Fig. 3

Author’s own calculations. Source: QLFS 2020Q1 and 2020Q2 (Statistics South Africa 2020b, c). Notes: This figure presents a coefficient plot of estimates of from specification (1) by dependent variable and lockdown level using varying industry capacity assumptions. ‘Very progressive’ = workers coded as being permitted to work if any share of the industry is permitted; ‘progressive’ = workers coded as being permitted to work if at least 25% of the industry is permitted; ‘conservative’ = workers coded as being permitted to work if at least 75% of the industry is permitted; ‘very conservative’ = workers coded as being permitted to work only if 100% of the industry is permitted. Markers represent point estimates and lines represent 95% confidence intervals. Lockdown levels range from 5 (most stringent) to 3 (most lenient). Sample restricted to the working-age population (15–64 years) as of 2020Q1. All model standard errors are clustered at the panel level. Estimated weighted using sampling weights. All models control for a vector of time-varying observable covariates including age, highest education level, and employment type. All models additionally control for individual fixed effects (FEs) and as such time-invariant observables are not included. ***p < 0.01, **p < 0.05, *p < 0.10