Imputing rent in consumption measures, with an application to consumption poverty in Canada, 1997-2009

We consider two econometric problems in the measurement of poverty, both relating to rent imputation. First, we account for quality differences correlated with selection into owneroccupied versus rental tenure. This correction increases estimated household consumption by 5% over uncorrected estimates and decreases estimated poverty rates quite dramatically. Second, we propose that measurement error induced by the imputation be corrected by imputing a consumption distribution, rather than a consumption level, for each household. This correction increases estimated poverty rates slightly. We use our methods to measure consumption poverty in Canada, and find that the imputation strategy used influences the patterns observed. For example, measured poverty among the elderly barely declines when one uses our methods, in contrast to the almost 6 percentage point reduction we find using traditional methods. In our assessment of the overtime evolution of consumption poverty, we find that substantial progress has been made on overall poverty and on child poverty, but that poverty among the elderly hardly changed.