A VARYING-COEFFICIENT APPROACH TO ESTIMATION AND EXTRAPOLATION OF HOUSEHOLD SIZE

Harry Haupt, Walter Oberhofer and Thomas Reichsthaler

Mathematical Population Studies, 2003, vol. 10, issue 4, 249-273

Abstract: Household formation analysis is both a multidimensional economical and statistical problem of great complexity. Since most of the literature tries to incorporate multiple economic aspects, there is, considering the extraordinary practical relevance of the problem, a remarkable gap between theory and application in this field. This paper tries to diminish this gap by a comprehensive treatise on the statistical site of the problem. Thus, we develop a model of household composition, where the evolution of the household membership rates is captured by a logit link-function and a multinomial distribution, which automatically fulfills the non-negativity and adding-up restrictions of the underlying probabilities. We use a varying-coefficients procedure by polynomially smoothing the household membership rates over age for every household size class and assuming a linear predictor in other variables. As an application we estimated and extrapolated the distribution of household sizes of an autonomous region using population register data. Our sample consisted of approximately 450,000 people living in about 170,000 households, grouped into nine different household size classes and classified into age classes from 0 to 90. The data covers a time span of 12 years, from 1986 to 1997. Empirical results show the robustness of the procedure even in case of low cell frequencies. Thus, there is no need for regional or age-group aggregations.

Keywords: Generalised linear model; multinomial logit model; nonparametric estimation; varying-coefficient model; local likelihood; household membership rate (search for similar items in EconPapers)
Date: 2003
References: Add references at CitEc
Citations: View citations in EconPapers (3)

Downloads: (external link)
http://www.tandfonline.com/doi/abs/10.1080/713644740 (text/html)
Access to full text is restricted to subscribers.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:taf:mpopst:v:10:y:2003:i:4:p:249-273

Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/GMPS20

DOI: 10.1080/713644740

Access Statistics for this article

Mathematical Population Studies is currently edited by Prof. Noel Bonneuil, Annick Lesne, Tomasz Zadlo, Malay Ghosh and Ezio Venturino

More articles in Mathematical Population Studies from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().