Engel Functions, Panel Data, and Latent Variables

Abstract

A system of consumer expenditure functions is estimated from Norwegian household budget data. Specific features o f our approach are: (i) Panel data on individual households are used, which offer far richer o p p o r t u n i t i e s f o r i d e n t i f i c a t i o n , e s t i m a t i o n a n d t e s t i n g t h a n u s u a l c r o s s s e c t i o n data. (ii) Measurement errors are carefully modelled. Total consumption expenditure is modelled as a latent variable, purchase expenditures on different goods and two income measures are used as indicators of this basic latent variable. (iii) The distribution of latent total expenditure across households, and its evolution over time, is estimated and important properties tested. (iv) Individual differences in preferences, represented by individual, time invariant latent variables, are modelled, identified, estimated, and tested. (v) We test the hypothesis that preferences are uncorrelated with total consumption expenditure, which is basic to all cross section estimation of consumer demand functions. (vi) The model can be formalized as a special case of the LISREL model, and the maximum likelihood algorithm of the computer program LISREL VI is applied. 1 Paper presented at the Econometric Society European Meeting, Bologna, August 29-September 2, 1988. This research has been financed in part by NORAS (project no. 214 1 802). Liv Daasvatn has given valuable programming assistance in the preparation of the data.

The environmental effect of car-free housing: A case in Vienna

“…These detailed surveys require that respondents record all their expenditures over a 2-4 week period followed by an extensive interview about larger purchases such as cars and holidays. Records from a single household are not seen as representative for that household, but only for the period that was recorded, so that artificial households are assembled from different records covering an entire year (Malinvaud, 1980;Aasness et al, 1993). Since the data quality of the consumption data in the inputoutput analysis is better than that in the CES (Kronsteiner-Mann and Schachl, 2006), that data was preferred for phase 1 and for the Austrian average, unless the detail of the CES was required.…”

Identification and estimation of nonparametric panel data regressions with measurement error

“…Other examples are the estimation of Engel functions from household panel data (Aas-ness, Biørn, and Skjerpen (1993)), the estimation of production functions (Olley and Pakes (1996)), studies of returns to research and development performed in private firms (Hall, Jaffe, and Trajtenberg (2005)), and analyzing the technology of skill formation (Cunha, Heckman, and Schennach (2010)). Since the first draft, the identification argument of this paper has also been used and extended for the study of nonlinearities in the dynamics of household income (Arellano (2014); Arellano, Blundell, and Bonhomme (2014)).…”

Heterogeneity in consumer demands and the income effect: evidence from panel data

“…They find that it is important to control for correlated heterogeneity ("fixed effects") and doing that they find evidence of habit formation in all three goods. However, they do not consider a full demand system, but only the three goods, Food at home, Transport and Services 1 . Labeaga and Puig (2002) also use the ECPF and find evidence of correlated heterogeneity.…”