Report on a Study to Crowdsource Farmgate Prices for Maize and Soybeans in Malawi

This report summarizes the findings from an innovative study to collect the prices that farmers received for maize and soybeans during the 2020 main marketing season in Malawi. Between April and July, whenever they sold maize or soybeans, farmers were asked to report the prices they received by calling or texting a toll-free number managed by Farm Radio Trust. Reported prices were then compared to the minimum farmgate prices set by the Ministry of Agriculture and Food Security. Our findings show that 75 percent of maize farmers and 90 percent of soybean farmers sold their crops below the official minimum farmgate prices. On average, prices received by these farmers were approximately three-quarters of official minimum farmgate prices.


Introduction
Since the early 2000s, Malawi's Ministry of Agriculture and Food Security (MoAFS) has announced minimum farmgate prices for around 21 crops at the beginning of each agricultural season. These minimum farmgate prices are notionally based on small farmers' typical costs of production and aim to protect farmers against extremely low prices. However, despite many anecdotal stories of traders offering farmers such low prices, consistent and accurate data on farmgate prices of agricultural commodities is scarce in Malawi: so, there is little hard evidence on whether minimum farmgate prices are enforced.
Crowdsourcing farmgate prices can augment the data on farmgate prices collected by MoAFS' Agricultural Markets Information System (AMIS) at relatively low cost. 1 Crowdsourcing is an innovative phone-based approach to collecting data on the price farmers receive. It relies on farmers reporting to a free phone call center the price and other attributes of their recent crop sales. To encourage smallholder farmers to report their recent crop sales, an inducement is offered in the form of a chance to win a coupon for farm inputs in a weekly raffle. 2 This facilitates the collection of robust, although self-reported, prices, which may be used to inform policy makers about the efficiency of agricultural markets and the effectiveness of price policies. Together with data on retail prices, such data can support the analysis of farm-retail margins and the competitiveness of alternative marketing arrangements. It can also help in examining the effects of the global COVID-19 pandemic, which has disrupted the global and national agri-food systems, with significant impacts on global food security, nutrition, and poverty (Arndt et al. 2020;Devereux, Béné, Hoddinott 2020;Laborde et al. 2020). In the longer term, better price information and more competitive farmgate prices are likely to stimulate increased production of many crops.
Most agricultural price monitoring in Malawi focuses on retail prices and, until now, only MoAFS through AMIS collects data on farmgate prices across the country. However, AMIS only collects farmgate prices over some time periods in selected years, which results in substantial data gaps. The only other source of farmgate prices is FAOSTAT (FAO 2020), a statistic web portal, managed by the Food and Agriculture Organization of the United Nations (FAO). Even though consistent farmgate price data is scarce in Malawi, available data from FAOSTAT show that the farmgate prices of maize and soybeans have fluctuated markedly over the years.  To close these data gaps, IFPRI crowdsourced the prices farmers receive for maize and soybeans during the 2020 main marketing season through the Farm Radio Trust (FRT) platform. Both crops play an important role in Malawi's agriculture as a source of farm income and for their contribution to farm household food and nutrition security (Ochieng et al. 2020). The study drew on lessons from a pilot crowdsourcing study covering pigeon peas and chickpea prices in 13 districts in southern Malawi conducted between August and October 2019. The pilot study found that crowdsourcing is a cost-effective method to source farmgate prices in a timely manner and that most farmers sold their pigeon peas and chickpeas below the minimum farmgate prices.
The rest of the report is organized as follows. The next section describes the methodology employed in the study in terms of sampling procedure, data and analytical approaches. Section 3 presents and discusses the study findings while section 4 concludes with some policy implications.
2 See Section 2 for further details. As far as the authors are aware, the crowdsourcing of farmer prices is set to applied in other countries. 3 It should be noted that the nominal producer prices reported by FAOSTAT are higher than AMIS prices for the same periods. In many instances, they were even higher than the average retail prices of the two crops collected by MoAFS.

Methodology
We employed a phone-based approach to crowdsource maize and soybean prices from farmers countrywide. To raise farmers' awareness of this exercise, radio jingles (adverts) were aired four times a day at prime times by three major radio stations in Malawi (Malawi Broadcasting Corporation, Zodiak and Radio Livingstonia) asking farmers to report on their recent sales of maize and soybeans to a toll-free line. See Appendix 1 for the script of these radio jingles.
Data for this study were collected in two phases. The first phase involved crowdsourcing of the prices maize and soybean farmers received. To encourage farmers' participation, callers were entered into a weekly draw and stood a chance of winning a coupon worth MWK 25,000 (approximately US$34). The winners could redeem the coupons at any store of a major agro-dealer (Agora/Farmers World) for agricultural inputs. This promotion and weekly draws lasted for four months (April-July 2020), with an additional grand draw during the last week of July, in which the winners received MWK 80,000 worth of coupons for agricultural inputs.
In addition to the sales prices, farmers who called the toll-free number were asked to provide information on their gender; volumes sold; type of buyer, location of sales; distances to and name of their nearest market. Farmers were also asked if and how the COVID-19 pandemic affected their maize and soybean marketing. See Appendix 2 for the checklist of questions that FRT call center operators asked farmers who called the free phone number.
In the second phase during August and September, a follow-up phone survey was conducted with those farmers who reported sales during the first phase of the study to better understand the marketing environment of the two commodities. Farmers were asked about their age; the size of their household and farm; area of the two crops cultivated; quantities harvested, sold and retained as seed; bargaining processes; quality assessment by both parties to the transactions; price expectations, awareness and usefulness of the minimum farmgate prices; and intra-household decision making on sales and use of sales proceeds. See Appendix 3 for the questionnaire used in the follow-up phone survey. Descriptive statistics, such as frequencies, percentages are presented in the following section. Table 1 presents a summary of farmers calling-in during the four months of the harvest season and those who responded to a follow-up phone survey. A total of 2,313 farmers (1,048 for maize and 1,265 for soybeans) reported their sales by phone during April-July. 4 A follow-up phone survey was conducted during August and September to better understand the marketing environment of the two commodities as detailed in the following discussion. In total, 850 of the maize farmers and 1,080 of the soybean farmers who reported prices between April and July were interviewed during the followup phone survey. Source: Crowdsourced data (April-August 2020). 4 A small number of farmers reported the same transaction multiple times, presumably to increase their chances of winning a coupon. These duplicate entries for both the initial calls and follow-up survey are excluded from our subsequent analysis. Figure 2 shows the distribution of farmers by district. Most of the farmers were from the main maize and soybean producing districts in central Malawi, although prices were reported from farmers in 27 of Malawi's 28 districts. 5

Figure 2. Number of farmers reporting by district
Source: Crowdsourced data (April -July 2020).

Study findings
This section presents the findings from both the initial calls from farmers and the follow-up phone surveys. We begin by presenting farmers' profile (gender, age, farming experience and land holding) and the summary statistics of buyers of maize and soybean, crop transactions (volumes, prices, buyers, rounds of negotiation and relationship with buyers), and market characteristics (distance to markets, location of sales).  Source: Crowdsourced data (April-July 2020). Table 3 presents additional summary statistics of participating farmers. Both maize and soybean farmers were similar in terms of demographic profiles. On average, farmers were 36 years of age with about twelve years of experience in farming. The farmers were largely small-scale by orientation, owning an average of four acres of land and allocating two acres to maize and one to soybean production. The average quantity harvested was 2,167 kg (about 43 bags) for maize and 961 kg (19 bags) for soybeans. About 46 percent of harvested maize was sold while 7 percent retained for seed. In contrast, the commercialization rate for soybean was higher with 70 percent of the harvest 5 There were no callers from Likoma district, which comprises several remote islands in Lake Malawi. Likoma is not a major maize and soybean producing district. sold and 7 percent retained for seed. The distance from farm to nearest main road (a proxy of access to transport) was 7 kilometers (km). Note: Kg = kilogram; km = kilometer.
Source: Crowdsourced data (April-July 2020). Figure 3 presents a summary of buyers of maize and soybeans and the location of sale of the two crops. About 79 percent of maize and 89 percent of soybean farmers sold to assemblers or retailers. The second most common buyer was the Agricultural and Market Development Corporation 6 (ADMARC) (19 percent of maize and 5 percent of soybean farmers). A few sales were made to other buyers such as consumers, farmers' cooperatives and food and animal feed processors. In terms of location of sales, farmgate and market were the most common points of sale. Most of the maize sales were made at a nearby market or trading center (41 percent) closely followed by the farmgate (39 percent). However, most soybean sales were made at the farmgate (52 percent) followed by market or trading center. Sales to ADMARC and large traders were often made at their depots, which are spread across Malawi.

Figure 3. Buyers of maize and soybeans and location of sales
Source: Crowdsourced data (April-July 2020). 6 ADMARC is a parastatal corporation with an extensive depot network, which mandated to market most agricultural commodities in Malawi.  Table 4 shows the characteristics of maize and soybean transactions reported in the April-July period in terms of volumes traded. The upper part of the table presents the summaries for maize (Panel A) while the lower part presents the summaries for soybeans (Panel B). Interestingly, larger volumes of the two crops were sold to food and animal feed processors, large traders, and farmers cooperatives even though these categories of buyers were the least common. The largest average sales of maize and soybeans were 1.4 MT and 1.2 MT respectively to processors, and 4.4 MT and 0.9 MT to large traders, respectively. However, some individual transactions were as large as 10 MT for maize and 5 MT for soybeans (column 6). In contrast, some sales of just 10 or 20 kg were reported (column 5) for both crops. The modal volume of maize and soybeans sold was 500 kg (about ten 50 kg bags). Figure A1 in Appendix 4 presents the distribution of volumes sold by farmers, while Figure A2 shows plots of the variation of ask-offer prices over time in two major maize and soybean producing districts in central Malawi. Source: Crowdsourced data (April-July 2020). Table 5 presents a summary of maize and soybean prices and transaction characteristics such as the reported average price/kg of the two crops, farmers' ask and buyers' offer prices, rounds of price negotiations, number of previous transactions with the buyer and number of buyers who visited the farm in the last seven days. For maize (Panel A), the average price was MWK 152/kg with reported prices ranging between MWK 50/kg and MWK 280/kg. The modal (most common) price for maize was MWK 200/kg and for soybeans MWK 250/kg. Figure A1 in Appendix 4 presents the distribution of prices reported to the FRT call center. Strikingly, 75 percent of farmers sold below the official farmgate price of MWK 200/kg during the 2019/2020 marketing season (MoAFS 2020). On average, farmers asked for prices above the official farmgate price (MWK 211/kg) whereas buyers offered prices below the official price (MWK 151/kg). The average transaction involved one round of price negotiation with buyers before a price was agreed. In typical price negotiations, farmers propose a price (ask price) while buyers also propose a price they were willing to offer (offer price) and haggling continues until both parties settle for their reserve prices, which then becomes the final sales price.

0%
On average, most sales were made to buyers who had transacted twice with the farmer on previous occasions. Visits to the farm by prospective buyers were limited to an average of three buyers in the seven days prior to the transaction. This is plausible given the market disruptions caused by the COVID-19 pandemic, which restricted trade in agricultural commodities during the study period (Aggarwal et al. 2020;Baulch, Botha, Pauw 2020;Chadza et al. 2020). We also calculated a simple measure of farmers' bargaining power, which shows their commitment to the initial ask price (sales price-offer price/ask price-offer price). This measure shows that both maize and soybean farmers had weak bargaining power (alpha value tending to zero). 7 About 95 percent of maize farmers accepted buyers' initial offer price.
For soybean farmers (Panel B), the average sales price was MWK 231/kg with reported prices ranging between MWK 80/kg and MWK 400/kg. On average, farmers asked for prices above the official price while buyers offered prices below the official price. Again, 90 percent of farmers sold below the official farmgate price of MWK 300/kg during the 2019/2020 marketing season (MoAFS 2020). On average, soybean transactions involved two rounds of price negotiations before a price was agreed. Most sales were to buyers who on average had transacted twice with the farmer before. Visits to the farm by prospective buyers were limited to about three buyers in the seven days prior to the transaction. Soybean farmers also had weak bargaining power (alpha= 0.04) as about 89 percent ended up accepting buyers' initial offer price.
As farmers' offer prices will normally be influenced by future price expectations, farmers were asked whether they expected prices to remain the same, fall, or rise during the remainder of the marketing season ( Figure 4). About 54 percent of maize and 35 percent of soybean farmers expected future prices to increase while less than 30 percent expected prices to fall. A further 31 percent of maize farmers and 39 percent of soybean farmers expected prices to remain the same during the rest of the season. Over 60 percent of maize and soybean farmers were aware that official minimum farmgate prices existed for the two crops ( Figure 5). However, when asked to state the actual official minimum prices, seven percent of those who were aware of the prices cited incorrect figures. Furthermore, only about 67 percent considered them useful in their price negotiations with traders. This is an important finding as it points to possible widespread lack of awareness and enforcement of the official minimum farmgate prices and helps explain why traders' offer prices were typically lower than the official prices (see Table 5). Towards the end of the interview, call center operators asked farmers about decision making on crop sales and the use of sales proceeds. From Figure 6, 57 percent of both maize and soybean farmers decided on sales and use of sales proceeds jointly with their spouses while 35 percent mentioned unilateral decisions by household heads. Interestingly, despite extensive women involvement in the production of both crops, only about 7 percent of male farmers mentioned that the decision on the use of crop proceeds was made solely by their spouses. As a wrap up question, farmers were asked about the use of crop sales proceeds and the summary is presented in Table 6. More than half of the farmers used the income from maize and soybean sales to pay for farm labor (ganyu) and other farm inputs such as seeds, fertilizers and other crop chemicals (56 percent and 54 percent of maize and soybean farmers respectively). Other major uses of sales proceeds included payments of household expenses, food and school fees and related expenses as cited by at least 10 percent of the farmers. This finding is plausible given that Ochieng et al. (2019) found that farmers make 'distress sales' at the peak of the main harvest marketing or lean seasons to meet pressing cash needs. In Malawi, school fee payments coincide with the peaks of the lean seasons and thus lead to distress sales (Dillon 2017  Purchase food 11 13 Pay school fees and related expenses 10 11 House construction/improvement 4 3 Purchase livestock 1 2

Repay loans 1 1
Start a business 1 1 Save the proceeds 1 0 Source: Follow-up phone interviews with farmers (August-September 2020).

Conclusion
This study used an innovative call center-based method to crowdsource the prices that Malawian maize and soybeans farmers received during the 2020 marketing season. Through phone calls with farmers who responded to radio jingles, we established that most maize and soybeans were sold either on farm (39 percent for maize and 52 percent for soybean) or at nearby markets (41 percent for both crops), and were mainly sold to assemblers and retailers. A small number of farmers sold directly to consumers. Most sellers (90 percent) were male and two-thirds (64 percent) came from the central region of Malawi, which contains country's principal maize and soybean producing districts. However, sales prices were reported from all districts in Malawi except Likoma.
About 75 percent of maize farmers and 90 percent of soybean farmers sold their crops below the official minimum farmgate prices of MWK 200/kg for maize and MWK 300/kg for soybeans. On average, prices received by these farmers were approximately three-quarters of official minimum farmgate prices. While about two-thirds of farmers were aware of official minimum farmgate prices, many of them could not state these prices accurately. Furthermore, about one-third of those who were aware of minimum farmgate prices did not find them useful when negotiating prices with buyers. Overall, farmers had weak bargaining power vis-à-vis buyers.
These findings point to a widespread lack of awareness and enforcement of minimum farmgate prices, and therefore their limited usefulness in raising prices farmers receive for their crops. Previous studies show that, while costly and informal, agricultural markets in Malawi are broadly efficient (Sitko and Jayne 2014). Therefore, rather than calling for stricter enforcement of minimum prices, we suggest encouraging greater competition at all levels of the market chain. This can be a more effective way to ensure that farmers receive better prices.
The results of this study are in line with findings from an earlier crowdsourcing study for pigeon peas and chickpeas in southern Malawi between August and October 2019 (Ochieng 2019). Both studies confirm that official minimum farmgate prices are weakly enforced and are therefore ineffective in guaranteeing minimum prices to farmers. Joint decision between household head and spouse iv)

ABOUT THE AUTHORS
Another adult household member 10. What did you spend most of the sales proceeds on? [Tick two most important] i) To pay school fees and other related expenses ii) To purchase farm inputs iii) To purchase farm inputs iv) To purchase food v) To pay medical expenses vi) To pay for household expenses vii) Other…