Poverty in Canada: Unidimensional and Multidimensional Measures. Presented by: Lori J Curtis, PhD Department of Economics

Transcription

1 Poverty in Canada: Unidimensional and Multidimensional Measures Presented by: Lori J Curtis, PhD Department of Economics IARIW-BOK Conference, Seoul South Korea, April

2 Background Unidimensional Measures of Poverty Can a single measure, such as GDP/capita, income or expenditure, adequately represent the experience of poverty (Atkinson, 2003; Sen, 2006)? Multidimensional Measures of Poverty Is aggregating multiple measures of attainments or deprivations into a single multidimensional measure better? subjective measures Weights Cut-offs PAGE 2

3 Background GDP measures have been used as an indication of the economic well-being (UNDP, 2016). viable indicator of aggregate economic growth limited as a description of the experience of wellbeing household production or the underground economy? health, leisure, environment, political freedom, or social justice? Income/expenditure historically commonly used measure in micro-level poverty studies. household production? health, leisure, environment, political freedom, or social justice? PAGE 3

4 Background Multidimensional measures of well-being/poverty gaining traction. In developing countries Multidimensional Poverty Index (MPI) (Alkire & Santos, 2010) Human Development Index (UNDP, 2010) Beginning in developed countries European Union (Alkire et al., 2014, Whelan et al 2014). US (Mitra & Brucker, 2014; Dhongde & Haveman, 2015). Canada (UNDP, 2016; Canada Index of Wellbeing, 2016) PAGE 4

5 Background PRESENTATION TITLE PAGE 5

6 Background PRESENTATION TITLE PAGE 6

7 Background PRESENTATION TITLE PAGE 7

8 Background Some countries have developed survey instruments for MDPI (see for example the Global MDPI project at Some have high quality longitudinal survey data to develop the instruments EU-SILC data in the European Union (Alkire et al., 2014, Whelan et al., 2014) Germany SOEP (Suppa, 2016) Others parse together indicators using available data American Community Survey (Dhongde & Haveman, 2015) Current Population Survey and American Community Survey data (Mitra & Brucker, 2014) PAGE 8

9 Data This study compares constructed MDP measures with traditional measures of poverty using multiple nationally representative data sources (pumf files) Cansim tables and are used to calculate the growth in GDP/capita over time. ( Survey of Household Spending (SHS) 2003, 2005, 2007, and 2009 Canadian Community Health Survey (CCHS) , 2005, 2007, 2009, and 2012 Waiting for access to Master files of SHS and CCHS PAGE 9

10 Methods Growth GDP/capita is % change in annual expenditure based real GDP/population (chain-linked). SHS Income and expenditure measures are estimated per Crossley and Curtis, Households are considered poor if they have adjusted income or expenditure 0.5*median. MDP dimensions, indicators, and thresholds a la Dhongde & Haveman s (2015) CCHS indicators are reported for households Statistics Canada weights are multiplied by household size poverty head counts. Paper presents sensitivity analysis of many measures here present growth in GDP/capita, income and expenditure from SHS, Multidimensional from CCHS PAGE 10

11 Multidimensional Poverty Dimensions and Indicators Dimensions Indicators No Regular MD Household does not have a regular Health family physician Disability A household member is limited in activities by a disability No High School Highest adult educational level in household is less than high school Education certificate No Eng/Fr Speaker No one in household speaks an official language No Employment Major source of household income is not employment income Standard of Living Low Income Household income (adjusted) is in the bottom quintile of the income distribution. Do Not Own Home Housing Crowded Food Insecure The house is not owned by a member of the household There is not a bedroom for parent(s) and each child Food Insecure Index

12 Sensitivity Mutlidimensional Poverty measure - CCHS MDP1 = No MD + Disability + No High school + No Eng/Fr + No employ income+ low income + crowded + house not owned MDP3 = No MD + Disability + No High school + No Eng/Fr + No employ income+ low income + food insecure + house not owned MDP 5 = MDPH without language (SHS comparison) MDP 7 = MDP H dropping PEI and NB (not available in 2009) MDP2 = MDP1 without home ownership MDP 4 = MDP 3 without home ownership MDP 6 = MDP 5 without home ownership MDP 8 = MDP 7 without home ownership MDP(H) = MDP1 (03 07) and MDP3 (09-12) MDP(NH) = MDP2 (03 07) and MDP4 (09-12)

13 Indicators in CCHS Proportion of Individuals with stated Indicator in CCHS No MD Disability Health No High school No Language No Employ Income Low Income Education Standard of Living No Own Home Crowded Food Insecure Housing *Canadian respondents 25 to 64 years of age (2009 where PEI and NB are excluded see text for explanation)

14 Tetrachoric Correlations between Indicators CCHS 2007 No Regular MD Disability No High School No Empl Inc Low Income Not Own Home Crowd Food Insecure No English Speaker No Regular MD Disability No High School No Employment Income Low Income Does Not Own Home Crowded (<1 rm/person) Food Insecure No English Speaker

15 Correlation Matrix Multidimensional Poverty Counts CCHS 2007 MDP1 Count MDP2 Count MDP3 Count MDP4 Count MDP5 Count MDP6 Count MDP7 Count MDP8 Count MDP1 Count MDP2 Count MDP3 Count MDP4 Count MDP5 Count MDP6 Count MDP7 Count MDP8 Count

16 Tetrachoric Correlations Multidimensional Poverty Score 2 CCHS 2007 MDP1 Count 2 MDP2 Count 2 MDP3 Count 2 MDP4 Count 2 MDP5 Count 2 MDP6 Count 2 MDP7 Count 2 MDP8 Count 2 MDP1 Count 2 MDP2 Count 2 MDP3 Count 2 MDP4 Count 2 MDP5 Count 2 MDP6 Count 2 MDP7 Count 2 MDP8 Count

17 Proportion of Population Sensitivity Analysis MDP H MDP NH MDPC 1 MDPC 2 MDPC 3 MDPC 4 MDPC 5 MDPC

18 Proportion of Population 0.25 Poverty Measures with and without NB and PEI MDP H MDP NH MDPsm H MDPsm NH Economics

19 Proportion of Population 0.25 Well being/poverty Measures GDP/cap Income Expend MDP H MDP NH

20 Percent of Population 60 MDP Counts with Housing CCHS mdp1u mdp1i mdp 2

21 Percent of Population 70 MDP Counts without Housing CCHS mdp2u mdp2i MDP

22 Multidimensional Poverty Index, MPI (M 0 )* M 0 (H) M 0 (NH) *weighted average deprivations experienced by the multidimensional poor.

23 Atlantic QUE ONT Prairies BC Atlantic QUE ONT Prairies BC Atlantic QUE ONT Prairies BC Atlantic QUE ONT Prairies BC Ranking 5 Regional Ranking by Measure GDP Income Exp MDP H MDP NH

24 Income, Expenditure & Multidimensional Poverty by Family Type Inc Exp MDP H MDP NH Inc Exp MDP H MDP NH Inc Exp MDP H MDP NH Inc Exp MDP H MDP NH Single Couple Two Parent Lone Parent

25 Multidimensional Poverty by Immigrant Status - CCHS H NH H NH H NH H NH H NH Native Born Immigrant < 10 yrs Immigrant 10 yrs

26 Conclusions. Individual Indicators are not highly correlated MDP measures are highly correlated Surveys provide different levels for indicators and thus poverty MDP higher head counts than income and expenditure (generally) Income higher levels of poverty than expenditure Generally uni and multidimensional trends not very similar Ranking changes at provincial level Subgroup analyses provide similar trends GDP growth most inconsistent with the other measures

27 Conclusions.???? Right measure??? Assumption seems to be MDP best captures experience but??? Policy No official poverty measure in Canada Who decides which measure is best? Who decides which indicators are important? If poverty alleviation important goal need consistent measure Then need consistent data LFS legislated to provide consistent data over time for Urate etc. Could we do the same for poverty measures?

28 Thank You Comments Appreciated