Income Sorting: Measurement and Decomposition

Thomas Davidoff
2004 Social Science Research Network  
This paper addresses the measurement of income sorting and the attribution of observed sorting to different causes. In terms of measurement, I show that a standard decomposition of variance of household income into within jurisdiction and between jurisdiction components understates sorting in the presence of measurement error. Using 1990 US Census data, I find that adjusting for this error approximately doubles the estimated extent of sorting. On average, across all US metropolitan areas (MSAs)
more » ... olitan areas (MSAs) I find that approximately ten percent of the variation in household income can be explained by differences across jurisdictions. I attempt further to identify the extent to which the observed sorting may be attributed to a "Tiebout" mechanism, by which income sorting follows from sorting by preferences over governance into differently governed jurisdictions. I find that zip codes are considerably more homogeneous than jurisdictions, and that on average, neighboring zip codes in different jurisdictions are only slightly more different from each other than neighboring zip codes in the same jurisdiction. This result implies that we cannot safely assume that observed sorting on characteristics is driven by differences in government, and also implies that extra-governmental neighborhood characteristics are an important source of sorting.
doi:10.2139/ssrn.555997 fatcat:lv3vjscdsvd5lhtrkve5dszrla