Europe‘s Population Change in a Global Context Wolfgang Lutz Increasing political concern The continuing demographic transition Low fertility in Europe.

Europe‘s Population Change in a Global Context Wolfgang Lutz Increasing political concern The continuing demographic transition Low fertility in Europe Global demographic trends Education and human capital Toward balanced demographic trends

Political Concern “After the end of the cold war population ageing will be the greatest challenge for Europe. But since these fundamental demographic changes comes so slowly we are only starting to pay attention to them“ Helmut Kohl around 1994 (at a meeting of EU heads of state and government)

European Banking Federation 2004 The chief economists of European banks and banking federations urge EU member states to face up to the challenge of ageing, which will affect all areas of public policy. Governments’ main efforts should be concentrated on tackling three problem areas in particular. 1) Increasing the birth rate is particularly important since in many EU countries births have been below replacement level for some years and are expected to stay below replacement level in the foreseeable future. 2) Switching to partially capital-funded pension systems, eventually considerably decreasing the pay-as-you-go share of the fi nal pension of the retired individual. 3) Limiting early retirement possibilities for existing employees and eliminating the additional incentives for early retirement. Even more, timely retirement should be encouraged, at least for the time being. A gradual and moderate increase of the retirement age should also be considered as an option.

Commission of the European Communities 2005, Green Paper Importance of demographic challenge for the Lisbon Strategy “Europeans would like to have more children. But they are discouraged from doing so by all kinds of problems that limit their freedom of choice … It is also the case that families — the structure of which varies but which still constitute an essential part of European society — do not find the environment in which they live conducive to child- rearing”. If Europe is to reverse this demographic decline, families must be further encouraged by public policies that allow women and men to reconcile family life and work.”

EU-15, Uncertainty Distribution of Support Ratio (2000-2050)

The big picture: Demographic Transition Demographic Transition: After initial mortality decline fertility falls with some time lag. This paradigm explains much of the demographic trends over the 20th century. Original assumption: fertility decline will stop at replacement level. But reality has been different. Today half of world population is below replacement level

0 10 20 30 40 50 60 70 1875188518951905191519251935194519551965197519851995 Crude birth and death rates (per 1,000) CBR CDR Year BIRTH AND DEATH RATES IN MAURITIUS since 1871 Source: Mauritius Central Statistical Office.

Empirical Fertility Trends in Europe In most countries of Europe there was a “baby boom” in the 1960s. This “boom” was particularly strong in Western Europe. It was followed by a steep fertility decline during the 1970s. Europe was the first continent to fall below replacement level.

Figure 1. Western Europe, Total Fertility Rate, 1960-2001

Figure 2. Southern Europe, Total Fertility Rate, 1960-2001

Figure 3. Northern Europe, Total Fertility Rate, 1960-2001

EST LAT SLO H A D P E I GR B L NL S DK M CY IRL LIT F UK SR CR PL

Cross-country correlation between Total Fertility Rate and Female Labour Force Participation Rate for 21 OECD countries, 1960-2000

Figure 4. Central / Eastern Europe, Total Fertility Rate

Figure 5. Central / Eastern Europe, mean age of childbearing, 1960-2001

Birth Deficit due to the Tempo Effect An increase in the mean age of childbearing results in a lasting loss of births, unless the childbearing age decreases again

EU-15 Population Scenarios Source: Lutz, O’Neill, & Scherbov, Science, 2003. All scenarios assume constant mortality and no net migation. Year Instant Replacement Level Fertility 20 Years of Fertility at 1.8, then Replacement 20 Years of Fertility at 1.5, then Replacement The tempo effect accounts for about 45% of the population decline due to continued low fertility

Why a policy focus on the tempo of fertility? (1) It is well established that in a period when the mean age of childbearing increases, the period TFR is depressed relative to what it would be without such an increase. So far the tempo effect has been mostly considered as an undesired disturbance that needs to be ironed out (tempo- adjusted fertility rates) in order to have a purer measure. The tempo effect greatly matters for the number of births born in any given year. It has a lasting effect on the age structure and size of a population. Tempo policies aim at influencing the tempo effect and its lasting impacts on population dynamics.

GR A D I E P UK B L FIN DK NL S IRL F CY M SLO CR LAT SR LIT EST H PL

Why a policy focus on the tempo of fertility? (2) While choice of family size (quantum) is largely considered a private matter, where direct government interference is not welcome, a focus on the timing of births might be more favoured, particularly because there is a strong health rationale against further delays (risk of not getting pregnant, risks associated with late pregnancies). There are essentially two ways of effecting the mean age of childbearing: A - Change the typical sequence of life cycle phases (e.g. having children while still at university) B - Shorten the duration of those phases that typically precede childbearing (e.g. shorten the study time to a certain degree)

What we know about demographic trends: We know the past trends in fertility, mortality and migration We know the current structure by age and sex (plus other characteristics) We know much about the future age structure of the population, because everybody above age 25 in 2030 has already been born (we know cohort size)

Population Pyramid for Austria, 2000

Probabilistic Age Pyramid for Austria 2030 Yellow area gives 95 percent interval, green 60 percent and blue 20 percent

What we do not know: We do not know whether fertility will recover or continue to decline We do not know whether we are already close to maximum life expectancy or still far away (if there is a limit at all) We do not know the future political conditions that will determine migration We do not know well how acceptable and affordable policies can enhance level of fertility But we know that the plausible range of these factors is such that it cannot significantly alter the ageing trend

Uncertainty Range of Future Support Ratio in the European Union 2000-2050

How low can fertility fall? The possible “low fertility trap” hpyothesis Observation that countries that fell below fertility level of 1.5 children have hardly recovered. Negative demographic momentum: Because of past low fertility there will be fewer potential mothers in the future. Economics: Gap between aspirations for consumption and expected income widens for young people due to negative consequences of ageing (cuts in social security systems, possible economic stagnation) Ideational change: Young people are socialized in an environment with few children – may result in lower ideal family size in next generation.

Western Europe, Uncertainty Distribution of Proportion above Age 80 (2000-2100) UN “low” UN “high”

Can Migration Compensate for the Missing Births? Alternative Projections of the Old Age Dependency Ratio for the EU-15 in 2050 based on different Fertility and Migration Assumptions (Black line gives the level in 2000)

Research Priorities and Institutional Needs Analysis of variation across Europe is most important source for understanding the nature and determinants of processes. Yet demographic studies in Europe have traditionally been conducted by national institutes with primarily national focus. There is an urgent need for comparative European demographic analysis that goes beyond the networking that already exists among national institutes. International efforts (ECE, Council of Europe, Eurostat, DG Empl, etc.) are well below critical mass.

The End of World Population Growth in the 21 st Century: New Challenges for Human Capital Formation and Sustainable Development A new IIASA Population Book W. Lutz, W. Sanderson and S. Scherbov (Eds.)

The End of World Population Growth in the 21st Century: New Challenges for Human Capital Formation and Sustainable Development The End of World Population Growth Applications of Probabilistic Population Forecasting Future Human Capital: Population Projections by Level of Education Literate Life Expectancy: Charting the Progress in Human Development Population-Environment-Development-Agriculture Interactions in Africa: A Case Study on Ethiopia Interactions between Education and HIV: Demographic Examples from Botswana China’s Future Urban and Rural Population by Level of Education Population, Greenhouse Gas Emissions and Climate Change Conceptualizing Population in Sustainable Development: From “Population Stabilization” to “Population Balance”.

World Population from the year 1000 to 2100 (historical data from 1000 to 2000; deciles of IIASA’s probabilistic forecasts to 2100) Source of historical data: UN 2001

World Population Trends: The Big Picture While the 20 th century was the century of population growth (with the world population increasing from 1.6 to 6.1 billion), the 21 st century will be that of population aging (with the proportion above age 60 increasing from currently 0.10 to 0.25-0.45 by 2100).

Forecasting the Population by Age and Sex For forecasting the population we need the current population by age and sex for each region. We need to make assumptions on the three components of change: Fertility (birth rates) Mortality (death rates, life expectancy) Migration The future paths of all three factors are uncertain. Therefore we produce probabilistic population projections.

What is wrong with the UN “Variants” Approach ? 1.High, medium and low variants only differ in fertility (mortality and migration uncertainty disregarded). 2.The high-low interval is supposed to give a “plausible range” of future outcomes without being specific (does it cover 99% or 50% of possible outcomes?) 3.It is probabilistically inconsistent when aggregating to regional and global totals (the global high variant is simply the sum of all regional high variants). 4.It assumes smooth and monotonous changes in components (i.e. no booms and busts)

Three Approaches to Probabilistic Forecasting Time series analysis (Lee and Tuljapurkar). Problems: assumption of no structural change, TS data often not available. Ex post error analysis: You assume that the errors of past projections will be replicated in the future (Keilman). Problems: mechanistic (why should we do the same errors now than others did decades ago?) Expert argument based uncertainty distributions: subjective probabilities (Lutz et al. 1996). Problems: How biased are experts? Proposed Synthesis Approach: “The end of world population growth” (Lutz et al 2001 in Nature) State of the Art Discussion: 2004 Special Issue of International Statistical Review: “How to deal with uncertainty in population forecasting? (Lutz and Goldstein, Guest editors).

Errors in past UN projections for South-East Asia

Nine sample paths (out of 2,000 simulated paths) of world population size from 2000 to 2001

Deciles of the resulting uncertainty distribution of world population size to 21000.

World Population from the year 1000 to 2100 (historical data from 1000 to 2000; deciles of IIASA’s probabilistic forecasts to 2100) Source of historical data: UN 2001

Probability of End of Population Growth: Proportion peaking prior to the indicated year (out of 2,000 simulated population paths)

1 st, 3 rd, 5 th (median), 7 th, and 9 th deciles of the forecasted distributions of world population size at 10-year intervals from 2000 to 2100. Note: Uncertainty measure is (9 th decile-1 st decile)/median.

Western Europe, uncertainty distribution of the proportion above age 80. UN “high” UN “low” UN “high”

Adding Education to Age and Sex Formal education is typically acquired at young ages and then does not change over the life course (goes along cohort lines, multi-state population models). This is why the educational composition of the total population changes only very slowly. Educational efforts made today will only improve the average education of the work force many years later.

Education Matters For individual life (Micro) For economic development (Macro) “Whereas at the micro case... it is established beyond any reasonable doubt that there are tangible and measurable returns to investment in education, such evidence is not as consistent and forthcoming in the macro literature “ (Psacharopoulos and Patrinos 2002) Findings are dependent on education indicators chosen (age-range). For health/mortality and fertility (cross-sectoral effects) Indirect effects of education on institutions and good governance (control by the educated)

Fertility Rates by Level of Education Region (1) No Education (2) Primary (3) Secondary and higher Difference (1) – (3) SS-Africa 6.45.53.72.7 North Africa 4.73.62.81.9 Asia (East +SE+South) 4.13.52.71.4 West Asia 6.44.63.52.9 Latin A.+ Caribbean 5.84.52.63.2

Measuring Formal Education Education Flows – Policy variable (Gross and Net Enrolment by Age, Repetition Rates) Education Stocks - Change very slowly due to great momentum –Mean years of schooling –Distribution by highest educational attainment –Functional literacy

Measuring Human Capital Stocks Directly from Census Data (Barro and Lee, 25+; Lutz and Goujon, 5-year age groups) For Forecasting: Perpetual Inventory Method (Nehru, Swanson and Dubey 1993, Ahuja and Filmer 1995). Sum over past school enrollment rates then superimpose it to UN population projections. Demogrphic Multi-state Projection Methods (Developed at IIASA in the 1970-80s by Rogers, Keyfitz and others) Full cohort component projections which consider educational attainment in addition to age and sex (can consider also fertility, mortality and migration differentials by level of education)

Principles of Population Projection by age and sex Migration Mortality Migration Fertility Migration Males Population by Age and Sex 2000 2005 MalesFemales

Migration Mortality Migration Fertility Migration Principles of Population Projection by age, sex, and education Population by Age, Sex, and Education 2000 2005

Projecting the population by level of education We need to know the current composition of the population by age, sex and education categories. We need to know how the birth rates differ for women with different levels of education. We need to know school enrollment at different levels and make alternative assumptions for the future education transition rates. We need to make assumptions about future trends in fertility, mortality and migration by level of education.

Country Specific Scenario: TFR 1.5 in 2030, Education Constant