The migration component of population projections for the Maltese Islands : a study of the demographic aspects

Abstract

This thesis focuses on demographic aspects of migration analysed for purposes of population projections of the Maltese Islands. International migration has not been introduced in the computation of Maltese national population projections so far, nor has internal migration been included in the computation of population projections at the sub-national level. A main question that the thesis offers an answer to is: How does migration influence the size and age-gender structure of the Maltese population on the national and sub-national level in the context of population projections making? This thesis presents the first attempt at statistical modelling, based on the Rogers and Castro double exponential equation migration model, for population projections in Maltese demography. The data are derived from the 1995 Census of Population and Housing. Two questions on internal migration are asked: (a) place of usual residence one year prior to the Census and (b) place of usual residence five years prior to the Census. These questions provided a basis for analysis of parity of migration, computation of respective gross migraproduction rates as well as some 28 'Key reports' on statistical modelling. Seven model parameters are estimated by means of non-linear regression analysis. The issue of paucity of migration data remains a chronic problem in demographic analysis. The thesis examines the qualities and defaults of migration data sources such as census, survey and population register. The latter one has been seen as a useful tool in collecting data needed for multiregional and multi-state demographic analysis. A more in-depth, causal analysis of migration leading towards life-event analysis would be possible using population register. Like other works of similar character this thesis seeks to achieve statistical significance of estimated parameters in a context of highly detailed demographic data. The 'Key reports' contain the important inputs needed for the Rogers and Castro migration model equations for: (a) in- and outmigration flows; (b) two genders and the total population; (c) usual place of residence one and five years prior to Census; and (d) six regions as well as the total population of the Maltese Islands. It has been demonstrated that internal migration needs to be introduced in the Maltese population projection practice since it directly contributes to processes of: (a) depopulation (the Inner Harbour Region), (b) aging (Gozo and Comino and the Inner Harbour Region) and (c) attractiveness and population increase (the Northern and the Western Regions). Another important finding is that the gross inter-regional migraproduction rates stood at 1.31 and 1.45 for migration five years prior to Census and one year prior to Census, respectively. The mean age of labour force component indicates a family of 'late peaking' migration curves in almost all analysed regions. The thesis confirms the applicability of the Rogers and Castro migration model using the Maltese migration data. The index of goodness-of-fit of 10.2% for modelled migration rates five years prior to Census indicates better results of modelling than the corresponding index of goodness-of-fit for modelled migration rates one year prior to Census of 16.4%. However, models being a simplification of a more complex social reality do not offer a panacea for all problems of forecasting. The exogenous variables other than the classical population system variables (namely births, deaths and migration) are becoming more and more important in contemporary demography. This makes the task of producing population projections necessarily an exercise of multidisciplinary character.