Academic Scientific JournalsBenefits and limitations of using probabilistic approach to forecast the population of Poland
Keywords:
demographic forecasts, Polish population, stochastic modelsAbstract
A probabilistic approach in demographic forecasting was developed in response to the need for more precise information about the uncertainty of demographic projections. Probabilistic projections methodology is however source of controversies, and most of the official population projections are still deterministic.
To discuss the benefits and limitations of probabilistic projections when applied to predicting number and structure of population of Poland, the results of the work carried out in the Central Statistical Office will be presented. The basic probabilistic projection, based on the CSO projection for 2014 -2050 which I prepared in 2015, gives the opportunity to compare confidence intervals for individual coefficients and population numbers with empirical data from the last three years. This comparison clearly indicates that the probabilistic approach has managed to include the variability of demographic processes within confidence intervals, whereas the results of the deterministic forecast, just three years after its creation, deviate significantly from the real data.
The example of this forecast also demonstrates serious constraints that this type of approach faces when applied to data for Poland. The largest of these is undoubtedly the official data for migration, which is far below understated and marginalizes the importance of this component in population projections. In the case of probabilistic forecasts, low migration flows, coupled with significant fluctuations in them, cause additional complications – with classical approaches to the confidence intervals computation giving negative values. More reliable estimates of migration based on so-called. „mirror statistics” are only available since 2008.
The probabilistic approach can undoubtedly also be used to prepare far more complex forecasts at lower levels than nationwide. It gives, for example, an opportunity to estimate the likelihood of extreme depopulation in particular regions of Poland.
Downloads
References
Alho J. (2002), The Population of Finland in 2050 and Beyond, The Research institute of Finnish Economy, Helsinki.
De Beer J. (2000), Dealing with Uncertainty in Population Forecasting, Statistics Netherlands.
De Beer J., Alders M. (1999), Probabilistic Population and Household Forecast for the Netherlands.
GUS (2014), Prognoza ludności na lata 2014–2050.
Keilman N. (1997), Ex Post Errors in Official Population Forecast in Industrialized Countries, „Journal of National Statistics”, No. 3.
Keilman N., Pham D.G., Hetland A. (2002), Why Population Forecasts Should Be Probabilistic? – Illustrated Case of Norway, „Demographic Research”, Vol. 6, No. 15.
Lee R., Tuljapurkar S. (1994), Stochastic Population Forecasts for the United States: Beyond High, Medium and Low, „Journal of American Statistical Association”, Vol. 89, No. 428.
Lutz, W., Sanderson W., Scherbov S. (1998), Expert-Based Probabilistic Population Projections, „Population and Development Review”, Vol. 24.
Matysiak A., Nowok B. (2007), Stochastic Forecast of the Population of Poland 2005–2050, „Demographic Research”, Vol. 17, No. 11.
ONZ, Department of Economic and Social Affairs, Population Division (2013), World Population Prospects: The 2012 Revision, Highlights and Advance Tables.
Rowam S., Wright E. (2010), Developing Stochastic Population Forecasts for the United Kingdom: Progress Report and Plans for Future Work, Office for National Statistics.
Scherbov S., Mamolo M., Lutz W. (2005), Probabilistic Population Projections for the 27 EU member States Based on Eurostat Assumptions.
