Session 7: Population projections for national populations • 7.1. Spectrum • 7.2. Evaluation of projection results • 7.3. Accounting for uncertainty – choosing alternative projection scenarios • 7.4. Lab time 7.1.2. Using Spectrum SPECTRUM is a suite of easy to use policy models which provide policymakers with an analytical tool to support the decision making process. SPECTRUM consists of several software models including: •DemProj: Demography •FamPlan: Family Planning •LiST: Lives Saved Tool (Child Survival) •AIM: AIDS Impact Model •Goals: Cost and impact of HIV Intervention •Resource Needs Module: Costs of implementing an HIV/AIDS program •RAPID: Resources for the Awareness of Population Impacts on Development •Safe Motherhood Model •Allocate 7.1.2. Using Spectrum SPECTRUM is at version 4.51. As it under continuing development, one should check for updates online: http://www.futuresinstitute.org/spectrum.aspx 7.1.2. Using Spectrum Spectrum 7.1.2. Using Spectrum • Advantages: – Reliable and well tested – Appealing user-interface – User base is large, but concentrated among health and policy professionals (UNAIDS) – Support, on-site Training available • Disadvantages – Complex package due to integration into a variety of other tools (AIDS, POLICY etc.) – Obtaining results can be cumbersome 7.1.2. Using Spectrum • Steps: • • • • 1. 2. 3. 4. Projection parameter settings Data input Executing the projection Obtaining, saving the results Hands-on exercise • Preparing a cohort-component projection with Spectrum Hands-on exercise • Executing the sample projection and examining the results • Preparing a new projection • Adding data • Obtaining results from Excel Using Spectrum 1 Sample.pjn Using Spectrum 2 Using Spectrum 3 Projection menu group Using Spectrum 4 Set last year Uncheck AIDS Using Spectrum 5 Using Spectrum 6 Using Spectrum 7 Using Spectrum 8 Using Spectrum 9 Using Spectrum 10 Using Spectrum 11 Using Spectrum 12 Using Spectrum13 Using Spectrum 14 Trouble shooting Spectrum • Spectrum stores results in files with the extension *.dp • By renaming the file to *.csv, it can directly be opened by Excel 7.2. Evaluation of projection results Projections are about the future, and therefore are invariably incorrect. It would not be useful to evaluate them in terms of their correctness (accuracy). Instead, one should assess them differently: 1. Sound Methodology – 2. Internal Consistency – – 3. The projection makes sense relative to the characteristics and assumptions of the methods The projection makes sense compared to historical demographic patterns The projection makes sense compared to local socioeconomic/ cultural characteristics External Consistency – The projection makes sense compared to contextual and/or comparison areas 7.3. Accounting for uncertainty -Choosing alternative projections scenarios • • • • • Reasons for uncertainty How to account for uncertainty? Construct scenarios Project past variability Expert opinions Reasons for uncertainty • Open future – several options for future trends • Input data • Assumptions How to account for uncertainty? • Construct scenarios • Probabilistic projections: – Project past variability – Expert opinions Scenarios • Common scenarios • • • • Constant fertility Constant Mortality No migration Instant replacement Hands-on exercise: • Preparing and comparing different projection variants Hands-on exercise: Scenarios • Spectrum: Comparison Projections • Start by creating a Projection, using data in Spectrum. • You may modify the data as you wish: Set fertility to constant, for example. Name it accordingly: BeninConst. • Or simply use the medium variant available in Spectrum as a starting point. Name it BeninMedium. • Select an appropriate base year (2010?). • Select an appropriate last year (2050?) • Inspect the settings. Save the projection • Next add more scenarios. Hands-on exercise: Scenarios • Add another scenarios. • You could just create another projection input file, apply your scenario setting and save it with an appropriate name. • There is a shortcut: Open the file you just saved again. Spectrum gives you a choice. Click on Load and rename. Hands-on exercise: Scenarios • We want the second scenario to keep fertility levels constant at base level. Name the second scenario BeninConstant. • Spectrum has now two projections loaded: BeninMedium and BeninConstant. BeninConstant is right now only a copy of BeninMedium, so we have to make changes to the fertility settings in BeninConstant. • Make sure that BeninConstant is set to be the active projection. Hands-on exercise: Scenarios • In order to keep the fertility constant, you can either copy the base fertility to the projection years, or use Spectrum’s copy and duplicate functions: • Highlight the fertility and click Duplicate. • Check if the fertility has been changed: Go to results and look at the fertility chart. • Note that there are now two fertility trends, one named BeninMedium, and one named BeninConstant. Hands-on exercise: Scenarios • Now you may add even more scenarios. • As the first scenario (BeninMedium) is the reference scenario, set BeninMedium to be the active projection. • Now re-load BeninMedium, and chose Load and rename. • Rename it to BeninInstant. Hands-on exercise: Scenarios • What do the scenarios reveal about the demographic future of the country chosen? • Discuss the results. 7.4.Lab time Other software • RUPEX US Census Bureau Rural-Urban Projection Program with EXCEL interface • United Nations MORTPAK projection program PROJCT • LIPRO multi-state projection program (LIfestyle PROjections) USCB RUPEX Description The program uses the cohort component method for projecting the population by age and sex. The projection is made following cohorts by single years of age, even if all or some of the input data are in five year age groups. Using interpolated values of central death rates from life tables, RUP estimates the number of deaths and subtracts them from each cohort. It also takes into account internal and international migration. Every year births are estimated using fertility rates and the female population in reproductive ages. The annual number of births constitute new population cohorts that are followed through in the future. USCB RUPEX Steps using RUPEX: – 1. Data input – 2. Projection parameter settings – 3. Executing the projection – 4. Obtaining, saving the results USCB RUPEX The program provides output of a wide variety of demographic measures for any specified year of the projection period. These outputs include: a) b) c) d) e) f) Population by sex and age (single years, 5-year age groups, special groups) and summary measures of age (e.g., percentages, sex ratios, median ages, dependency ratios). Summary vital rates (e.g., crude rates, life expectancy, infant mortality rates, and total fertility rates). Life tables. Net numbers of migrants or migration rates by age and sex. Number of deaths, by age and sex. Number of births, by age of mother, and age-specific fertility rates. USCB RUPEX – Advantages: • Reliable and well tested methodology • Large user base • Professional support • On-site Training available • Integration with analysis tools – Disadvantages • User interface somewhat non-intuitive (Excel based) • Data format of input file text based and not self explanatory • Still in testing for Windows Vista/Win7 United Nations MORTPAK Population Projection LIPRO • Multistate projection program developed by NIDI, Netherlands • Although originally developed for household projections, the LIPRO computer program can be used for a wide range of calculations in multistate demography. LIPRO has been extensively used for various applications, in the Netherlands as well as in many other countries. LIPRO • Consistency algorithms exist and can be expanded. • Powerful, but demanding. • No further development • Runs only under Windows XP LIPRO LIPRO • Projection data must be prepared outside: age-specific rates of change for each region/unit. Their evolution over time can be controlled through a scenario module. Alternatively, the number of events by age and sex can be inputted; LIPRO will then calculate the rates/probabilities. Challenge: Age-specific migration rate matrix: Origin and destination, by age. LIPRO • • • • • Belgium - LIPRO household typology China - population projections Shanghai Marital status projections for England & Wales. Spain - population projections Catalonia Paper on: Internal migration scenarios and regional population projections for the European Union. International Journal of Population Geography • http://www.nidi.knaw.nl/Pages/NID/24/841.bGFuZz1VSw.html Thank you