Stefan Humer and Severin Rapp (2020) The Cost of Children - Methods & Bandwidths

How much resources do families with children need relative to childless households? Which methods are suitable to measuring such costs of children? These questions are highly relevant, not at least when it comes to calibrating the generosity of public transfers, such as means-tested minimum income schemes. This paper reviews the vast literature on the cost of children and equivalence scales, highlighting the blind spots of scholarship and summarising the empirical estimates of child-costs in Austria. We begin with an assessment of a number of methods applied to compute the costs of children, showing that no single approach outperforms the others in all dimensions. In view of their theoretical foundations, the most popular approaches in Austria are subject to criticism in the international literature.

Given the broad variety of methods, we argue that the purpose of the analysis is crucial to the choice of the methodological approach. The costs of children computed for assisting prospective families with their financial planning will be an ill-suited evidence-base for the design minimum income schemes. The meta-analysis of the costs of children in Austria suggests that the needs of the first child range between one and two thirds of an adult’s needs. At the same time, the evidence for the existence of economies of scale is ambiguous. Constructing bandwidths for child costs shades the diversity of the results, shaped by different methodological approaches. Deriving costs in currency units crucially depends on the assumptions made regarding the reference value of the needs of childless households.  

Looking forward, our results point out that future attempts to compute the costs of children should pay particular attention to capturing the living conditions of single households, multigenerational households and patchwork families. At the same time, research needs to address the heterogeneous needs of households along the income distribution. Lastly, scholarship should consider how to combine approaches and synthesise results appropriately to produce more accessible and user-friendly statistics.