Multi-states Projections: A Window on the Dynamics of Heterogeneous Populations

1
Multi-states Projections A Window on the
Dynamics of Heterogeneous Populations
Anne Goujon (goujon_at_iiasa.ac.at)
International Institute for Applied Systems
Analysis (IIASA), Austria Vienna Institute of
Demography (VID), Austrian Academy of Sciences,
Austria
2
Outline

• Multi-educational states
• Principles
• Why? (3 criteria)
• Example India
• Multi-religious states
• Projections of Austrias main religions

3
PART 1 Population Projections by Level of
Education

• Already several case studies
• Pioneer work in Mauritius (Lutz et al. 1994) and
  Cape Verde (Wils 1995)
• North Africa (Yousif Goujon Lutz 1996)
  Algeria, Egypt, Libya, Morocco, Sudan, Tunisia.
• Middle Eastern Countries (Goujon 1997 2002)
  Jordan, Lebanon, Syria, West Bank and Gaza Strip.
• Lebanons six administrative regions (Goujon
  Saxena 1999, unpublished)
• Yucatan (Goujon et al. 2000).
• 13 world regions (Lutz Goujon, 2001)
• Indias 15 administrative states (Goujon
  McNay, on-2003
• Egypt and Egyptian governorates (Goujon et al.
  2007)
• Southeast Asia (Goujon K.C., 2007)
• 120 countries (Lutz et al. on-going)

4
Principles of Population Projectionby Age and Sex
Mortality
Males
Females
Males
Females
Migration
Migration
Migration
Fertility
Population by Age and Sex Population by
Age and Sex 2005 2010
5
Principles of Population Projectionby Age, Sex,
and Education
Mortality
Males
Females
Males
Females
Migration
Migration
Migration
Fertility
Population by Age, Sex, and Education
Population by Age, Sex, and Education
2005 2010
6
Why Education???

• Education answers the three main criteria of why
  to explicitly consider a particular dimension in
  population projections
• It is interesting as such and is a desirable
  explicit output parameter
• It is a source of demographic heterogeneity and
  has an impact on the dynamic of the system
• It is feasible to consider the dimension
  explicitly

7
Why Education??? Interesting as such a
desirable explicit output parameter

• Output of the projection the level of
  educational attainment of the population by age
  and by sex for a defined period
• Picture of human capital composition (age-group
  20-64) in absolute values.
• Show long term effects of education policies The
  momentum of population and education change in
  development planning
• Assess according to present pace of improvements
  the likelihood of the realization of certain
  education/development goals
• Education is a good proxy for quality of life,
  autonomy of women, level of economic development.
  

8
Education and Economic Growth(Lutz
Crespo-Cuaresma, 2007)

• The educational attainment of younger adults is
  key to explaining differences in income across
  all countries.
• For the poor countries, it turns out that not
  only universal primary education, but also
  secondary education of broad segments of the
  population boosts economic growth.

9
Why Education??? A source of demographic
heterogeneity with an impact on the dynamic of
the system

• No other socioeconomic variable shows a similar
  degree of association with fertility (result
  shown from WFS and DHS).
• Female education is also related to infant and
  maternal mortality mortality differentials exist
  at almost all ages and for both sexes
• The education-fertility relationship is very
  relevant because the education level of a society
  can be directly influenced by government policy.
  This brings the State to be the key variable in
  the demographic transition.

10
Fertility (TFR) differentials by womens
education in 2001-2006
Source Demographic and Health Surveys
11
Heterogeneity in the Level of Heterogeneity

• Fertility differentials between upper and lower
  education groups tend to cluster regionally, with
  linkages to the level of socioeconomic
  development, the stage of the demographic
  transition, the stage in the level of mass
  education in the country and the cultural setting
  (Jejeebhoy 1995, Cochrane 1979, UN 1987)
• Narrowest fertility gap countries quite advanced
  in the process of development and demographic
  transition
• Largest differentials Countries in settings of
  medium development and halfway through the
  process of demographic transition.
• Developed world narrow gap with a diminishing
  negative effect of education and in some
  countries a high education even turns into a
  stimulating factor (Kravdal, 2001).

12
Infant Mortality by Mothers Education
Factor by which IMR is higher for uneducated
women than for women with secondary or higher
education
Source Macro-International, Demographic and
Health Surveys, 2007
13
Ability to Perform Daily Activities Activity of
Daily Living scores by education Southeast Asian
countries
Source Lutz and K.C. 2007
14
Why Education??? Feasibility to consider the
dimension explicitly

• Multi-state population projection tools exist
• For instance

PDE Population Projection Software (IIASA)
PopEd (Sergei Scherbov, VID)
15
Multi-State Cohort Component Method the
Extended Leslie Matrix

• The multi-state population projection method
  allows division of the population to be projected
  into any number of states originally
  geographic regions (Rogers 1975) and for our
  purpose educational categories
• Combination of the discrete time cohort component
  projection used for single-state populations
  (Leslie 1945), and an adapted form of the
  multi-state population projection method first
  compiled in complete form by Rogers (1975) and
  Wilson and Rogers (1980).
• The demographic method of cohort-component
  projection is most appropriate to educational
  projections because education is typically
  acquired in childhood and youth and then changes
  the educational composition of the population
  along cohort lines.

16
The Extended Leslie Matrix

• Multi-state projection method the age- and
  sex-specific population is further divided into
  states and the transitions between these states
  are included in the projection.
• Transitions are specific to each age and gender
  group, and are represented by age- and
  sex-specific transition matrices.
• These transition matrices can replace the age-
  and sex-specific birth, death, and net migration
  scalars in the Leslie matrix.
• The multi-state population projection is then
  represented as an extended Leslie matrix.
• The population vector is also extended to include
  the population by states.
• The matrix is arranged as the original one-state
  Leslie matrix, but now, each scalar in the matrix
  has been replaced by a small transition matrix
  and each scalar in the population vector is a
  small vector of the population states.
• Transitions refer to movements from one state to
  another and are distinct from mortality or its
  inverse, survivorship. Each transition can be
  called Tij (a) which means the transition rate
  into state i out of state j in age group a. In
  every period, each person is exposed to a certain
  probability of making a socio-economic transition
  and to dying. Thus, in the matrix of transitions,
  survivorship S(a) and the transitions Tij (a) are
  included.

17
Data AvailabilityPopulation, Fertility,
Mortality, Migration, Transitions

• Population by age, sex and education can be
  extracted directly from censuses, but also from
  UNESCO publications, and others.
• Fertility data by education can be extracted from
  DHS, and other surveys.
• Mortality data are more difficult to obtain for
  all age groups but exists for some countries.
• Migration data by education can sometimes be
  extracted from censuses or surveys.
• Transitions probabilities have most of the time
  to be calculated, e.g. based on two surveys or
  along cohort lines.

18
Example India (1970-2050)
19
India in 1970
Source Lutz, Goujon, K.C. and Sanderson 2007
20
India in 1975
Source Lutz, Goujon, K.C. and Sanderson 2007
21
India in 1980
Source Lutz, Goujon, K.C. and Sanderson 2007
22
India in 1985
Source Lutz, Goujon, K.C. and Sanderson 2007
23
India in 1990
Source Lutz, Goujon, K.C. and Sanderson 2007
24
India in 1995
Source Lutz, Goujon, K.C. and Sanderson 2007
25
India in 2000
26
India in 2005
Source Lutz, Goujon, K.C. and Sanderson 2007
27
India in 2010
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
28
India in 2015
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
29
India in 2020
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
30
India in 2025
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
31
India in 2030
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
32
India in 2035
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
33
India in 2040
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
34
India in 2045
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
35
India in 2050
Goal Scenario
Constant Enrolment Scenario
Source Lutz, Goujon, K.C. and Sanderson 2007
36
Total Population 1,658,270,000
India in 2050
Goal Scenario
Constant Enrolment Scenario
Total Population 1,807,725,000
Source Lutz, Goujon, K.C. and Sanderson 2007
37
New Times, Old Beliefs
PART 2 Population Projections by Religion

• Predicting the future of religions in Austria
• Anne Goujon, Vegard Skirbekk, Katrin
  Fliegenschnee, Pawel Strzelecki

38
Austrian Population by Religion 1900-2001
Source Statistic Austria, Census 1900 to 2001
39
Religious Influences on Demographic Events
Most major religions contain texts and commands
to increase their number of followers. The
Bible promotes childbearing (Gen 128) And God
blessed them, and God said unto them, Be
fruitful, and multiply, and replenish the earth.
While Mohammed says Marry women who are loving
and very prolific for I shall outnumber the
peoples by you (al-Masabih 1963, p 659)
Marriages are endorsed in all religions and
divorced are largely forbidden in Catholicism and
Islam. Protestants permit divorce. Interreligious
marriages are allowed in Islam only if the
husband is Muslim. All major religions promote
transmission of religions to their children.
Conversion or secularization is strongly
discouraged in all religious, although the degree
of punishment differ according to religion and
society.
40
Fertility Differences
 
 


 
41
Different Fertility Patterns
42
Main Questions for the Projections

• Question 1 If secularization and the increase
  of other religions in the population continue,
  when will Roman Catholics make up less than 50
  of the total population?
• Question 2 Will the Muslims or those without
  religion become the dominant group in Austria?
• Question 3 What is the influence of migration
  on the religion structure of the country?
• Question 4 Could a change in the religious
  composition lead to increased fertility in
  Austria?

43
12 Scenarios from 2001 to 2051 Fertility
2 fertility scenarios
Constant Fertility by religion
Converging Fertility by religion
44
12 Scenarios from 2001 to 2051 Secularization
18 Scenarios from 2001 to 2051 Secularization
3 transition/secularization scenarios
Constant secularization trend ( 2001-05)
High secularization trend (2 2001-05)
Low secularization trend (0)
45
12 Scenarios from 2001 to 2051 Migration
2 migration scenarios
46
Results Total Population of Austria, 2001-2051
47
Results TFR of Austria, 2001-2051
Results Total Fertility Rate
48
Results Proportion Roman Catholics in Total
Population, 2001-2051
49
Results Proportion Protestants in Total
Population, 2001-2051
50
Results Proportion Muslims in Total Population,
2001-2051
51
Results Proportion Other Religions in Total
Population, 2001-2051
52
Results Proportion Without religion in Total
Population, 2001-2051
53
Age and Religion A Clash of Generations
54
The Answers to the Questions for the Projections

• Question 1 If secularization and the increase
  of other religions in the population continue,
  when will Roman Catholics make up less than 50
  of the total population? Starting from 2031
• Question 2 Will the Muslims or those without
  religion become the dominant group in Austria?
  Not before 2051
• Question 3 What is the influence of migration
  on the religion structure of the country? Quite
  important
• Question 4 Could a change in the religious
  composition lead to increased fertility in
  Austria? Yes, but not really

55
Conclusion
Thank you
Conclusion

• Global aggregate figures of any kind tend to have
  little meaning
• Information content typically lies in variation
• Variation can be over time, space or over
  individuals (sub-populations)
• Such variation is the source of information for
  studying change as well as its determinants and
  consequences.
• To also make sense of such information we need
  theories, hypotheses, models.

Questions comments