|
Imputing Consumption from the
Consumer Expenditure Survey to the Panel Study of Income Dynamics
|
I present in some detail the data work and estimation techniques
used by Blundell, Pistaferri, and Preston (BPP) to impute total
consumption to households in the PSID. The BPP procedure requires
that food consumption be monotone in total consumption; I present
the results of a non-parametric test of this hypothesis. In particular
I estimate a PLR model using the higher-order difference technique
of Yatchew. I find a small region of non-monotonicity at very high
consumption expenditures.
|
Apr. 2002
|
