Demographic projections and forecasts abstracts

Strand organiser: Professor Ludi Simpson, University of Manchester 

Demographic projections: Ethnicity and projections – Monday 7 September 1.30pm 

Projecting the regional explicit development of the population structure and social heterogeneity in Nepal
Samir K.C., Markus Speringer, Wittgenstein Center for Demography and Global Human Capital, (IIASA, VID/ÖAW, WU) 

Population projections at national level and smaller administrative units can provide essential information for planning and implementing government policies, including the allocation of budget and resources. Often, such projections are using crude methods and are not based on evidence and argument based assumptions about the future. Consequently, the results, when compared later with actual demographic rates, are further away from reality. Acknowledging the fact that the future is uncertain, we attempted to minimize the level of uncertainty, firstly, by understanding the important forces (behavioral, cultural and socioeconomic factors) that affect future demographic events. Secondly, we constructed a baseline demographic scenario based not only on the continuation of past trends but also on how the other forces develop and will impact the demographic events in the future. In addition, in collaboration with the Ministry of Health and Population, we developed multiple policy (population and health) relevant projection scenarios for more than 4000 administrative units in Nepal for the period 2011 to 2031. We used data, published tables and sampled microdata, from the latest two Census 2001 and 2011 along with several rounds of Demographic and Health survey data. In this paper, we will present our methodological approach of estimating and projecting the fertility, mortality and migration at small spatial level as well as the population projection model. Preliminary results show that, in Nepal, both internal and international migration are and will be the main (direct and indirect) cause of changing population dynamics at different spatial levels. 


Alternative methods to estimate mortality for UK ethnic groups
Pia Wohland, Institute of Health and Society, Faculty of Medicine and Newcastle University Institute for Ageing 

Previously, in the course of developing subnational ethnic population projections for the UK, we produced the first estimates of ethnic mortality for local areas in the UK for the year 2000. These estimates were vital input variables to our model. We found that only few ethnic minority groups could expect to live longer compared to the White British majority group and most ethnic minorities could expect to live shorter lives. However, these early estimates have been challenged by subsequent work by Scott and Timaeus (2013) and Wallace and Kulu (2014). In light of this, we reviewed our methods, taking into consideration literature on immigrant and ethnic group mortality differentials starting in 1984. Merging the available information led to the development of new improved methods to estimate national and subnational mortality for UK ethnic groups in the absence of actual deaths data by ethnicity. This presentation will detail alternative estimation methods of mortality rates by ethnicity, nativity, age and gender, which are also important population health indicators. 


Ethnic population projections for the UK and local areas, 2011-2101: New results for the fourth demographic transition
Philip Rees1,Pia Wohland2, Nik Lomax1, Paul Norman1,  1School of Geography, University of Leeds, 2Institute of Health and Society, Newcastle University, Newcastle 

This paper outlines a revised model for projecting the ethnic group populations of local authorities in the UK and the likely projected populations. We implement the revision in two steps. First, we update estimates and assumptions for our previous ETHPOP model using 2011 Census and associated data and run a new projection for the UK (England LADs with Wales, Scotland and Northern Ireland) as single zones. We compare results with our 2001 based ETHPOP projections. Second, we describe our ongoing work to revised the ethnic projection model to incorporate new features found necessary as a result of an evaluation of ETHPOP performance against 2011 Census outcomes and of consultation with national statistical offices, local governments and other organisations. The NewETHPOP projections will differ from previous ETHPOP projections: they will cover all LADs in the UK; they use 24 ethnicity-nativity groups harmonised across 2001 and 2011 Censuses and the four home countries. Nativity (UK born, Foreign Born) will be used because of its impact on fertility and mortality rates. The conceptual framework is switched from the transition to the movement approach in line with that used by the National Statistical Offices. To estimate ethnicity-nativity specific components of change we will extend the NSO scheme for reconciling population change to the 2001 and 2011 censuses. To inform assumptions for the projections we will use a 14 year time series from 2001 to 2014 and employ time series models along with expert judgement to design assumptions. We will explore the alternative approaches to projecting international migration in model variants. We plan to report on a principal projection that is constrained to the official national and sub-national population projections and on our own alternative projection based on different assumptions. We also propose a set of impact scenario projections, which estimate the influence of component assumptions and each group’s 2011 age structure. Our projections will track the “Diversity Explosion” for the UK as whole (the Third Demographic Transition) and its spatial diffusion across the UK (the Fourth Demographic Transition). 


Ethnic group migration patterns: a UK time series analysis
Nik Lomax, School of Geography, University of Leeds 

Internal migration in the United Kingdom (UK) shapes the size and composition of the population at local authority scale. This paper draws upon a time series of origin-destination migration data which is disaggregated by age, sex and ethnic group to reveal how patterns and propensities have changed between 2001 and 2013 at the sub-national level in the UK. The analysis uses the 2001 and 2011 Censuses as anchors (where comprehensive disaggregated information are available) and develops further a time series proposed by Lomax (2013) which has been used widely for the analysis of patterns of migration by age and sex. Adding the ethnic dimension to these data is important as different groups have distinct age structures, health care needs and socio-economic characteristics. The way in which the ethnic composition of areas is changing due to migration is explored through assessment of levels of segregation or integration, inequality and resource provision. Analysis will assess different directional patterns of ethnic minority groups compared with the whole population using the Migration Efficiency index (MEI), which will be plotted on maps using three classes (higher than all people, about the same as all people, lower than all people). These results will indicate how stable ethnic group populations are across local authority areas over time. The time-series data used in this paper form a key input for a projection system for sub-national ethnic group populations (NewETHPOP) so consideration of future trends and possible scenarios will be offered, which are informed by the historic estimates. 


Demographic projections: National and global scenarios – Tuesday 8 September, 11.00am 

Long run net migration assumptions and their impact on projected populations
Guy Abel, Vienna Institute of Demography, Austrian Academy of Sciences 

Migration is a complex process influenced by multiple demographic, economic and political factors which make any form of long run prediction difficult. In general, little attention is given to migration in global population projection models and yet as fertility and mortality rates fall, its role is becoming more pronounced. In this paper, three simple migration scenarios are developed to provide a range of alternative future populations in the long run of all countries. The alternative migration scenarios are used in a global cohort component projection model to obtain populations up until 2100. Results illustrate the important role of migration assumptions in long run projections, especially in post demographic transition countries, answer some common “what-if”' questions and quantify the impact of expert judgements used in the UN medium migration scenario on projected population levels. 


The dynamics of human capital-specific old-age dependency ratio in Europe Dimiter Philipov, Anne Goujon, Paola Di Giulio,  Wittgenstein Centre for Demography and Global Human Capital (IIASA, VID/ÖAW, WU)   

A rise in human capital can boost economic growth mainly through innovation and increase in competitiveness and hence have the potential to alleviate economic problems related to population ageing. Yet little research is available on the dynamics of this effect and on conditions under which it holds. We have constructed a human capital-specific old age dependency ratio (HC-OADR) where populations in both the numerator and the denominator are distinguished by their level of human capital, measured with earnings and pensions differentiated by education and age. The dynamics of the HC-OADR was first examined using data for Italy (published in Demographic Research in 2014). Multistate population methods were utilized for long-term projections under two scenarios: with constant and with increasing rates of transition to higher education. The HC-OADR under the constant scenario produces a trend of population ageing that is faster than the trend received with the conventional OADR i.e. under specific conditions, a constant or a moderately increasing human capital may cause aggravation of consequences of population ageing rather than their alleviation, which can be achieved under a faster increase in human capital. Based on the dataset of the Wittgenstein Centre population projections (2014) and European Union Statistics on Income and Living Conditions (EU-SILC), we will apply the same methodology to a large set of European countries (about 30 countries) to explore whether our findings hold in different educational, income, and pension settings in a dynamic way. Preliminary results show that they do tend to hold. 


Population ageing and population decline in Europe, 2013-2080: Estimating the sensitivity and elasticity of projection results
Nora Sánchez Gassen and Hal Caswell, Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam 

Member states of the European Union face two demographic challenges: their populations will age rapidly in the coming decades, and in many countries populations will also decline. Recent population projections estimate the size of these trends and quantify the effect of demographic changes themselves, for instance on labour force sizes. These projections depend on fertility, mortality and migration assumptions as well as assumptions on future changes in labour force participation rates that underlie the projection models. Methods to systematically analyse the sensitivity of projection results with respect to changes in these underlying parameters have only recently been introduced. Here we present such sensitivity analyses for 31 European countries, using Europop2013 national population projections published by Eurostat. We formulate the projections in matrix form and calculate how changes in mortality, fertility, and migration parameters influence projected population sizes and age structures, using matrix calculus methods. From these results, we also calculate the sensitivity and elasticity of projected developments in labour force sizes in Europe. These analyses will identify which demographic or behavioural parameters have the largest influence on projection results, and may therefore be an important target of policy intervention. 


Projections and local planning: Tuesday 8 September, 1.45pm 

Labour force projections for Districts of Britain
Ludi Simpson, University of Manchester 

The paper will review methods of projecting the labour force, and assess the fitness for purpose of practical options for making standard projections for all local authorities in Britain. Projections of the labour force and of jobs are required for the statutory development plans drafted by each local authority in Britain, but are not part of the official statistics provided by any of the national statistical agencies. Projections for the UK by ONS were discontinued after the 2004-based round. The paper will review the published accounts of past and current methods and strategies for projecting the labour force. Current projections include the Office for Budget Responsibility (for the UK), the University of Warwick’s Working Futures project on productivity (for Wales, Scotland, N Ireland and regions of England), and local authorities and planning agencies’ projections for specific local areas. The paper will draw on this review to develop a series of options for a strategy and method appropriate for all local authority areas. Criteria for fitness for purpose of a method, which must itself be available for scrutiny, include: - Practical: using data available for all districts of England, Wales and Scotland. - Suitable for sensitivity tests and alternative scenarios: a model must allow straightforward alteration of key assumptions to inform the development of local development plans. - Efficient: can be developed and maintained within available resources. The research has been commissioned to provide data for labour force projections as a Data Module in the demographic software POPGROUP. 


Household projections and planning: geographical variations in England
Greg Ball, Independent demographic consultant

The official projections of household numbers in England are of vital importance for debate and decision-making about the amount of land for housing development in English local authorities. These decisions have major social, environmental and financial consequences. The official projections are objective and based purely on past trends, but the concept of housing need is value-based. Consequently, there are arguments over whether the projections merely perpetuate past housing shortages, particularly in areas where there is the highest demand for housing. Furthermore, the 2012 official projections are not based on a full revision or updating of household formation trends in the light of the 2011 Census.  Difficulties with data are cited, and the possibility has been mooted of adopting a simpler methodology, similar to those in other UK countries. In these circumstances, how can local authorities get a better understanding of future changes in housing needs? While advanced research has been carried out, for example by Glenn Bramley at Herriot Watt University, this has yet to achieve widespread use in the preparation and public examination of local plans. Data from the 2011 Census and the official household projections will be used to compare household representative rates across local authorities in England. This is an initial exploratory analysis to establish if significant differences between areas are discernible. If differences exist then this raises questions about the reasons behind them. One outcome might be to stimulate the development of a simple analytical tool to assist local policy makers to make more intelligent use of the projections in decision-making. A second aim is to increase awareness of the issue among policy-makers and so stimulate the further development and adoption of more complex approaches. 


Preparing for housing projections in the Sheffield City Region
Richard Cooper, North-East Derbyshire District Council 

The Government's latest Planning Guidance indicates that housing provision should be 'benchmarked' against economic growth and its likely job growth. Localism has also been a characteristic of Government policy, meaning that plans are now prepared rather more in isolation. These two factors have diluted consideration of the spatial relationship within strategic job and housing market areas, such as travel to work areas and City-regions. Attempts to address this include Local Authorities working together on shared evidence for housing and attempting to agree shared approaches.  There are technical aspects to the work involving population projections. A suitable housing provision for districts may be considered in the light of prospective jobs or future migration. In both cases adjustments from past rates, which inform the projection, might be made to plan the future. Typically a range of possibilities - scenarios - might be prepared. However study of the results indicates that population and housing growth at individual local authority level may not properly reflect the way job growth across a wider area impacts on that local authority. This is particularly where job growth ambitions are applied - giving a misleading outcome, with a distorted housing requirement to deliver an anticipated labour force. This presentation will hopefully prompt discussion of how valid the case is for doubts about certain scenarios, and technical approaches to address the question of how much housing is needed across, say, a city-region. Also relevant is how results should be presented to help policy-makers understand the evidence underlying housing and labour force requirements. The presentation will be mostly discursive. 


Demographic projections: Improvements in official forecasts – Tuesday 8 September, 5.00pm 

Improvements in the Scottish sub-national population projection model: Results and planned developments
Esta Clark1, Luke Main1, Kirsty MacLachlan1 Philip Rees2, Paul Norman2, Nikolas Lomax2, 3Pia Wohland; 1Population and Migration Statistics Branch, Demographic Statistics, National Records of Scotland, 2School of Geography, University of Leeds, 3Institute of Health and Society, Newcastle University 

ONS is revising methods used in the National Population Projections (NPP), testing recommendations in the Bijak 2012 review. These changes imply that National Records of Scotland (NRS) should revise their Sub-National Population Projection (SNPP) methodology. NRS have collaborated with Leeds University in reviewing methods. The most important recommendation is to adopt the spatial framework of the 2014 NPP model and the ONS England SNPP. In the past the Scotland SNPP has been constrained only to Scotland’s NPP projected populations, births and deaths. The proposal is to constrain to four migration inflows to Scotland (from the other Home countries plus immigration) and to the equivalent four migration outflows. These flows within Scotland (by age) need to be estimated for Council Areas (CAs). This could be done using allocation factors for six inter-Home country migration streams computed using data from the 2001 and 2011 Censuses and patient registers, building on estimates by Lomax (2013) of migration flows between UK local authority districts. Another recommendation is that a multi-regional model for forecasting migration between CAs in Scotland be adopted. A final proposal is a general method for converting CA projected populations to overlapping areas, needed for planning purposes. The method draws on recent census small area populations and mid-year estimates for data zones and ensures consistency across the different projections. After hearing about the recommendations from Leeds team, NRS will then talk about those which have been incorporated into the methods for the 2014-based SNPP and present some results based on 2012 data. 


Updating the methods and tools for producing the UK National Population Projections
Amy Large, Paula Guy; Office for National Statistics 

The National Population Projections for the UK are currently produced by the Office for National Statistics (ONS) every two years. They provide a picture of how the population may develop in future years based on a variety of assumptions about future fertility, mortality and migration. The results are used by government departments to help plan for things like future pension provision, school place requirements and the demand for health services. Further to a recommendation from the ESRC CPC review of projections methodology, ONS intend to implement changes to the migration methodology in the 2014-based National Population Projections. These changes move from incorporating cross border migration as gross flows between countries of the UK to setting assumptions as rates. The methodology includes the option to apply an adjustment to the rates, similar to the method recently adopted by Statistics Canada, to ensure migration figures are responsive to the projected size of each of the UK countries. With these changes to the methodology, and with the civil service move to Windows 7 in 2014, the Excel based system previously used for producing the projections was no longer sustainable. This paper will discuss the approach taken to ensure a robust and flexible system, the new methodology which was incorporated into the new system, and how the Population Projections team collaborated with Statistical Computing Branch (SCB) to achieve a successful result. 


Forecasting age patterns of interregional migration 
Arkadiusz Wiśniowski1, James Raymer2, Jakub Bijak1,  Jonathan J. Forster1, Peter W. F. Smith1; 1University of Southampton, 2Australian National University 

Age-specific interregional migration is an important component of subnational population change yet models for forecasting the patterns for use in population projections are largely non-existent. National statistical offices tend to rely on simple deterministic assumptions regarding net migration or gross flows of in-migration and out-migration. These models do not take into account the linkages between origins and destinations and often have to be adjusted to ensure zero net migration and the same totals for in-migration and out-migration. In this paper, we focus on the full matrix of flows to avoid this problem. To deal with the large number of possible flows, we develop a Bayesian hierarchical time series approach to forecast age-specific interregional migration in Italy. The results demonstrate the differences that arise from different models specifications and the promise of the general approach. 


Estimating the population of the very old at sub-Scotland level
Gail Sinclair, Luke Main,  Esta Clark; National Records of Scotland 

As the number of those reaching very old ages increases, local authorities require more information on this section of the population in order to help with the planning of health and social care provision. Estimating at very old ages is considered problematic due to increased potential for incomplete or incorrect information. This study produces estimates for the 32 Scottish Council areas at single year of age from 90 to 100+ and considers a number of alternative methods. The Office for National Statistics and the National Records of Scotland currently produce estimates for ages beyond 90 at a national level using the Kannisto-Thatcher (KT) method; this method was given special attention. The KT method uses deaths data to estimate a survival ratio and assumes negligible migration. Whilst international migration is low, moves between Council areas are more likely, therefore data from the Scottish NHS Central Register and 2011 Census are used to estimate levels of migration at very old ages. Other established population estimation methods, including the Cohort-Component, Constant Ratio and Ratio Change methods, are used to produce estimates and the results analysed. Further exploratory work is carried out by making adjustments to the KT method. The number of cohorts/years used to calculate the survival ratio is altered. Results are recalculated starting at 85 instead of 90 years old. Results indicate that a close to constant ratio can be assumed and could be used as part of a combination method, where results are constrained to the existing official mid-year estimates for ages 90+. 


Small area projections – Wednesday 9 September, 11.30am 

Electoral projections for output areas
Michelle vonAhn, Local Government Boundary Commission for England 

Can accurate electoral projections be created for output areas? The methodology utilises Census data for output areas, local authority electoral registration data, ONS mid-year estimates and population projections. Population aged 18+ is derived from the mid-year estimates and used as a denominator against the relevant year's electoral register, repeated for as many year's data as available. The average of these proportions is applied to aged 18+ projections to create electorate projections. From Census data, the distribution of 18+ population is calculated: (OA population/LA population). These percentages can then be applied to the projected authority electorate to create electorate projections for output areas. The accuracy of the projections at English local authority level, and for their constituent wards is difficult to determine. The methodology does not address areas where there are large developments having local impact. However, this approach can provide initial GIS-based data for electoral reviews that can be modified utilising additional local information to reflect demographic change derived from residential development. The accuracy of this approach down to output area level can only be assessed in limited ways, as there are very few reviews that have been completed, implemented and with electoral data available for new wards. The accuracy of the data for local authority level is dependent on the ONS projections accurately reflecting change over time as well as stability in electoral registration rates - and with the introduction of Individual Electoral Registration there is likely to be trend disruption in areas with population turnover and student populations. 


Use of origin-destination census data to produce small area population projections for London
Ben Corr, Monica Li, Greater London Authority 

The Greater London Authority produces annually-updated local authority and ward-level projections of population to support planning within the city. These projections make use of census origin-destination outputs to determine migration parameters within the models. With the recent availability of data from the 2011 Census, the GLA has updated its models and produced a new round of projections. This presentation describes how the GLA uses census origin-destination data to produce population projections for London's six hundred-plus wards. This process includes the production of local authority- level projections and the creation of ward-level migration proxies. An exploration of the results will follow. 


Forecasting at ward level – the Manchester model
Elisa Bullen, Manchester City Council 

Manchester City Council Forecasting Model (MCCFM) brings together local intelligence to build on official estimates and projections by enhancing them with finer detail and outputting for small areas. Rather than applying city-level longitudinal trends and rates to project future population, longitudinal ward-level data have been used to calculate fertility and mortality rates. 2011 Census distributions have been used along with HESA data and school censuses to improve prediction. Local data continues to be gathered and input to keep refining the model and two main outputs are released per year. This presentation covers why the in-house model was created and how it was achieved.