The IMAGE Project - Comparing Internal Migration Around the GlobE: Data, Methods, Variations and Explanations Martin Bell and Elin Charles-Edwards Presentation to Vienna Institute of Demography September 2015
Context Compared with other demographic processes, such as fertility and mortality, little attention has been given to the way in which internal migration varies between countries around the world. Comparative indicators conspicuous by their absence from international statistical tables.
Significance of Internal Migration Estimated 865 million people outside region of birth Aggregate Level: Major force shaping patterns of human settlement Underpins functioning of the economy Integral to human development Individual Level Essential link to life course transitions Mechanism to meet individual aspirations Means of escape from risk & danger
Why Make Comparisons? Aids understanding Promotes analytical rigor Enhances migration theory Assists policy development Rising recognition of significance
Impediments Multifaceted nature of migration Lack of ready access to data Disparities in the way internal migration is captured Differences in spatial and temporal frameworks The absence of commonly agreed statistical indicators
The IMAGE Project An international collaborative research program which aims to provide a robust basis for comparing internal migration between countries around the world Focus on 193 UN Member States Funded by Australian Research Council Discovery Project 2011-15 http://www.gpem.uq.edu.au/image The IMAGE Inventory of Migration data The IMAGE Repository of Migration data The IMAGE Studio: computes metrics Focus: Migration intensity Migration distance Migration impacts Migration connectivity Outputs: Thematic Methodological Regional
Outline 1. Summarise the global inventory and data repository 2. Describe league table of migration intensities 3. Introduce the IMAGE Studio 4. Summarise the extent of global variation 5. Explore why mobility varies between countries
Who collects what?
Towards harmonisation Censuses, Registers, and Surveys collect different types of data (e.g. moves/movers, duration of residence) Migration is measured over different intervals No analytic solution for temporal harmonisation Spatial frameworks differ widely from one country to another No simple method of spatial harmonisation
Comparing migration intensities Strict comparability requires: Estimates of all changes of address (combining residential mobility and migration) Crude Migration Intensity ACMI = (100*M) / P All moves (M) only available for 30 countries Our solution: Estimate overall intensities based on Courgeau et al (2012) Make separate comparisons for one and five year intervals
Courgeau s k (1973) Defines an index, k, to standardise for the number of zones in a country: if the propensity to move is a function of distance, then the crude migration intensity (CMI) must be related to the number of zones into which a territory is divided. The finer the spatial mesh, the larger the number of migrations that will be captured and hence the greater the recorded intensity k represents the slope of a line derived by plotting the CMI from two or more sets of geographies against the log of the number of zones in each geography Algebraically, CMI=k log (n)
Crude Migration Intensity (%) Courgeau et al. (2012) Plots CMI against average households (H) per zone (j) CMI j = w + k ln (H/j) When H/j = 1 (i.e. average of 1 household per zone) then ACMI = w (representing aggregate mobility) Implementation constrained by available geographies (e.g. states, provinces, counties) Create additional geographies using the IMAGE Studio 12 10 8 6 4 2 0 Germany Administrative levels 0 5 10 15 20 ln (number of households/number of spatial units)
The IMAGE Studio Data Preparation (Flow matrix, PARs, Shapefiles of Basic Spatial Units (BSUs)) Spatial Aggregation (Randomly aggregates BSUs into Aggregate Spatial Regions (ASRs)) Computation of Indicators (Intensity, Impact, Connectivity) Spatial Interaction Model
IMAGE Studio Subsystem Interfaces Data Preparation Spatial Aggregation Internal Migration Indicators Spatial Interaction Modelling
How and why of spatial aggregation? Creates a range of additional geographies by random aggregation of Basic Spatial Units (BSUs) into progressively larger regions 500..490..480.110..100 Multiple configurations at each spatial scale Generates additional geographies for estimation of ACMI using Courgeau Enables us to explore the impact of the MAUP on migration indices Scale component: How does the indicator vary according to the number of Aggregated Spatial Regions (ASRs)? Pattern component: How does the indicator vary according to the configuration of ASRs at any spatial scale?
Example of aggregation: Germany 412 BSUs 200 ASRs 150 ASRs 100 ASRs 50 ASRs 10 ASRs
Crude Migration Intensity (%) Using the Courgeau et al. (2012) method 14 12 1 year event Estimated overall mobility (a) 10 8 6 4 Sweden_E Germany_E Belgium_E Finland_E 2 0 0 5 10 15 20 Ln (No of Households / No of ASRs)
IMAGE Repository Region Total countries Countries for which data are held Data on all changes of address Origindestination matrix Africa 54 18 2 18 Asia 46 23 4 18 Europe 44 36 13 35 Latin America 32 23 2 23 North America 3 3 2 3 Oceania 14 4 2 4 Total 193 107 25 101
League Table Coverage Region Number of countries Percentage of population Total countries in region Africa 17 43.8 54 Asia 17 83.4 46 Europe 34 97.1 44 Latin America 21 98.6 32 North America 32 100.0 3 Oceania 3 76.3 14 Total 95 81.1 193 81% of global population 35/98 countries missing have pop n<1 million 3/98 have pop n >100 million (Pakistan, Nigeria & Bangladesh)
One year intensities Fiver year intensities Migration intensities Iceland Australia Finland Zambia Kenya Canada Norway Sweden USA Denmark Tanzania UK Switzerland Belgium Netherlands Germany Sudan Colombia Greece Austria Japan Hungary Turkey Israel Ireland Belarus Lithuania Latvia Italy Portugal Estonia Malta Bulgaria Czech Republic Burkina Faso Romania Cyprus Spain Slovakia Croatia Ukraine Slovenia Russian Federation Poland Macedonia 19,1 17,6 17,0 13,3 15,4 16,0 12,8 12,7 12,5 12,2 12,0 10,9 10,7 9,0 10,1 10,3 8,8 8,6 8,1 8,1 7,9 7,4 7,1 7,0 6,6 5,7 5,5 5,1 5,1 5,1 4,8 4,7 4,5 4,2 4,0 4,0 3,8 3,1 2,7 2,7 2,7 2,5 2,0 1,3 1,0 0,0 5,0 10,0 15,0 20,0 ACMI (%) New Zealand 54,7 South Korea 52,8 USA 44,3 Australia 42,4 Fiji 41,3 Canada 38,5 Panama 37,1 Chile 36,2 Switzerland 36,1 Senegal 34,7 France 34,0 Cameroon 32,9 Paraguay 31,0 Japan 28,9 Israel 28,2 Mongolia 27,4 Barbados 26,5 Bolivia 22,9 Kyrgyzstan 22,4 Peru 21,8 Uruguay 21,5 Guinea 21,3 South Africa 21,2 Morocco 20,8 Malta 19,8 Uganda 19,1 Cambodia 18,4 Rwanda 18,1 Greece 18,1 Brazil 17,5 Argentina 17,5 Malaysia 17,1 Costa Rica 16,7 Tunisia 15,9 Ghana 15,0 Guatemala 14,4 Dominican Republic 13,9 Portugal 13,5 China 12,8 Vietnam 12,6 Indonesia 12,4 El Salvador 12,3 Mauritius 12,0 Cuba 11,9 Honduras 11,9 Thailand 11,2 St Lucia 10,9 Nicaragua 10,1 Ecuador 9,6 Haiti 9,5 Philippines 9,3 Spain 8,6 Iraq 8,5 Nepal 8,3 Venezuela 7,8 Mali 7,4 North Korea 6,3 Mexico 6,1 Egypt 5,4 India 5,2 0,0 20,0 40,0 60,0 ACMI (%)
Standardised Migration Intensities
Modes of explanation Culture of mobility (e.g. Long 1991) High mobility in new world countries reflects the peripatetic tradition of immigrant forbears A mobility transition (e.g. Zelinsky 1971, Skeldon 1997) definite patterned regularities in the growth of personal mobility through space-time associated with modernisation Life course transitions (e.g. Rogers and Castro 1981, Bernard et al 2014b) focus on the role of life course transitions and housing adjustment in triggering migration Measures of development: Geographic, Social, Economic, Demographic Variables Undertaken separately for 1 and 5 year data
Explaining cross-national differentials Explanatory Variables Geographic One-year ACMI Five-year ACMI n r p n r p Geographic area (Square root) 44 0.44 ** 60 0.15 Population density 44-0.10 59-0.11 Urbanisation 40 0.65 ** 60 0.39 ** Economic GDP per capita 40 0.67 ** 56 0.61 ** Gini coefficient (Income inequality 2000) 28 0.05 33-0.01 Foreign direct investment /GDP (2000) 43 0.04 55 0.01 Female labour force participation (2000) 43 0.53 ** 60 0.18 Labour force participation (2000) 42 0.39 * 60 0.22
Explaining cross-national differentials Explanatory Variables Social One-year ACMI Five-year ACMI n r p n r p Human development index (2000) 40 0.62 ** 58 0.48 ** Mobile phone subscribers (2000) 40 0.65 ** 60 0.54 ** Literacy (2000) 25-0.76 ** 48 0.06 Per cent males 20-24 living at home 11-0.87 ** 4-0.97 * Demographic Growth rate (2000-2005) 45 0.41 ** 59-0.25 E 0 (2000-2005) 45-0.03 60 0.25 Total Fertility Rate (TFR) (2000-2005) 40 0.44 ** 58-0.15 Median age 40 0.05 60 0.37 ** Net international migration rate (2000-2005) 40 0.35 * 55 0.48 ** Remittances as % of GDP (2000) 41-0.26 53-0.35 *
1 year migration intensity 5 year migration intensity Level of Urbanisation 20 18 16 14 12 10 8 6 4 2 Ireland Turkey Switzerland Greece Portugal Belarus Cyprus Romania Poland Croatia Slovakia Slovenia Macedonia Sweden Austria Hungary Lithuania Latvia Estonia Bulgaria Italy USA Germany Colombia UK Netherlands Spain Canada Japan Czech Republic Finland Australia Norway Denmark Belgium Iceland Israel Malta 0 40 50 60 70 80 90 100 Urban Population (%) 60 50 40 30 20 10 0 Senegal Switzerland Cameroon Paraguay Barbados Fiji Mongolia Panama New Zealand South Korea Canada Chile France Japan Israel Uganda Guinea Kyrgyz Republic Bolivia Morocco South Africa Peru Uruguay Tunisia Malta Cambodia Brazil Ghana Costa Rica Greece Malaysia Argentina Vietnam Guatemala Rwanda China Indonesia Portugal Dominican Republic Mauritius St Lucia Ecuador El Salvador Thailand Haiti Philippines Nicaragua Nepal Iraq Spain Mali Venezuela Honduras India Egypt Mexico USA Australia 0 20 40 60 80 100 Urban Population (%) r=0.65, n=40 r=0.39, n=60
1 year migration intensity 5 year migration intensity Human development index 20 18 16 Finland Iceland Australia 60 50 New Zealand South Korea 14 12 10 8 6 4 2 Turkey Colombia Greece Hungary Austria Japan Norway Sweden Canada Denmark UK Switzerland Belgium Germany Lithuania Israel Belarus Portugal Ireland Latvia Malta Estonia Bulgaria Cyprus Romania Poland Czech Republic Spain Croatia Slovakia Italy Slovenia Macedonia USA Netherlands 0 0,6 0,65 0,7 0,75 0,8 0,85 0,9 0,95 HDI 40 30 20 10 0 Senegal Cameroon Paraguay Japan Mongolia Barbados Israel Kyrgyz Republic Bolivia Guinea Peru Uruguay Morocco Malta South Africa Rwanda Tunisia Greece Cambodia Uganda Malaysia Brazil Costa Rica Guatemala Dominican Republic Argentina Ghana Indonesia Portugal Vietnam Mauritius Haiti Nicaragua China Thailand Honduras Ecuador Mali Philippines Spain Venezuela Iraq Nepal India Mexico Egypt Fiji Panama Chile USA Australia Canada Switzerland France 0 0,2 0,4 0,6 0,8 1 HDI r=0.57, n=38 r=0.48, n=58
Typical age Profile of Migration Bernard, A., Bell, M., & Charles-Edwards, E. (2014). Life-Course Transitions and the Age Profile of Internal Migration. Population and Development Review, 40(2), 231-239.
Country Variations in the Age Profile Bernard, A., Bell, M., & Charles-Edwards, E. (2014). Life-Course Transitions and the Age Profile of Internal Migration. Population and Development Review, 40(2), 231-239.
Comparative age profile metrics Existing approach Parameters of model schedules (Rogers and Castro, 1981) Issues related to their estimation (variability, sensitivity and instability) and interpretation (comparability and interpretability) Alternative indicators Normalised intensity at peak Age at peak + Overall migration intensity 2/3 of inter-country variance Bernard, A., Bell, M., & Charles-Edwards, E. (2014). Improved measures for the cross-national comparison of age profiles of internal migration. Population Studies, 68(2) 179-195
Regional migration profile clusters late and dispersed early and dispersed Three solution k-means clusters based on variables normalised to unit variance early and concentrated
Life-Course Transition Metrics Transitions measured from status by age Completion of higher education Entry into labour force Union formation Family formation 1,0 0,8 0,6 0,4 China Brazil France Metrics Prevalence (Modell et al. 1976) Proposition of a population that has experienced a transition at age 35 Timing (Hajnal 1953) Singulate mean age computed between ages 15 and 35 Spread (Carter and Glick 1970) Interquartile range 0,2 0,0 15 20 25 30 35 40 Marital status by age Bernard, A., Bell, M., & Charles-Edwards, E. (2014) Life-Course Transitions and the Age Profile of Internal Migration. Population and Development Review, 40(2), 213-240
Spread Timing Prevalence Summary Indices of the Life-Course Factor analysis, females Factor 1 Timing Index Factor 2 Spread Index Higher education 0.86 0.09 Union formation -0.70-0.46 Education completion 0.94 0.15 Union formation 0.92-0.01 Parenthood 0.85-0.36 Education completion 0.58 0.55 Union formation -0.10 0.91 Share of total variance 0.58 0.22 Bernard, A., Bell, M., & Charles-Edwards, E. (2014) Life-Course Transitions and the Age Profile of Internal Migration. Population and Development Review, 40(2), 213-240
Migration and life course transitions timing
Migration and life course transitions spread Brief transition
Conceptual Framework The Proximate Determinants of Internal Migration Age Patterns Underlying determinants Proximate determinants Migration outcome Context Social Economic Cultural Religious Demographic Life-course transitions Completion of education Entry into labour force Union formation Chilbearing Migration age patterns
Conclusions First global dataset on internal migration Wide diversity in data collection practice Problems of harmonisation statistical solutions Wide variation in migration intensities Positive association with development indicators Differences in age profile as well as intensity Links to life course transitions Other dimensions spatial impact; connectivity; distance decay Implications for data collection Implications for policy