We use South African survey data to study whether inflation forecasts are unbiased. Depending on a fore- caster’s information set, we evaluate whether forecasts are biased due to forecaster herding. Forecaster herding is strong when a forecaster’s information set contains no information on the contemporaneous forecasts of others. When we randomly allocate forecasters into a group of early forecasters who can only observe the past forecasts of others and late forecasters who can also observe the contemporaneous forecasters of their predecessors, evidence of forecaster herding weakens. Evidence of forecaster (anti-) herding is strong and significant in times of high (low) inflation volatility.