Forecasting Portuguese Population: the special and spatial contribution of mortality and migration

There are no sub national probabilistic population forecasts for Portugal. We aim, in this project, to establish population forecasts based on what-if scenarios. The establishment of those scenarios (as well as the variability of forecasts) will rely on an expert-judgement approach. We assume that the population forecasts should be developed, primarily, according to the cohort-component model. So, to provide the expert team the information for the argued-specification of the underlying assumptions used in the scenarios methodology, we should be able to inform about the type of age-specific trend and the main determinants of each component change. The methodology places greater emphasis on expert argument rather than only on expert opinion. We will analyse the recent past changes on Portuguese population mortality and migration using two different approaches. Through the first one (a traditional demographic approach), each component of the demographic change (mortality, internal migration, emigration and immigration) will be separately analysed, aiming the definition of patterns of behaviour, by age and sex. The second approach is an explanatory one through which we intend to identify the main determinants of demographic behaviour and to define an econometric model that describes trends, taking into account the effect of changes on the predictors. These two types of analysis will be both achieved for each component in a micro perspective (based on individual data) and complemented by the study of the links between demographic trends and macroeconomic indicators (using aggregated data even at the regional level). For the identification of internal migration schedules, micro data from 1991 and 2001 census will be used, applying new indirect methods of estimation, in the framework of a multistate and multiregional approach. At the explanatory level, spatial econometric methodology will be applied, which will enable the treatment of geographical location (origin-destination stream), being obviously important, as an explanatory factor to be taken into account when studying these phenomena. Concerning the immigration, a multistate approach will be applied, using the individual data collected by the “Serviço de Estrangeiros e Fronteiras” from each process of legalization. The analysis of emigration will rely on the data available from the annual national survey to the “Movimentos Migratórios de Saída”; lastly, the analysis on the returned emigrants will also be based on micro data from census. The analysis of mortality will follow the traditional demographic approach and will be based on official data available at the macro level (to assess life expectation trend and to estimate the pattern of mortality, by age and sex). The estimation of the most relevant determinants of mortality will rely on individual data (death declaration). The analysis of the Portuguese demographic situation will be done, in a first stage, using disaggregation and decomposition principles, and using disaggregated spatial data at the nuts IV level (municipality level). In a second stage, a second level of aggregation will be done (larger spatial units) by re-defining geographies of population changes, which are based on homogeneity from the point of view of the determinant variables. The identification of those spatial units will utilize multivariate analysis. The aggregation at the administrative unit level (nuts II level) will also be simulated. The obtained results will allow the definition of new forecasting regions. The final choice of the geography for population forecasts will be based on expert opinions, although being aware that the use of administrative units as projections regions is more relevant for institutional purposes. The final results will reveal what will be the dimension and the structure of the future population, conditional to different patterns of fertility, mortality, internal and international migration. We will be able to derive individual, deterministic projections that can be used within scenarios that describe future socio-economic developments. The project will establish, for the first time, probabilistic forecasts of Portuguese population for the next 50 years, through an innovative adaptation and development of the multistate projection technique.