The effects of changes in household demographics and employment on consumer demand patterns

Abstract

This study examines to what extent changes in consumer demand patterns over the last two decades in the Netherlands can be attributed to changes in household demographics, employment and total expenditures. The dominating changes in consumer demand are decreasing budget shares of food & beverages and clothing & footwear and increasing budget shares of housing and services. The changes in households’ composition – away from the traditional one-earner family with children – together with the increase in household total expenditures account for about one-third of the decrease in the budget share of food & beverages, half of the increase in the budget shares of services and only a minor part of the increase in housing. Once controlled for budget effects, the quadrupling of the proportion of employed women with young children accounts for about one-third of the increase in the budget shares of personal & health care – including childcare – and food away, holidays & entertainment.

Acknowledgements

This study is part of the EU project ‘Demand Patterns and Employment Growth in Europe and the United States’ and its financial support is gratefully acknowledged (no. HPSE-CT2001-00089). We thank the anonymous referee and the editor Mark Taylor, all seminar participants at the Amsterdam and Seville Project meetings and Rob Alessie for very useful comments and discussions.

Notes

¹ See e.g. the seminal work on the service economy of Fuchs (1968) and Schettkat and Yocarini (2004) for an overview.

² Kalwij and Machin (2005) is a forthcoming book chapter and is output of an EU funded research project of a collaboration of researchers from six countries where the main objective was to make a cross-country comparison of household expenditures patterns by making use of household level budget surveys. Parks and Barten (1973) and Selvanathan and Selvanathan (1993) examine cross-country differences in consumption patterns using aggregate data.

³ Results of a first attempt to use this approach can be found in Kalwij and Machin (2005). These first results were conditioned by cross-country comparability issues. We extend on these first results in several directions using the Dutch data. To mention a few, housing consumption is included, female employment when having young children is explicitly examined, and a quadratic Engel curve is used, hereby explicitly allowing for the endogeneity of total expenditures, household employment variables and homeownership.

⁴ Unit nonresponse is over 70%. This requires a weighting of the final sample to derive a representative picture of household expenditures of the Dutch population (CBS, 1992).

⁵ See Kalwij and Salverda (2004) for a detailed description of how the Statistics Netherlands imputes rents.

⁶ Basic health premiums are mandatory and especially in case of public health insurance largely paid for by the employer. The privately insured often get the basic premium reimbursed by their employer. See Kalwij and Salverda (2004) for more details.

⁷ Interest payments, which are presumably increasing with income, are not taken into account.

⁸ Note that Kalwij and Machin (2005) exclude housing from the final analysis.

⁹Kalwij and Salverda (2004) presents a detailed (and favourable) comparison of our budget data with National Accounts data.

¹⁰ We wish to emphasize that we include durables and zero-expenditures in the analysis to provide a complete picture of consumer demand changes in the Netherlands.

¹¹ The absence of information on wage rates and hours of work prevent us from modelling leisure or labour supply as a commodity (see, e.g. Folkertsma, 1995).

¹² Note that prices are the same for all households, hence we need time variation for the identification of price effects.

¹³ Pooled estimation results. Estimating this equation by year does not alter the main conclusions of this study.

¹⁴ Furthermore, at this high level of aggregation there are very few zero expenditures reported.