DYNAMIS A Portable Dynamic Socio-Demographic Microsimulation Model - - PowerPoint PPT Presentation

▶
dynamis
SMART_READER_LITE
LIVE PREVIEW

DYNAMIS A Portable Dynamic Socio-Demographic Microsimulation Model - - PowerPoint PPT Presentation

Sep 06, 2023 547 likes •854 views

DYNAMIS A Portable Dynamic Socio-Demographic Microsimulation Model for Developing Countries Martin Spielauer, Olivier Dupriez DYNAMIS-POP-NPL DYNAMIS o Dynamic Micro-Simulation POP o Focus on Population projections o The demographic

slide-1
SLIDE 1

DYNAMIS

A Portable Dynamic Socio-Demographic Microsimulation Model for Developing Countries Martin Spielauer, Olivier Dupriez

slide-2
SLIDE 2

DYNAMIS-POP-NPL

  • DYNAMIS
  • Dynamic Micro-Simulation
  • POP
  • Focus on Population projections
  • The demographic core for other applications
  • NPL
  • Country application for Nepal
  • First version was for Mauretania DYNAMIS-POP-MRT

http://ihsn.org/projects/dynamis-pop

slide-3
SLIDE 3

What is Dynamic Micro-Simulation?

Computer-simulation of a society in which the population is represented by a large sample of its individual members and their behaviors.

  • Macro Models project cell-sizes
  • Dynamic microsimulation projects individual life courses

and the interaction between people

slide-4
SLIDE 4

When does it make sense?

When macro models are technically too restrictive

  • Number of variables
  • Types of variables (non-categorical)
  • Process types (non-Markov)

When longitudinal consistency is required

  • Realistic life-courses
  • Longitudinal accounting

Modeling of interactions

  • Life course interactions and downstream effects
  • Linked lives, transmission
  • Policies
slide-5
SLIDE 5

Limitations

Transitory limitations

  • Computer power
  • Development costs
  • Data requirements

Randomness affecting prediction power

  • Prediction power depends on model specification and

randomness which increases with detail

  • Difficulty to find optimal point between too simple

models (misspecification error) and too detailed models (randomness)

  • Not a limitation for population projections
slide-6
SLIDE 6

How are MS Models Created?

  • Start from a population data base
  • Individual characteristics
  • Links to other persons
  • Behavioral models for updating individual characteristics
  • Discrete time models: updates in fixed time steps

using models of probabilities

  • Continuous time models: competing risk approach

based on rates and corresponding waiting times.

  • Accounting routines
  • Aggregated model output tables
  • Tax-benefit accounting
slide-7
SLIDE 7

DYNAMIS– Characteristics & Philosophy

  • Portable platform
  • Based on data available for most countries
  • Refinable, extendable & adaptable to specific contexts
  • Modularity
  • Library of analysis tools and models
  • Selection of models, geographic depth, variables to be

included in country context

  • Automated generation of model parameters
slide-8
SLIDE 8

DYNAMIS– Characteristics & Philosophy

  • Start from the ‚known‘ (available macro projections)
  • Model can reproduce macro models: same

assumptions, parameter tables -> same outputs

  • More refined models can be added and selected with

and without alignment to macro projections

  • Reproducible
  • Step-by-step documentation (stats & programming)
  • Freely available software (R, Modgen, DYNAMIS)
slide-9
SLIDE 9

DYNAMIS – Characteristics & Philosophy

  • User friendly
  • Fully documented GUI and model (help files)
  • Parameters typically have intuitive interpretation
  • Scenario management: parameters stored together

with results

  • Analysis tool
  • Easy to specify meaningful scenarios
  • Overall trends vs. trends in relative differences
  • Downstream effects
  • Decomposition of effects
  • Composition vs. behavioral effects
slide-10
SLIDE 10

DYNAMIS – Characteristics & Philosophy

  • Rich Output
  • Extendable hierarchical list of tables
  • Micro-data output (projected cross-sectional and

panel data)

  • Database of individual histories for graphical

visualization (BioBrowser)

  • R-Integration
  • Data analysis and parameter file generation
  • DYNAMIS can be run from command line/from R
  • Post-processing of simulation results
slide-11
SLIDE 11

DYNAMIS – GUI

slide-12
SLIDE 12

Data

Census Micro-Data DHS Micro-Data

Fix Prepare Analyze

Starting Population Parameter Files

slide-13
SLIDE 13

DYNAMIS – Fertility

  • Base Version
  • Age-specific fertility distribution by year
  • Total Fertility Rate (TFR) by year
  • Extended Version
  • First births by age, union status, education, province
  • Higher order births by education, time since last birth
  • Separate trends by birth order
  • Alignment Choices (extended version)
  • Not aligned
  • Aligned to total births of base version
  • Aligned to total births by age of base version
slide-14
SLIDE 14

DYNAMIS – Mortality

  • Base Version
  • Standard life table of age-specific rates by sex
  • Life expectancy by calendar year and sex
  • Refined child mortality model (ages 0-4)
  • Age baseline
  • Relative risks by mothers education and age group
  • Age-specific overall trends
  • Alignment options (refined model)
  • Without
  • Initial alignment to base model – trends from base
  • Initial alignment to base model – specific trends
slide-15
SLIDE 15

DYNAMIS – Internal Migration

  • Base
  • Probabilities to leave by province, age group and sex
  • Distribution of destinations by origin, age, sex
  • Refined
  • Education added to probability to leave
slide-16
SLIDE 16

DYNAMIS – Immigration and Emigration

  • Immigration
  • Immigration numbers by year and sex
  • Age distribution by sex
  • Destination distribution by sex and age
  • Emigration
  • Emigration rates by province, age and sex
slide-17
SLIDE 17

DYNAMIS – Primary Education

  • Base: Probabilities of school entry and graduation by

year & province of birth, sex

  • Based on proportional models (logistic regression)
  • Typically high and persistent inter-provincial differences
  • Refinements:
  • Choice of geographical level (region, district)
  • Transmission: adding mothers education
  • Grade system: tracking of students by grade
slide-18
SLIDE 18 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 2019 2022 2025 2028 2031 2034 2037 2040 2043 2046 2049

Primary School Entry

Female Kathmandu Female Rautahat Male Kathmandu Male Rautahat 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 2019 2022 2025 2028 2031 2034 2037 2040 2043 2046 2049

Primary School Retention

Female Kathmandu Female Rautahat Male Kathmandu Male Rautahat

DYNAMIS – Primary Education

  • Nepal
  • Mauretania
slide-19
SLIDE 19

DYNAMIS – School Progression

  • Parameters for intake, success, progression, repetition
  • ...or a „best guess“: optional automatic calibration to

meet target graduation rates.

slide-20
SLIDE 20

DYNAMIS – School Progression

  • Illustration: Primary students by grade in Mauretania
  • Scenario 1: Scenario 2: Improvements

Current trends towards universal graduation

slide-21
SLIDE 21

DYNAMIS – First Union

  • Option A: Age-specific rates
  • Option B: Parametric model by Coale & McNeil
  • Parameters: lowest and average age at first union

formation and final outcome of ever entering a union

  • Simulation results can be used as base for option A

(which can be easily modified e.g. to a scenario banning child marriages)

slide-22
SLIDE 22

DYNAMIS – First Union - Analysis

slide-23
SLIDE 23

DYNAMIS – First Union - Analysis

slide-24
SLIDE 24

Example: Effects of Education

  • 2 Scenarios:
  • Base: Continuing observed trends
  • Alternative: Phased-in universal primary, cohorts

2005-2010

slide-25
SLIDE 25

Example: Effects of Education

0% 20% 40% 60% 80% 100% 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023 2025 2027 2029 .. 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 Base Szenario .. Alternative Szenario

Primary Education Of 18 Year Old - Kathmandu

Never entered primary school Primary school non-completer Primary school graduate 0% 20% 40% 60% 80% 100% 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023 2025 2027 2029 .. 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 Base Szenario .. Alternative Szenario

Primary Education Of 18 Year Old - Rautahat

Never entered primary school Primary school non-completer Primary school graduate

slide-26
SLIDE 26

Example: Births by Mother‘s Education

0% 20% 40% 60% 80% 100% 2004 2007 2010 2013 2016 2019 2022 2025 2028 2031 2034 2037 2040 2043 2046 2049 2004 2007 2010 2013 2016 2019 2022 2025 2028 2031 2034 2037 2040 2043 2046 2049 Kathmandu .. Rautahat

Distribution of Births by Mother's Education - BASE SZENARIO

Never entered primary school Entered primary school Graduated from primary school 0% 20% 40% 60% 80% 100% 2004 2007 2010 2013 2016 2019 2022 2025 2028 2031 2034 2037 2040 2043 2046 2049 2004 2007 2010 2013 2016 2019 2022 2025 2028 2031 2034 2037 2040 2043 2046 2049 Kathmandu .. Rautahat

Distribution of Births by Mother's Education - ALTERNATIVE SZENARIO

Never entered primary school Entered primary school Graduated from primary school

slide-27
SLIDE 27

Example: Teenage Births, Infant Deaths (0-4)

5000 10000 15000 20000 25000 30000 35000 40000 2001 2005 2009 2013 2017 2021 2025 2029 2033 2037 2041 2045 2049

Teenage Pregnancies

Base <17 Base 17-18 Alternative <17 Alternative 17-18 5000 10000 15000 20000 25000 30000 35000 40000 45000 2003 2006 2009 2012 2015 2018 2021 2024 2027 2030 2033 2036 2039 2042

Child Deaths (Age 0-4)

Base Szenario All Alternative Szenario All
  • Nepal
  • Mauretania
slide-28
SLIDE 28

Validation - Nepal

2000000 1500000 1000000 500000 500000 1000000 1500000 2000000 (min,5) [5,10) [10,15) [15,20) [20,25) [25,30) [30,35) [35,40) [40,45) [45,50) [50,55) [55,60) [60,65) [65,70) [70,75) [75,80) [80,max) 2001_Data Male 2001_Data Female 2011_Projection Male 2011_Projection Female 2011_Data Male 2011_Data Female

slide-29
SLIDE 29

Validation - Nepal

2000000 1500000 1000000 500000 500000 1000000 1500000 2000000 (min,5) [5,10) [10,15) [15,20) [20,25) [25,30) [30,35) [35,40) [40,45) [45,50) [50,55) [55,60) [60,65) [65,70) [70,75) [75,80) [80,max) 2001_Data Male 2001_Data Female 2011_Projection Male 2011_Projection Female 2011_Data Male 2011_Data Female

slide-30
SLIDE 30

Next Steps

  • Current work on model improvements
  • Ethnicity (in Nepal: Casts) as additional dimension
  • Refined models for migration: temporary work

migration, scenario building

  • Refinements of fertility models: regional and ethnic

differences

  • Full R integration: post-processing of output
  • Model validation
  • New applications
  • Research collaborations