European Regional Science Association
|
European Regional Science Association
|
|
|
|
|
The abstract for paper number 279:
Naoru Koizumi, University of Pennsylvania, Philadelphia, PA, USA, Philip McCann, University of Reading, Reading, USA
Housing Decisions in Panama: A Log-Linear Analysis
A log-linear regression was conducted in order to investigate the relationship between the characteristics of households (and household heads) and housing tenure decision in Panama. In contrast to most other papers on this subject which tend categorize housing tenure choices in terms of the decision to own or renter, in our analysis it is necessary to capture the effects of an additional category, namely that of house building. The reason for this is that housing acquisition decisions in Panama are inextricably tied up with the issue of land-plot purchases which are undertaken in order to allow for future individual family house-building. House purchasing is undertaken by both wealthy and poor families, but the mode of purchasing differs substantially according to the plots. The effect of this is that in Panama, approximately 70% of owners are mere plot purchasers who plan to build houses on their own in future, while only about 30% of owners are purchasers of plots which already have some form of existing house structure as part of the plot. In contrast, more than 90% of renters occupy plots with at least partially finished houses constructed on the site. Hence, in terms of the level of development of the dwellings they occupy, renters can be considered to be far more homogeneous as a group than owners. For these reasons we categorize housing-tenure decisions in terms of three categories: renters, buyers and builders, in which the latter two categories are subdivisions of the decision to own. Buyers are defined here as the households who purchase a house with a plot, as opposed to builders who purchase only a plot with a view to subsequently building a house.
Our analysis will first show that a log-linear regression with the traditional two dichotomous dependent categories of owners versus renters, as a function of a range of categorical explanatory variables, is not statistically robust. However, introducing our three alternative dependent categories produces very satisfactory results. Here, we test and compare seven such models, which differ in terms of the interaction terms we employ. The determined of the best model was based on the results of basic statistical tests as well as residual analysis. The results showed that the characteristics of a buyer and a builder in terms of the age of household head, and the number of dependents in a family, are very similar. In contrast, the characteristics owned by a buyer and a builder in terms of education level of household head and household income are seen to be highly dissimilar. Consequently, the analyses revealed that age and number of dependents are key factors in determining the tenure choice of purchasing versus renting. On the other hand, within the group deciding to purchase plots, the levels of education and income are the principal determinants in the tenureship decision.
Unfortunately full paper has not been submitted.