Beschreibung:
This paper examines measurement error in the share of income going to the top 1% (and other subgroups) that comes from estimation of the denominator. We compare two approaches. First, the default ‘external’ approach used in the current literature, which relies on different data sources for the numerator and denominator. Second, an alternative ‘augmented internal’ approach that uses the same data source for the denominator as for the numerator, but augments this to fill gaps in coverage. We set out four principled criteria for selecting between these approaches and argue that the ‘augmented internal’ approach is to be preferred. On this approach, the UK top 1% share is 2 percentage points higher than under the alternative ‘external’ approach.