J. Conlisk, R. Ramanathan; Expedient Choice of Transforms in Phase-Diagramming1, The Review of Economic Studies, Volume 37, Issue 3, 1 July 1970, Pages 441

A common method of estimating permanent income is to classify households into groups and then take the average income as an estimate. This has two weaknesses: (1) no account is taken of expected future receipts, and (2) households in the same group may have different permanent incomes because of differences in factors other than those used in forming the groups. The present study suggests a method of correcting for these defects by using a two-stage procedure. Empirical results indicate that irrespective of the choice of variables used to group households, the method suggested yields better results than the common method of using the mean values.