivery of public services remains an important problem. Public infrastructure, such as roads or boreholes for drinking water, is poor. The quality of service provided in hospitals or schools is low. Absenteeism and corruption are endemic.
Uganda is a case in point: its public service sector suffers from high levels of elite capture, ineffective monitoring and weak accountability. In response to this, the government of Uganda, under the stewardship of the Office of the Prime Minister, initiated community based monitoring and accountability meetings — popularly known as barazas — with the general objective of
enhancing public involvement in holding the government accountable for service delivery in relation to the resources spent.
In these town-hall style meetings, citizens receive information and are allowed to call officials out on not delivering on their promises. Baraza events often become very emotional and attract considerable media attention.
Barazas were first piloted in 2009. Since then, efforts have been under way to roll them out in all sub-counties in the country. The programme became a high profile policy intervention which received broad support within government and among citizens alike.
But a formal impact evaluation was still outstanding. Therefore, in 2015, with the support of the International Initiative for Impact Evaluation, which funds rigorous impact evaluations in low- and middle-income countries, we designed a large study aimed at evaluating the effectiveness of the baraza programme. The study involved over 12,000 households in 250 sub-counties spread throughout the country.
Overall, we found that barazas increased the quality and quantity of some public services, though not in all sectors, and some services were affected more than others.
Barazas organised at the sub-county level had a clear impact on service delivery in the agricultural sector. They encouraged farmers to organise themselves in associations and cooperatives, allowing them better access to government provided agricultural inputs and advisory services.
In other sectors, we found significant impacts of sub-county level barazas on access to drinking water and schooling infrastructure. More specifically, barazas reduced the time that one has to wait at the water source by 11 minutes. The intervention also increased the percentage of public schools with electricity by 50% and the percentage of schools with a water source by 15%.
Three key questions
Our study took the form of a cluster randomised control trial, a method often used in the bio-medical sciences that enables the researcher to attribute changes in a pre-defined set of outcomes to an intervention. In our study, we assigned sub-counties to host barazas at random, leaving the rest as the control group. Because all sub-counties are similar in terms of public service delivery before the baraza, any difference in service delivery after the study must be the result of the baraza intervention.
We set out to answer three questions. We firstly simply wanted to assess the impact of the baraza programme as it was originally designed.
Second, we zoom in on the two main components of barazas. On the one hand, barazas are platforms that are supposed to increase transparency by empowering citizens with information. On the other hand, barazas encourage citizens, politicians and civil servants to actively engage with each other in the spirit of deliberative decision making.
Third, in 2012, a decision was taken that subsequent barazas would target larger administrative areas (districts) instead of the smaller sub-counties. The aim was to increase participation at a higher jurisdictional levels and reduce costs. So we checked if these larger barazas were equally effective.
What we found
Our findings suggest that barazas reduced in-kind contributions of citizens to the common good – such as voluntary labour – during construction of roads. But at the same time there was increased cash contributions for the construction of boreholes for drinking water, schools and health infrastructure.
The greater willingness to contribute cash for the common good suggests that barazas have the potential to increase trust in officials and strengthen the social contract.
There were also signs that barazas may work differently in particular situations or settings. For example, we found that barazas were particularly effective for households in remote areas. In addition we found that the effectiveness of barazas seemed to dissipate over time.
We also found that barazas may be an important catalyst to organise communities around public infrastructure in the absence of social cohesion due to ethnic fractionalisation.
Overall, our analysis suggests that impact from barazas may be highly localised and context specific. A baraza in one sub-county could evolve in a very different direction to one in another sub-county. They could be effective at solving the specific problems that came up during the meetings, but when outcomes were averaged to estimate overall impact, this could be diluted.
Finally, barazas are fairly cheap to organise and affect large groups of people. Even small and uncertain effects may significantly outweigh the cost of organising a baraza every once in a while.
Tewodaj Mogues and Nassul Kabunga contributed to the research on which this article is based. This article first appeared on The Conversation.
Providing information to empower women in agriculture: Evidence from Uganda
Els Lecoutere, David Spielman, Bjorn Van Campenhout
Targeting women directly with relevant information in ways that are appealing increases their agency, access to resources, and achievements in farming
In many developing countries, women often have limited ability to make important strategic choices, let alone transform those choices into desired action and outcomes. Research suggests that targeting women with relevant information in formats that are both accessible and appealing can change this reality. However, all information campaigns are not equally successful. Often, seemingly small design attributes can have substantial impacts on effectiveness. Interest is growing in understanding how the content, format, sources, and targeting of information affect empowerment. For example, the influence of role models on women’s decision-making and empowerment is potentially important in information campaigns (Porter and Serra 2020, Riley 2017).
Empowering women in agriculture
In the smallholder agriculture systems found in many developing countries, women work long days in the field but rarely have a voice in major decisions. These can include what crops to grow, which technologies and inputs to use, and whether to consume or sell crops. One constraint to women’s influence is asymmetric information between spouses regarding the existence and application of productivity-enhancing technologies and practices (Magnan et al. 2015). When women possess less information, they are less able to participate in decision-making processes on the farm (Fisher and Carr 2015). Targeting women with information that reduces such asymmetries may therefore be empowering.
The source of information may also contribute to women’s empowerment in agriculture. Prior studies capture this by studying the singular and combined influence of social learning, peer effects, social identification, and gender homophily (a tendency of people to align with others of the same gender and background) (e.g. BenYishay and Mobarak 2019, Beaman and Dillon 2018). Targeting women with information conveyed by women of similar characteristics may be a further channel for empowerment.
Yet, very few of these ideas find their way into practice. Agricultural extension systems in developing countries tend to target male farmers with information that is rarely tailored to their female counterparts. A recent survey on public service delivery in Uganda found that only 16% of extension agents are women. This suggests that the extension system does not sufficiently recognise the gendered power dynamics governing intra-household information exchanges. This serves to reinforce the disempowerment of women in agricultural decision-making processes.
Testing cognitive and behavioural channels
To investigate the consequences of male-biased services, we designed an experiment in eastern Uganda involving 3,300 smallholder maize-farming households that varied both the provision of and exposure to information by gender (Lecoutere et al. 2019). We produced videos consisting of a 10-minute inspirational story in which a farmer (a man, a woman, or a couple) recount how they used to struggle with low maize yields. The videos include information about a range of productivity-enhancing strategies and discuss the costs and benefits of the different practices and inputs being promoted. They also recommend that viewers take a long-term perspective on improving their maize cultivation by starting small and reinvesting profits on increasingly large areas of land. We further varied who saw the videos in the participating households: the male co-head, the female co-head, or the co-heads together.
Receiving information empowers women
We find that targeting women within the household (as opposed to only the male co-head) with extension information has a positive effect on different domains of empowerment. This includes women’s knowledge of agronomic practices, their participation in agricultural decision-making, and their adoption of recommended practices and inputs. This is illustrated in figure 1 below, which shows average treatment effects for indices created for these four groups of outcomes. Our analysis also reveals positive effects on production-related outcomes, such as maize yield, as well as market participation.
Figure 1 Effect of involving women as recipient of information and as messenger
Most of the impact on women empowerment seems driven by the women who received the information alone, suggesting that both male and female co-heads tend to monopolise information. Whilst there is some evidence that the women who received information together with their husbands somewhat increased cooperation within the household , the effects are much weaker.
Changing perceptions
The empowerment impacts of involving women as messengers, the role model effect, seem more complicated. While we find no effect on women’s knowledge, decision-making, or adoption of recommended agronomic practices, involving women as messengers may lead to an increase in women’s use of organic fertiliser.
Looking only at households where women co-heads were targeted, we find that if these women were shown a video featuring a woman proving the information, they are more likely to take the lead in decision making. This suggests that gender homophily may be at work. Furthermore, we find that if men were shown a video featuring a woman, they are less likely to make decisions unilaterally. This suggests that involving women as messengers may have challenged beliefs and stereotypes about women being less able to make decisions related to agriculture (Beaman et al. 2009).
Policy implications: Providing information to women can empower them
Most extension strategies target the male household head, implicitly assuming that a married woman’s interaction with her husband will provide her with the necessary information on agriculture. Yet, this assumption relies on the alignment of preferences of male and female co-heads, sharing of household resources including information, and cooperation that means households reach Pareto-optimal outcomes (Fletschner and Mesbah 2011). Our research adds to the large empirical literature that rejects this assumption.
Three key policy recommendations follow from our study:
- To empower women at an individual level, boosting yields on plots that they autonomously manage, information should be provided to them directly and this can lead to considerable returns.
- To empower women in collaboration with their male household co-heads, information should be provided to them together and this could contribute to greater involvement of women in joint decision-making and joint action. However, this may not translate into better agricultural outcomes on jointly managed plots, increases in joint sales, or improvements in women’s independent agricultural outcomes.
- Featuring women as role models in information campaigns may be useful in reducing men’s dominance in the domain of agriculture, and therefore can create opportunities for greater involvement of women. Including women as role models in information campaigns targeted at women can stimulate women’s individual decision-making and action.
References
Beaman, L, R Chattopadhyay, E Duflo, R Pande and P Topalova (2009), “Powerful women: Does exposure reduce bias?”, Quarterly Journal of Economics, 124 (4): 1497–1540.
Beaman, L and A Dillon (2018), “Diffusion of agricultural information within social networks: Evidence on gender inequalities from Mali”, Journal of Development Economics, 133: 147–161.
BenYishay, A and A Mobarak (2019), “Social learning and incentives for experimentation and communication”, Review of Economic Studies, 86 (3): 976–1009.
Fisher, M and E Carr (2015), “The influence of gendered roles and responsibilities on the adoption of technologies that mitigate drought risk: The case of drought-tolerant maize seed in Eastern Uganda”, Global Environmental Change, 35: 82–92.
Fletschner, D and D Mesbah (2011), “Gender disparity in access to information: Do spouses share what they know?”, World Development, 39 (8): 1422–1433.
Lecoutere, E, D Spielman and B Campenhout (2019), “Women’s empowerment, agricultural extension, and digitalisation: Disentangling information and role model effects in rural Uganda”, International Food Policy Research Institute (IFPRI), Discussion Paper 1889..
Magnan, N, D Spielman, K Gulati and T Lybbert (2015), “Information networks among women and men and the demand for an agricultural technology in India”, International Food Policy Research Institute (IFPRI), Discussion Paper 1889.
Porter, C and D Serra (2020), “Gender differences in the choice of major: The importance of female role models”, American Economic Journal: Applied Economics (forthcoming).
Riley, E (2017), “Role models in movies: The impact of Queen of Katwe on students’ educational attainment”, CSAE Working Paper Series 2017-13, Centre for the Study of African Economies, University of Oxford, Oxford.
This article first appeared on VoxDev.
Designing gender sensitive agricultural extension information campaigns
Bjorn Van Campenhout, Els Lecoutere, David Spielman
People often end up making suboptimal choices because they lack critical pieces of information, fail to notice important features of the data they possess, or believe things that are not true. Indeed, these “informational inefficiencies” are considered an important reason why farmers in developing countries miss out on the benefits of modern technology and improved agricultural practices. In Uganda, previous research as part of the Policy Action for Sustainable Intensification of Cropping Systems (PASIC) research project suggests that a lack of information about the existence, use, and profitability of modern technologies and recommended practices constrains technical change among rice farmers. Research also shows that a simple piece of information can have a big impact on technical change. Small design attributes, such as the way in which the information is packaged or who delivers it, can dramatically improve the effectiveness of an information campaign.
An ongoing IFPRI study under the Developing Local Extension Capacity (DLEC) project—a collaboration with Digital Green funded by the U.S. Agency for International Development under the Feed the Future Initiative looks—at gender in the context of information provision by agricultural extension services. Given that extension services in many countries tend to employ male extension workers and target male household heads, little is known about what can be achieved by changing the gender of the information provider(s) and its targeted recipient(s).
This study tests if household-level outcomes—awareness, understanding, experimentation, or adoption of a technology or practice—can be influenced by small changes in the extension approach through the following pathways:
- Reducing information asymmetries between men and women in (dual) households. This is tested by comparing outcomes in which information is given to a single individual to outcomes from households in which the same information is given to both the man and the woman together.
- Promoting a “cooperative approach” to farming. This is tested by comparing outcomes in which information is given by a single individual to outcomes in which the same information is given by both the man and the woman as a couple.
- Leveraging a “homophily effect” whereby information is more effective if it is provided by someone of the same gender. This is tested by comparing outcomes in which the gender of the messenger and recipient of the message is the same to outcomes in which the genders of the messenger and recipient differ.
- Providing information. This is the simplest effect expected from extension, and is tested by comparing outcomes based only on the provision of information to any household member.
The study involves almost 4,000 maize farmers in Uganda in five eastern districts. The information intervention is implemented as a short video shown to farm households according to the experimental design: videos in which the extension information messages were presented and narrated by either a man, a woman, or a couple were shown to either the man, woman, or a couple.
The video features recommended practices on maize farming, such as inorganic fertilizer use and timely weeding and proper spacing. It also promotes a more commercial approach to farming more generally, such as starting small and re-investing part of the profits.
The videos were shown to study participants in August and September, right before the start of the growing season. Farmers will be visited again at the end of the season in February 2018 to collect information on outcomes. Data from surveys conducted during these visits will be used to evaluate the effectiveness of the video interventions.
It is too early to say anything substantive about these outcomes, so stay tuned. But a first look at data collected during the first video screenings immediately reveals the devastating impact of the fall armyworm in Uganda. While a farmer might normally harvest between 600 and 800 kg of maize per acre, the armyworm infestation in the first season of 2017 has reduced yields to only about 280 kg of maize per acre.
A look at data collected immediately after the video screening also reveals some interesting insights. To check if people actually learned something by watching the video, participants completed a short test (just three multiple choice questions) related to the content. Our analysis shows that the information treatment seemed to work: households shown a video are significantly more likely to answer the multiple-choice questions correctly. In addition, households in which the video was shown to a couple are also more likely to answer the questions correctly when compared to households in which a single individual (either the man or the woman in the household) viewed it.
However, no difference was found when the information was presented by an individual versus a couple. Also, households in which a man received information from a man, or a woman received it from a woman, scored equally well on the test as households in which the information was conveyed between different sexes. However, the latter does not necessarily mean there is no effect from leveraging a homophiliy effect on outcomes other than test scores: Even though there is no difference in terms of what people learn from the video, people may act upon information differently depending on who delivers the information and who is targeted, as farmers may deem the information more credible if it comes from someone of the same gender.
We must wait until after the maize harvest to see if paying attention to the sex of messenger and receiver in the context of ICT-mediated information provision also leads to changes in the adoption of modern technologies and improved agronomic practices. But even in this early stages of this research, two results emerge. First, extension video messages are an effective way of increasing knowledge on improved technologies, inputs, practices among our sampled maize farmers in Uganda. Second, results confirm that targeting information to couples increases extension effectiveness, at least in terms of improving their knowledge about the technologies and practices.
This article first appeared as an IFPRI research blog.