ACS Data Users Group

It has long bothered me that small geographies have to rely on 5-year sample estimates, but there is observable structural bias in those estimates. Namely, previous year observations are inflated using CPI growth, which differs nationally (much less regionally) from rent and income growth.

Does anyone have a well-tested methodology for using public data to improve the accuracy of those smaller geography 5-year estimates?

My easy button solution would be to simply use the regional differences in 1-year vs. 5-year estimates to adjust the 5-year estimates for the geographies within that region, but I'm curious about more sophisticated strategies that others have already worked out. 

Jonathan Kurzfeld, Ph.D.

Director of Planning & Research

MaineHousing

------------------------------
Jonathan Kurzfeld
Director of Planning and Research
MaineHousing
------------------------------

The Census SAIPE (Small Area Income and Poverty Estimates) may have some additional estimates. This is the data that supports the free and reduced price lunches, and the longest running small area program (since 1970s). The income data are produced annually. The rent won't be there of course; you would have to build your own SAEs.

------------------------------
Stas Kolenikov
NORC
------------------------------