Applying entropy econometrics to estimate data at a disaggregated spatial scale

A relatively frequent problem when cross-classified data is needed (for example region <InlineEquation ID="IEq1"> <EquationSource Format="TEX">$$\times $$</EquationSource> <EquationSource Format="MATHML"> <math xmlns:xlink="http://www.w3.org/1999/xlink"> <mo>×</mo> </math> </EquationSource> </InlineEquation> industry) is that only aggregate (not cross-classified) data exists. Filling the gaps by combining data from diverse sources usually requires data conciliation. Ecological inference and entropy estimation techniques can be useful tools for this type of problem. This paper tests an estimation procedure based on entropy econometrics to recover disaggregated information from more aggregated data. We use U.S. Bureau of Economic Analysis data to estimate the 2011 personal income of local areas grouped into labor market size-classes in each State. The estimation is performed blindfolded, using only the distribution of personal income across States and, for the United States as a whole, across labor market size-classes, with local employment data as an a priori proxy indicator. Official local area personal income is aggregated into labor market size-classes for each State and used as a benchmark to compare them with the estimates. The results suggest that this technique could be an efficient way of estimating information at local level when different databases of aggregated information could be combined. Copyright Springer-Verlag Berlin Heidelberg 2014