Production- and Consumption-Oriented Interventions to Accelerate Crop Varietal Turnover in Nigeria

This study will evaluate the impact of production- and consumption-related interventions to address informational and behavior constraints of farmers and consumers in adopting new varieties of maize and cowpea.

The interventions involve (1) provision of seed trial packs (each 6-kg minikit consists of 1.5 kg each of white maize, orange maize, white cowpea, and brown cowpea) and (2) a consumption-side intervention that provides grain/flour mini-packs and village-level cooking/tasting demonstrations.

The varieties to be promoted are (1) newly bred varieties, with commercial seeds available in the market but not widely known and adopted by the farmers; and (2) varieties with good potential for widespread adoption based on published traits of these varieties and recommendations from breeders and feedback from seed companies. This experiment will promote the following varieties:
• SAMPEA19 (white cowpea): early medium maturity (70–75 days); medium seed size; tolerant to striga, drought, and Alectra; potential yield of 2.7 t/ha; erect; good for Sudan and Northern Guinea savannas
• FUAMPEA3 (brown cowpea): high-yielding with a potential yield of 2.4 t/ha; large seed size; medium maturity (80–85 days); semi-indeterminate; good for intercropping; suitable for Northern and Southern Guinea savannas
• SAMMAZ51 (white maize): resilient to multiple stresses; tolerant to drought and Striga hermonthica; high grain yield (8.5 t/ha); suitable for processing and consumption in the form of thick porridge and other food products
• SAMMAZ52 (orange-fleshed maize): high yielding with a potential yield of 6 t/ha; medium maturity period (100–120 days); biofortified for extra nutrition benefits, with intermediate levels of provitamin A content (9.8µg/g); mostly suitable for green maize and processing into diverse foods including child-weaning foods; suitable in Northern Guinea and Sudan Savannas.

The baseline survey shows that farmers in the study sites overwhelmingly grow and prefer white cowpea and white maize. The qualitative interviews and FGD discussions reveal, however, that some dishes are prepared with brown cowpea or orange maize.

2023-07-03

2023-12-31

The primary outcome is adoption of improved varieties promoted, with “adoption” defined as farmers planting the variety in the cropping season (binary variable)

“Improved variety” is defined as (1) an officially released variety or a local name that matches an official released variety by extension agents, agrodealers, and breeders; (2) seed acquired in a certified/labeled/sealed bag; and (3) seed acquired in the last two years for OPV maize (if OPV maize is not recycled within the last two seasons, seeds lose vigor), newly acquired with no recycling for hybrid maize, or acquired in the last three years for cowpea (can only be recycled within the last three seasons).

The ITT effects (of secondary outcome indicators):
• Land cultivated with the varieties promoted (in hectares)
[f02a, f02b, f02b_oth, f03a + variety planted above]
• Planned cultivation of varieties promoted in the next season (in hectares)
• Share of land area with maize and cowpea planted with varieties promoted (percent)
• Planned share of land area with maize and cowpea planted with varieties promoted (percent)
• Land cultivated with any improved variety (in hectares)
• Planned land cultivation with any improved variety in the next season (in hectares)
• Share of land area with maize and cowpea planted with improved varieties (percent)
• Planned share of land area with maize and cowpea planted with varieties (percent)
• Yield: value of production (naira per hectare) is used due to commonly practiced intercropping in the study area [maize_harvest_type, mz_G8_01a, mz_G8_01_units, mz_price_on_farm, mz_price_on_farm_unit, mz_price_off_farm, mz_price_off_farm_unit, cw_G8_01a, cw_G8_01_units, cw_price_on_farm_unit, cw_price_off_farm, cw_price_off_farm_unit]

The spillover of non-treated (SNT) involves the same indicators as the ITT effect, except that the sample is the within-cluster nontreated households, compared to the control group:
• Adoption of any variety promoted (0/1 if the household planted the variety promoted)
• Planned adoption of any variety promoted in next season (0/1)
• Land cultivated with the varieties promoted (in hectares)
• Planned cultivation of varieties promoted in the next season (in hectares)
• Share of land area with maize and cowpea planted with varieties promoted (percent)
• Planned share of land area with maize and cowpea planted with varieties promoted (percent)
• Land cultivated with any improved variety (in hectares)
• Planned land cultivation with any improved variety in the next season (in hectares)
• Share of land area with maize and cowpea planted with improved varieties (percent)
• Planned share of land area with maize and cowpea planted with varieties (percent)
• Yield of maize and cowpea: value of production (naira per hectare) is used because of commonly practiced intercropping in the study area
Other outcomes to explore:
• Income from maize and cowpea sales [mz_G8_01k, cw_G8_01k + yield and price data above]
• Gender parity indicators (composite index, and several subindexes, following the Abbreviated Women’s Empowerment in Agriculture Index, A-WEAI—see Malapit et al. 2017) [7 modules on A-WEAI]
Other indicators to collect related to the valuation and experience of farmers on the varieties promoted, in comparison to current varieties:
• Production cost
• Other (desirable and undesirable) production traits
• Processing, milling, and cooking qualities
• Desirable and undesirable qualities of the maize and cowpea in terms of color, size, taste, texture, appearance, and others
• Desirable and undesirable qualities of the maize and cowpea when cooked (in different dishes)

The study will employ a cluster randomized controlled trial, with villages randomly assigned to different treatment arms:
• Treatment 1: Villages randomly assigned to receive seed trial pack (before Year 1 rainy season planting)
• Treatment 2: Villages randomly assigned to receive consumption-oriented intervention (grain/flour trial pack and cooking/tasting demonstration and facilitation in which the primary woman and man decision-makers within the household are invited (after harvest in Year 1)
• Treatment 3: Villages randomly assigned to receive seed trial pack (before Year 1 rainy season planting) + consumption-oriented intervention (grain/flour trial pack and cooking/tasting demonstration and facilitation in which the primary woman and man decision-makers within the household are invited) (after harvest in Year 1)
• Treatment 4: Villages randomly assigned to receive seed trial pack + training on good agricultural practices (before Year 1 rainy season planting) + consumption-oriented intervention (grain/flour trial pack and cooking/tasting demonstration and facilitation in which the primary woman and man decision-makers within the household are invited) (after harvest in Year 1)
• Control: No intervention

Not available

Randomization done in office by a computer

cluster=village

Yes

252 clusters

2016 households

T1: 344 households (seed trial pack treatment)
T2: 344 households (consumption pack treatment)
T3: 344 households (seed trial pack + consumption treatment)
C: 344 households (pure control group)
T4: 320 households (seed trial pack + consumption + agriculture training)
S: 320 households (non-treated in treatment T4 villages to study spillover)

We use the following assumptions for the power calculations. The primary outcome we use is a binary indicator for use of improved variety at the farmer level. According to LSMS data for 2018, 11 percent of maize growers used improved varieties and 3 percent of cowpea growers used improved cowpea varieties. In the two focus states (Bauchi and Kaduna), the average adoption rate was 12 percent for maize and 2.5 percent average adoption rate of improved varieties for cowpea. Intracluster (within-village) correlation for this outcome has been estimated to be 0.12. As in equations 1–3, we measure the impact of the treatment assignment by testing these null hypotheses— β_1= 0, β_2= 0, β_1= β_2—and measuring pooled effects for the treatments. In order to minimize costs and best use our available resources, we targeted 252 villages (8 households per village), given the setup of 2 enumerators per village, questionnaire length, and survey team setup. At baseline, we surveyed a total of 2,016 households. A total of 50 villages were randomly assigned as the control group (400 households); 101 villages were randomly assigned as Treatment Arm 1 (seed trial pack) (4 households randomly assigned as treatment and 4 households randomly assigned as within-cluster control); and 101 villages were randomly assigned as Treatment Arm 2 (seed trial pack + training) (4 households randomly assigned as treatment and 4 households randomly assigned as within-cluster control). With this setup, the minimum detectable ITT and SNT effects (β_1 and β_2) are 5.5 percentage points for cowpea and 9 percentage points for maize, for the sample size of 420 households per treatment arm. The assumption is a two-sided test with a 5 percent significance level and a power of 80 percent, and standard errors are clustered at the village level. The minimum detectable effects are reasonable and well below the expected impact of the interventions. Some breeders and key informants expect that the interventions will more than double the adoption rate for maize (i.e., an 18-percentage-point increase in maize adoption from 11 percent before to 29 percent after intervention) and at least quadruple the adoption rate for cowpea (i.e., a 7.5-percentage-point increase in cowpea adoption from 2.5 percent before to 10 percent after intervention).