A methodology for household projections: the HPROM model (Household PROjection Model)
This Working Paper presents the methodology the Federal Planning Bureau has currently developed to draw up long-term Belgian household projections. This methodology allows detailed projections of the number of households (at the district level) by household type according to living arrangements and not legal situation. Thus, the projections include the different forms of living arrangements, such as cohabitation, single-parent families, single households ("one person"), etc. They also guarantee coherence with the national population projections, which have been published by the Federal Planning Bureau and Statistics Belgium for several years and are based on the so-called component method.
Since 2008, Statistics Belgium and the Federal Planning Bureau have been publishing annual population projections for Belgium at the district level (NUTS3) and by gender and age. Several federal and regional institutions (including the FPB) use those projections as explanatory variables in numerous short-, medium- and long-term projection models (w.r.t. the economy, long-term health care, energy, transport, etc.) and for specific projects and requests. For several years now, the demand for a demographic projection at the household level has been increasing. Understanding household dynamics is particularly useful for various aspects of social life (identification of, for instance, the growing number of single-parent families - often single women - or single households - often elderly; these are susceptible to poverty problems or a lack of support) and economic life (impact on housing, transport, mobility, consumption, taxes, etc.). In order to meet those demands, a household projection model was developed for Belgium. This Working Paper aims to specify further the methodology used.
This methodology belongs to the group of static household projection models, as opposed to the dynamic models. While the latter study the transition probabilities from one state (i.e. one position in a household) to another by analysing flows, the former focus on the stocks and rates of each state in the studied population. It is, however, not limited to the so-called Household Headship rate method, which determines and projects the number of households by the share of heads of households in the population; instead, it expands that method by also taking into account the rate of membership of a specific position within the household for each member of the household. Therefore, the methodology allows detailed projections (at the district level) by household type and according to living arrangements (de facto situation) and not legal situation. The legal situation represents the administrative situation of each individual as he/she is registered with the Belgian National Register (single, married, divorced, widowed). Given the new forms of living arrangements which have been observed for several years, the legal situation does not always correspond to the actual situation of a household. From a socio- demographic point of view, the de facto situation is often more relevant than the legal situation.
More concretely, the household projection starts from the population projection by age and gender at the NUTS3 level. An individual household membership rate is associated with each group of individuals (by age, gender and NUTS3 level). Individual household membership rates are defined according to their living arrangements. The number of individuals at time t with a given household position (by age and gender at the NUTS3 level) is obtained by multiplying the population at time t (by age and gender at the NUTS3 level) by the corresponding individual household membership rate for the given position (by age and gender at the NUTS3 level ).
The population projection comes from the Belgian population projection published by Statistics Belgium and the Federal Planning Bureau. The future evolution of the rates of membership of a position within the household includes the recent developments regarding the different forms of living together by age and gender of the individuals. However, it is presumed that the different household types will coexist in the long run, whereas the current trends should not continue at the same pace. Hence, a saturation effect is introduced in the long run.