Model Uncertainty and Measures of Inequality of Opportunity

We argue that substantive normative disagreements in the literature on equality of opportunity regarding which factors constitute circumstances imply substantial model uncertainty in empirical specifications. We address this issue using model averaging methods applied to linear and regression tree models. We investigate the implications of model uncertainty in estimating measures of absolute and relative inequality of opportunity (IOp) using data from the 2011 wave of the European Union Statistics on Income and Living Conditions (EU-SILC) for 31 countries. We find that failure to account for model uncertainty and nonlinearity risks significantly overstating the degree of inequality of opportunity across the set of countries. We also explored using our IOp measures that are robust to model uncertainty in place of the intergenerational elasticity of earnings (IGE) in the Great Gatsby Curve (GGC). In this case, we find that not accounting for model uncertainty and nonlinearity would lead to an overstatement of the relationship between the degree of inequality of opportunity and income inequality.