Effect of government intervention in relation to COVID-19 cases and deaths in Malawi

The study examines whether and to what extent government non-pharmaceutical interventions (NPIs) in Malawi reduced COVID-19 confirmed cases and deaths. Globally, countries deployed a range of nine yardstick interventions (e.g., school/workplace closures, restrictions on gatherings and movement, public information campaigns, travel controls) with varying degrees of strictness and adherence. Malawi declared a national disaster on 20 March 2020 and registered its first two cases on 2 April 2020. The government implemented measures such as international travel bans, school closures, cancellation of public events, decongesting workplaces and public transport, mandatory face coverings, and testing for symptomatic individuals. A proposed 21-day national lockdown (from 21 April 2020) was blocked by the High Court. Given mixed global evidence on the effectiveness of stringency measures, and the reliance on NPIs in low- and middle-income countries due to vaccine inequities, Malawi provides a pertinent case. As of 25 April 2022, Malawi recorded 85,747 confirmed cases and 2,633 deaths—far fewer deaths than early expert projections. The core research question is whether the relatively low deaths and high recovery rates are attributable to government stringency measures or other factors. Methodologically, the paper addresses endogeneity between policy stringency and outcomes by using an instrumental variable strategy and focuses on a single-country case study to avoid cross-country heterogeneity biases.

The literature presents mixed evidence on the efficacy of government stringency measures against COVID-19. Several studies report strong negative effects of stringency on transmission and deaths (e.g., Hsiang et al., 2020; Achuo, 2020; Arshed et al., 2020; Haldar and Sethi, 2020; Hadianfar et al., 2021). African-focused works (Achuo, 2020; Carlitz, 2021) find inverse long-run effects of stringency on incidence rates. Global analyses across multiple waves (Hale et al., 2021) suggest that policy effectiveness persists across successive waves, with timing and strictness (Dergiades et al., 2020, 2022) and economic status (Ratto et al., 2021) influencing outcomes. Conversely, others find limited or no measurable impact of stringent measures on cases or deaths (Berry et al., 2021; Gibson, 2020/2022). Ratto et al. (2021) note ineffectiveness in the initial wave but significant negative associations in subsequent waves. In LMICs, vaccine inequity (COVAX shortfalls) heightened reliance on NPIs (Hassan et al., 2021; Tatar et al., 2021; Kunyenje et al., 2023), underscoring the policy relevance of evaluating stringency effects in contexts like Malawi.

Data: Daily country-level data for Malawi from the Oxford COVID-19 Government Response Tracker (OxCGRT)/Our World in Data, from 20 February 2020 to 25 April 2022. Additional dummy controls capture seasonal and political periods: planting season (November–March), harvest season (April–June), and an election dummy (0 through July 2020; 1 thereafter) to account for shocks from political demonstrations, campaign rallies, and inauguration events. Variables: Outcomes are cumulative confirmed COVID-19 cases and confirmed deaths. The key policy variable is the OxCGRT stringency index (0–100), a composite of nine interventions: school and workplace closures; cancellation of public events; restrictions on public gatherings; closure of public transport; stay-at-home requirements; public information campaigns; restrictions on internal movement; and international travel controls. Controls include planting, harvest, election dummies, and a time trend (as used in regressions). Identification strategy: To address potential reverse causality between rising cases/deaths and tighter government policies, the study employs an instrumental variable (IV) approach, instrumenting the stringency index with the exogenous policy shift marked by the date of constitution of Malawi’s COVID-19 task force. The instrument is argued to be exogenous (policy-timing driven, decided at a presidential cabinet meeting) and affects outcomes only through its impact on policy stringency. Instrument relevance is assessed via first-stage F-statistics using the Stock–Yogo rule-of-thumb. Empirical specification: Y_i = β0 + β1 Stringency_i + β2 Election_i + β3 Harvest_i + β4 Planting_i + (Time trend) + ε_i, where Y_i is confirmed cases or deaths. Both IV and OLS estimates are reported for robustness. Sample size: N = 805 daily observations. First-stage diagnostics and descriptives: The first-stage shows strong instrument relevance (Adj. R^2 = 0.62; F = 465.27; p < 0.01). Correlations show an inverse co-movement between stringency and outcomes: corr(stringency, cases) = -0.408 and corr(stringency, deaths) = -0.418 (both p < 0.01). Descriptively, by 25 April 2022, Malawi had 85,747 cases and 2,633 deaths (~97% survival), with the stringency index averaging ~45 (max ~65).

- Instrument relevance: First-stage F-statistic = 465.27 (p < 0.01), indicating a strong instrument. - Correlations: Stringency is negatively correlated with outcomes: cases r = -0.408 and deaths r = -0.418 (both p < 0.01). - Main IV effects: • Cases: A 1-point increase in the stringency index is associated with approximately 179 fewer confirmed cases (IV: -178.826, t ≈ -11.93, p < 0.01). • Deaths: A 1-point increase in the stringency index is associated with approximately 6 fewer deaths (IV: -6.230, SE 0.432, p < 0.01). - Robustness: OLS estimates are directionally consistent but smaller in magnitude (cases: -92.146; deaths: -5.357), suggesting OLS underestimates policy effects relative to IV. - Model fit: Very high explanatory power in both outcomes (Adj. R^2 ≈ 0.97). - Controls: Election periods are associated with higher cases (~20,780 IV) and deaths (~709 IV), while planting and harvest seasons are associated with lower cases and deaths. A positive time trend term is observed in both models. - Descriptive outcome context: As of 25 April 2022, cumulative totals were 85,747 cases and 2,633 deaths, implying ~97% survival; stringency trended downward over time despite wave-like increases in cases/deaths.

The findings directly address the research question by demonstrating that increases in government policy stringency are associated with significant reductions in COVID-19 confirmed cases and deaths in Malawi, even without a full nationwide lockdown. The IV strategy mitigates reverse causality concerns, lending credibility to a causal interpretation that stricter NPIs reduced transmission and fatalities. Results align with multi-country evidence (e.g., Hale et al., 2021; Dergiades et al., 2022) indicating that non-pharmaceutical policies, particularly when timely and sufficiently strict, curbed the pandemic’s impact. Mechanistically, Malawi’s measures—restrictions on public gatherings and events, school closures, reduced public transport capacity, time-limited markets, and enforcement via potential penalties—likely reduced mobility and contact rates. The positive association between election-period dummy and outcomes underscores the role of mass gatherings. Policy significance is high in LMIC contexts where vaccine access lagged, indicating that coordinated, enforced, and well-communicated NPIs can meaningfully mitigate public health crises.

The study provides the first Malawi-specific empirical evidence that government stringency measures substantially reduced COVID-19 cases and deaths. Leveraging a strong instrument, the analysis indicates that each one-point rise in stringency lowered cases by about 179 and deaths by about 6, with results robust to OLS checks. Despite resource constraints and the absence of a full lockdown, coordinated NPIs—guided by the Presidential Taskforce on COVID-19—were effective. Policymakers should prioritize timely, appropriately strict, and enforceable NPIs during future waves or similar outbreaks, along with sustained public communication and community mobilisation to enhance adherence. Future research should exploit subnational data and decompose the stringency index to identify the most effective policy components and context-specific strategies.

- Aggregation: The main policy variable (stringency index) aggregates multiple measures, obscuring the marginal contributions of individual interventions (e.g., school closures vs. market time restrictions). - Spatial heterogeneity: Analysis at the national level masks district/city differences; enforcement and compliance likely varied between urban and rural areas, limiting tailored policy guidance. - Data granularity: Lack of district-level, intervention-specific, and enforcement/compliance data constrains more nuanced causal inference and targeting. - Future research: Decomposing the index into components and employing subnational analyses would help identify the most effective measures and improve policy prioritization.