Estimating Self-Employment Income-Gaps from Register and Survey Data: Evidence for New Zealand

Abstract

This paper provides estimates of the income-gap of the self-employed – defined as the proportion of undeclared to true income – in New Zealand using traces of expenditure to infer true income holdings following the approach of Pissarides and Weber (1989) and Feldman and Slemrod (2007). This uses the relationship between expenditure and income for the employed (with lower opportunities to evade) to infer the true income of the self-employed.We use a unique dataset – New Zealand’s Integrated Data Infrastructure (IDI) – that matches individual data including incomes and expenditures from the Household Economic Survey with register incomes declared to the tax administration. This has several advantages in our context. Firstly, register data minimises income measurement error by employed and self-employed as it does not rely on accurate recall by survey participants. Secondly, the approach allows us to measure evasion under different incentives for misreporting. We estimate that the self-employed underreport on average around 20% of their income (with a 95% confidence interval around 10-30%), and that the income-gap varies significantly by gender and region. Our results are also found to be highly robust to a range of sensitivity tests for measurement error.