ABSTRACT
Household income has been identified as one of the major determinants of demand for household goods. In addition, other household characteristics, such as household size and composition are also found to be important factors that influence household consumption decisions. This study, using four waves (2006/07, 2009/10, 2012/13 and 2016) of Sri Lankan Household Income and Expenditure Survey data, estimates three different specifications (namely, household expenditure, per-capita expenditure and expenditure per equivalent adult) of a complete system of Box-Cox Engel curves to incorporate household size and compositional differences into the model specification. A comparison of elasticity estimates across the three specifications indicates that amongst the three, the best performing model is the one utilizing household expenditure. An intertemporal analysis of expenditure elasticities indicates that although the magnitude of expenditure elasticities has changed, the necessity or luxury classification of household commodities has mostly remained unchanged for the period 2006 − 2016 in Sri Lanka.
Acknowledgments
The authors wish to thank the anonymous reviewers and the editor of this journal for their constructive comments. They also would like to thank the Department of Census and Statistics, Sri Lanka, for providing access to Household Income and Expenditure Survey Data and the Griffith Asia Institute for providing the research grant to undertake this research project. They also acknowledge the comments received from Professor John O’Brien and other participants at the 14th International Conference of WEAI in Newcastle, Australia, 2018, on an earlier version of the paper.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Expenditure elasticity given by (1) is .
2 Expenditure elasticity given by (2) is .
3 The derivation of Equations (14) and (15) is similar to that of the Equation (11), but with PCE and EPEA, instead of M.
4 Guillerm, M. (2017). ‘Pseudo-panel methods and an example of application to Household Wealth data.’ Economics and Statistics: 109–130 suggests that for pseudo panel estimates, fixed effects model is more appropriate than random effects models.