54-country panel finds no robust African production decline during COVID-19

A longitudinal mixed-methods study of 54 African countries and Zimbabwe found no robust evidence that the COVID-19 period was associated with lower agricultural productivity or cereal yields, while estimates for undernourishment were suggestive but sensitive to model specification. The study, published in Frontiers in Sustainable Resource Management, did not establish that agriculture was unaffected or resilient: it found that production continuity could coexist with impaired food access, heightened vulnerability and incomplete recovery in Zimbabwe’s post-land-reform farming systems.

The researchers combined a 2000–2022 panel covering 54 African countries with a thematic synthesis of 13 documents representing 11 evidence families relevant to Zimbabwe. Depending on data availability, model-specific samples contained 41–53 countries. Segmented fixed-effects models estimated immediate pandemic-period deviations and subsequent trajectory changes in agricultural value added per worker, cereal yield and the prevalence of undernourishment. Two-way fixed-effects specifications, dynamic models, interaction tests, country-placebo analyses and numerous robustness checks examined Zimbabwe’s relative trajectory and the moderating roles of inflation and irrigation. The documentary analysis was designed to identify mechanisms, contextual conditions and competing explanations rather than to validate individual coefficients.

In the principal panel models, pandemic-period terms were not statistically associated with agricultural productivity or cereal yield. For productivity, the immediate pandemic-period coefficient was 0.0154 in the principal specification, with a p-value of 0.580, while the trajectory coefficient was −0.0051, with a p-value of 0.789; their joint test was also insignificant, F(2, 50) = 0.164, p = 0.850. A channel-adjusted model, which included macroeconomic and financial variables and used a smaller complete-case sample, produced a positive immediate coefficient of 0.0608 and a negative trajectory coefficient of −0.0383. Its joint result, p = 0.0505, was just above the study’s 5% threshold. Cereal-yield terms were insignificant in both principal and adjusted models, with joint p-values of 0.787 and 0.583.

The food-security results were less consistent. In the principal model, the immediate pandemic-period estimate for the prevalence of undernourishment was 1.222 percentage points, with p = 0.058, and the trajectory estimate was 0.464 percentage points, with p = 0.077. Their joint test was significant only at the 10% level, F(2, 48) = 2.695, p = 0.078. After adjustment for macroeconomic and financial conditions, both coefficients were insignificant and the joint test was F(2, 40) = 0.780, p = 0.465. The researchers therefore did not support the hypothesis of a positive pandemic-period association with undernourishment at the conventional 5% threshold, but they described the principal result as suggestive rather than as evidence that food insecurity was unaffected. Some alternative covariance and climate specifications also produced positive significant coefficients, confirming sensitivity in the food-security estimates.

Rainfall was more consistently associated with cereal yield than were the pandemic-period terms. A one-unit increase in the standardized rainfall anomaly was associated with approximately 3.9% to 4.1% higher yield in the principal and channel-adjusted models, whose coefficients were 0.0385 and 0.0402. The analysis also found that agriculture’s share of GDP increased after 2019, by 1.918 percentage points in the principal specification, with a 95% confidence interval of 0.854 to 2.982. The authors cautioned that this could reflect sharper contraction in other sectors rather than genuine productivity growth. The distinction matters because agriculture can absorb displaced labor while productivity remains low and household purchasing power deteriorates.

Zimbabwe’s estimated trajectories differed from the pooled pattern for the rest of Africa across all three outcomes. Its 2020 differential was negative for agricultural productivity, with an interaction coefficient of −0.211, followed by a positive annual trajectory differential of 0.131; the modeled gap narrowed to −0.079 in 2021 and reversed to 0.052 in 2022, although the 2021 marginal estimate was not significant. Zimbabwe’s cereal-yield differential was positive in 2020, with a coefficient of 0.489, but declined as the trajectory coefficient of −0.173 was applied in subsequent years, producing a modeled differential of 0.144 in 2022. The undernourishment differential was negative and widened: the immediate coefficient was −2.608 percentage points and the trajectory coefficient was −3.447 percentage points, producing modeled differentials of −6.054 points in 2021 and −9.501 points in 2022. Joint interaction tests for all three outcomes remained significant after Holm adjustment.

Those results identify differences from the pooled African comparison, not uniquely Zimbabwean performance. Country-placebo analyses, in which each eligible African country was treated as the focal case, produced empirical p-values of 0.760 for productivity, 0.176 for cereal yield and 0.208 for undernourishment. Zimbabwe’s estimates were therefore not exceptional within the wider distribution of country-level patterns. The interaction results also do not identify a causal effect of land reform: the analysis was not a staggered-adoption difference-in-differences design, and the pandemic-period coefficients may also reflect unobserved events common to African countries during 2020–2022. The findings consequently describe conditional temporal associations and relative trajectories rather than the effects of a single cause.

The documentary synthesis indicated that Zimbabwe entered the pandemic with hyperinflation, drought and rainfall variability, rising input costs, livestock disease and institutional constraints. Restrictions and responses operated through interconnected labor, input, transport, market, financial and extension channels. Production did not collapse uniformly: favorable rainfall, climate-proofed practices and coordinated input programs supported aggregate maize recovery, while adverse outcomes were concentrated in particular provinces, commodities, livestock activities and market-dependent enterprises. Food access, however, did not necessarily track output. Inflation, unstable income, rising retail prices, lower farmgate prices and transport barriers weakened purchasing power and dietary quality even where food remained available or production improved. Effects also varied by geography, commodity perishability, market channel, gender, employment status, program participation and access to resources.

Adaptation included collective action, savings groups, digital communication, repair alliances, livelihood changes and conservation practices, but some responses were temporary, exclusionary or risky. Extension services, input programs, social assistance, market linkages and institutional coordination provided support, although access depended on funding, transport, inputs, infrastructure and implementation capacity. Recovery was partial and non-linear: production, employment, debt and food indicators sometimes improved or stabilized, while service access worsened, assistance declined and severe coping continued. The study’s evidence therefore supports the proposition that land access becomes productive and adaptive when complemented by knowledge, finance, infrastructure, inputs, markets and capable institutions, rather than demonstrating that redistribution itself caused resilience. Direct evidence comparing Zimbabwe’s A1 smallholder and A2 farming systems was sparse, limiting conclusions about differences between those categories.

The analysis also tested whether inflation and irrigation moderated the pandemic-period outcomes. Inflation did not significantly moderate the association with undernourishment, with a joint p-value of 0.461, despite documentary evidence that price pressures weakened purchasing power. Irrigation’s immediate interaction with productivity was positive and individually significant, with a coefficient of 0.006 and p = 0.031, but the trajectory term was insignificant and the joint test was p = 0.064 before Holm adjustment and p = 0.129 afterward. Irrigation did not robustly moderate cereal yield. A supplementary interaction between irrigation and rainfall was positive for productivity, indicating a more favorable rainfall association at higher irrigation coverage, but not necessarily that irrigation buffered adverse rainfall. These findings leave irrigation as a conditional capability rather than a general shock absorber.

Several limitations constrain interpretation. The panel had no conventional untreated pandemic control, contained only three pandemic-period observations and used unbalanced model-specific samples. National annual indicators can obscure seasonal, crop-specific and subnational variation, while the slow-moving undernourishment measure may miss short-term changes in affordability, diet quality or household access. Channel-adjusted models relied on smaller complete-case samples, dynamic fixed-effects estimates may face finite-panel bias, and full year fixed effects could not be included with the common post-2020 terms in the baseline segmented models. The documentary corpus was purposively selected, its sources differed in design and scope, some outputs were dependent, and coding was conducted by one analyst, so intercoder reliability could not be claimed. Pandemic restrictions also coincided with rainfall variability, drought, inflation, input costs, livestock disease and longer-standing structural vulnerabilities. The study’s central conclusion is therefore conditional: its evidence does not support a simple collapse narrative or a claim of proven resilience, and neither production continuity nor Zimbabwe’s distinctive trajectory can be attributed to land reform alone.

Subject of Research: Mixed-methods analysis of agricultural production, food security and land reform in 54 African countries with a Zimbabwe case study

Article Title: Land reform, agricultural resilience and food security under systemic shock: evidence from Africa

Article References: Tabani Mpofu, R. (2026). Land reform, agricultural resilience and food security under systemic shock: evidence from Africa. Frontiers in Sustainable Resource Management, 5, Article 1912341. https://doi.org/10.3389/fsrma.2026.1912341

Image Credits: AI Generated

DOI: 10.3389/fsrma.2026.1912341

Keywords: African agriculture, COVID-19, land reform, food security, agricultural resilience, Zimbabwe, cereal yields, undernourishment, mixed methods, irrigation, agricultural productivity, systemic shocks