Household projections by the headship rates method: The case of Serbia

Abstract

The headship rates method (HRM) of household projections based on the share of household heads in the total population of the same demographic characteristics (age, sex, nationality, marital status, etc.) is the most commonly used method, especially by statistical institutes and planning institutions. The specific rates of household heads by age are calculated by dividing the number of household hold-ers of a certain age with the total number of residents of the appropriate age. The future number of households is then simply projected on the basis of population projections by age and assumptions about the future changes of HR. The HRM is based on the projection of the future age structure of the population. In that sense, the choice of methods of population projection, as well as the meth-od of projecting HR-s have determining impact on the outcome of household projections. Given the methodological inconsistency typical for official popula-tion projections in Serbia and significant differences in addressing uncertainty of the future population change between deterministic and probabilistic approach in making population projections, the decision to use a probabilistic projection of the population of Serbia as the basis for calculating the future number of house-holds and their structure according to the age of the household head proved to be a logical choice. However, as the basic aim of this article is to show the simple method of household projections, the above-mentioned stochastic projection is used in utterly deterministic manner. The median of the prediction interval of the population distributed across age is interpreted as the most probable future, or as a prognosis. The HR-s based on the age structure estimates and estimated number of households by age of the household head from Household budget survey (HBS) are used for the purpose of HR projecting so that the number of observa-tions would be large enough for calculating inclination parameters. The obtained rates show a tendency to decline during the observed period, however, in certain age categories, the rates are expressed by extreme values that are certainly the result of random sampling in the HBS for the purpose of analyzing consumption rather than analyzing the demographic characteristics of households, and must be taken with a certain reserve. Although the tendency of declining rates in most age categories is not unexpected, surely the intensity of decline is unexpected. For this reason, in the formation of the regression function, the extreme values of the rates are intentionally excluded in the following way: after calculating the regres-sion line parameters, all the values of the rates that deviate from the regression values by more than 20 per cent are rejected, after which the regression parame-ters are recalculated. On the basis of the second calculation of the regression line, parameters are obtained. However, as the obtained parameters led to unexpected-ly large HR changes according to the age of the household head until the end of the projection period (2040), it was assumed that the inclination parameter (b) would be reduced by 10 per cent annually compared to the start year of the re-gression line. On the basis of the rates according to the 2011 census data and the hypothesis on the slowdown of the observed trends in the future, future HR-s are calculated. Furthermore, based on the projected HR-s by age and future age struc-ture of the population, the number of households by the age of the household head for the projection years is calculated. Based on the results of the projection, the total number of households will be reduced on average by over 11 thousand households per year. Also, compared to the 2011 census, it can be expected that the number of households in all age groups will be reduced by the end of the projection period, except in the category of household heads aged 65 and over that stabilizes to around 900 thousand households by the end of the projection period. Due to the decline in the number of households, the average household size will be reduced by 0.18 members in 2040 compared to 2011, from 2.89 to 2.71. The largest number of households in Serbia are family households, the share of single person households in the population under the age of 50 is small, and the structural barriers to the establishment of an indigenous household in persons under the age of 30 are significant. All of this makes it difficult to withdraw par-allels with other European populations in terms of a possible path that the popula-tion and households in Serbia should follow in the projection period. Some of the projections of households produced by the HRM of a newer date for populations also found in the post-transition demographic stage show that the age at which the household is based, the mechanisms that affect the generation, change, and extin-guishing of the household, which are characteristic for each society, result in significantly different values of age-specific HR-s. Of course, HR-s by age vary considerably among different populations. It is obvious that the key differences in Serbia in relation to other countries occur precisely at the age when individuals base their own household. The existence of postponing marriages and parenting that is recognized as key life-changing milestones in the transition to adulthood and the founding of one's own household, the chronic lack of systematic housing policy towards young people and high youth unemployment are the main causes of the late establishment of their own household and the maintenance of low HR-s for persons under 30 years of age in Serbia. Nevertheless, during the first dec-ade of the 21st century, there is a certain shift in the financial independence of young people, which gives some hope that in the future HR-s in the category between the ages of 30 and 39 can be slightly increased, which is confirmed on the basis of the sample of households from the HBS for the period 2006-2013. Namely, the tendency of a slight increase in the value of the rate for persons aged between 30 and 39 years is certainly the result of an increase in the age at which the household is based, which can be noticed on the basis of the reduction in rates for persons under the age of 30. On the other hand, a certain decline in the value of the rate characteristic for the households of the holders in their middle age (between 40 and 64 years of age) has an explanation in the increase in number and share of multi-family households in the period 1991-2011, especially in urban areas. During the 1990s, in the conditions of a deep socio-economic crisis, with the continuation in the next decade during the transition of the economic system, in conditions of significant poverty and the phenomenon of the re-traditionalization of partnership arrangements within multi-family households, it is obvious that a significant number of families in the middle of their life cycle lived in within parental households whose carriers are aged 65 and over. In fact, as the increase in the HR-s during the thirtieth year of age is the result of depriva-tion of rates in younger persons, this is, by and large, a rise in rates for persons aged 65 and over due to a reduction in rates among carriers aged between 40 and 64 years. The presented method of household projections is not characterized by methodo-logical sophistication, elegance and precision in reflecting changes in the struc-ture of households according to the family composition and a detailed presenta-tion of changes in the family status of individuals, but it certainly represents an simple way of household projecting according to the age distribution of carriers, the average size and the number of households. It seems that this approach, based on the stability of age-specific rates of household heads, without getting involved in the field of sociology, is quite precise in the medium term, especially given the simplicity in household projecting based on HRM.