Measuring poverty, hardship and living standards in New Zealand: a brief overview Bryan Perry Ministry of Social Development, NZ 2nd Peter Townsend Memorial Conference Bristol, January 2011 1 Introduction • New Zealand does not have an official measure of poverty or material hardship … • the Ministry of Social Development reports on: – income distribution and income poverty using a range of measures – material hardship and deprivation using non-monetary indicators • Statistics New Zealand sends NZ income distribution and income poverty information to the OECD • involvement by state agencies is relatively recent – Brian Easton – NZ Poverty Measurement Project Bristol Bryan Perry, Jan 2011 Introduction (cont’d) • MSD also publishes an annual ‘Social Report’ – includes income distribution and poverty information … – but it has a wider scope - 10 domains, 43 indicators • some limited use of ‘social inclusion’ / ‘social exclusion’ discourse … some on ‘social cohesion’ … more about ‘social and economic wellbeing’ • some use of an ‘underclass’ notion by current PM • the major data gap in NZ is for the dynamics of poverty and hardship … social mobility Bristol Bryan Perry, Jan 2011 Outline • active promotion of a common language and understanding around poverty and hardship • using HH incomes – relative and fixed (anchored) poverty lines – an illustration of the limitations of the incomes approach (older NZers) • using non-monetary indicators to construct more direct measures of material wellbeing – – – – the ELSI measure – its development, properties and value addressing some of ELSI’s limitations international comparisons next steps? Bristol Bryan Perry, Jan 2011 Active promotion of the big ideas • idea of ‘poverty’ in richer nations – we use the EU’s high level conceptualisation • income poverty & deprivation – different constructs • poverty and hardship are multi-dimensional … even more so if ‘causes’ and ‘consequences’ are included … but we try to keep an uncluttered focus on the goods, services and opportunities that money can buy • cannot avoid judgment in setting thresholds … but some thresholds are much more plausible / defensible than others Bristol Bryan Perry, Jan 2011 Income and living standards Living standards HH income Bristol Bryan Perry, Jan 2011 Govt services & subsidies Current income (eg last 12 months) + Access to other services & amenities + + expected future income Previous income + Assets Living standards + _ Contributions to budget not picked up in ‘current income’ eg HH production, help from outside the HH Special demands on budget eg unexpected bill health/disability costs high housing costs support for others o/s HH Bristol Bryan Perry, Jan 2011 Govt services & subsidies Current income (eg last 12 months) + Access to other services & amenities + + expected future income Previous income + Assets Living standards + _ Contributions to budget not picked up in ‘current income’ eg HH production, help from outside the HH preferences & wants? Special demands on budget eg unexpected bill health/disability costs high housing costs support for others o/s HH Bristol Bryan Perry, Jan 2011 Trends in HH incomes and income poverty, 1982 to 2009 Bristol Bryan Perry, Jan 2011 Median HH income ($2009), 1982-2009, NZ Equivalised HH income in $2009 (000's) 40 35 30 25 20 15 10 5 0 1980 82 84 86 88 90 92 94 96 HES Year Bristol 98 00 02 04 06 08 2010 Bryan Perry, Jan 2011 Income poverty rates, 1982 to 2009, NZ Proportion of population below threshold 35% 30% 60% REL 25% 20% 15% 10% 5% 0% 1980 82 84 86 88 90 92 94 96 98 00 02 04 06 08 2010 HES year Bristol Bryan Perry, Jan 2011 Income poverty rates, 1982 to 2009, NZ Proportion of population below threshold 35% 30% 60% 98 FIXED 60% REL 25% 20% 15% 10% 5% 0% 1980 82 84 86 88 90 92 94 96 98 00 02 04 06 08 2010 HES year Bristol Bryan Perry, Jan 2011 The limitations of relying on an incomes approach alone: the case older New Zealanders (65+) • NZ has a two-tier system of retirement income provision – NZ Superannuation - a universal pension, neither income nor asset tested – private provision by citizens – no earnings-related component • around half of older NZers have incomes of NZS plus less than $100 pw from other sources Bristol Bryan Perry, Jan 2011 Income poverty rates using different thresholds, by age group, 2008 70 Proportion below threshold (%) 65+ 60 under 18 total 50 18-64 40 30 20 10 0 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 Threshold as a proportion of the median (%) Bristol Bryan Perry, Jan 2011 Using NMIs: our general approach • information from individual items is useful … but need multiple items working together for a robust picture • we don’t directly use the ‘socially defined necessities’ approach with, say, a 3+ lack defining ‘deprivation’ • we use the ‘enforced lack’ notion … … but are very conscious of the impact of ‘adaptive preferences’ • deprivation (and material living standards more generally) is conceptualised as a latent variable reflected in the pattern of association between a number of observable indicators Bristol Bryan Perry, Jan 2011 Using NMIs: our general approach (cont’d) • the indicators or survey items cover the ‘necessities’, ‘desirables’, and ‘(common) luxuries’ categories – development of a full range living stds index (low to high) – reality of consumption decisions within HHs – even the relatively poor have some ‘desirables’, even a few ‘luxuries’ • no explicit attempt to derive an income poverty line using NMIs … – but we use the steepening of the gradient of the core necessities vs income curves to provide credibility for the thresholds we do use Bristol Bryan Perry, Jan 2011 Can the ‘enforced lack’ NMI approach to measuring hardship be extended to cover a wider range of living standards, from low to high? Bristol Bryan Perry, Jan 2011 Extending the enforced lack NMI approach to cover a wider range of living standards? • the development goal • candidate items to use in the extension • the model and the model fit (CFA) • a user-friendly version (ELSI and ELSI levels) • validation • point in time findings using ELSI • a closer look at the underlying conceptualisation of living standards for ELSI • implications of this for measuring changes over time • next steps Bristol Bryan Perry, Jan 2011 Stylised summary of development goal Cumulative percentage (%) 100 mainly necessities, a few desirables 80 60 40 20 0 0 10 material hardship 20 30 40 50 60 70 80 from 'just getting by' to very good living standards Bristol Bryan Perry, Jan 2011 Stylised summary of development goal Cumulative percentage (%) 100 80 mainly necessities, a few desirables 60 more desirables and 'luxuries' added the development goal 40 20 Non-linear ordinal scale, running from low to high material living standards 0 0 10 material hardship 20 30 40 50 60 70 80 from 'just getting by' to very good living standards Bristol Bryan Perry, Jan 2011 Wider range of items • in addition to the usual ‘necessities’ and ‘near necessities’, we needed some items that generally reflected higher living standards • tried various consumer durables … such as a dishwasher (too much variation across groups ...) • overseas holiday once each 3 years (ie no EL) • economising items (‘not at all’) eg – cut back on local trips to visit friends and relatives – put off buying new clothes as long as possible – bought cheaper or less meat than wanted to • self-rated material standard of living Bristol Bryan Perry, Jan 2011 The first model Ownership ELs Participation ELs Economising Material living standards Financial hardship Housing problems SR material LS SR income adequacy Bristol Bryan Perry, Jan 2011 The first model Ownership ELs Participation ELs Economising Material living standards Financial hardship Housing problems SR material LS SR income adequacy Bristol Bryan Perry, Jan 2011 The current model Ownership ELs Participation ELs Added to stretch the top end more Economising Material living standards SR material LS SR satisfaction with LS SR income adequacy Bristol Bryan Perry, Jan 2011 Extending the NMI deprivation approach …. • CFA analysis for a five factor model reflecting a single latent variable - model fit very good • reliability also very good (Cronbach of 0.79) • we created a user-friendly version of the scale, with scores ranging from 0 (low) to 60 (high living standards) – correlation of 0.98 with generic scale – ‘ELSI’ (the Economic Living Standards Index) – for presentation purposes, created 7 ‘levels’ from very low to very high living standards – Levels 1 and 2 – clear hardship zone – Level 3 – hardship? Bristol Bryan Perry, Jan 2011 Distribution of ELSI scores by ELSI level whole population, 2008 30 26 Population percentage 23 20 18 10 10 9 7 6 0 ELSI levels (1-7) Bristol Bryan Perry, Jan 2011 Validity? • to what degree can we trust an instrument to be measuring what it claims it measures? • requires a careful marshalling of evidence from different sources – cumulative, never ‘final’ ********* - items … close examination - not covered today - internal structure - CFA and reliability - relationship with other variables - next section - consequences in practice - later - re-examination of underlying construct - later Bristol Bryan Perry, Jan 2011 Non-ELSI items: negative association with ELSI 60 Population percentage 52 40 33 31 23 23 20 18 13 9 7 4 3 0 0 0 0 0 3 0 0 0 0 0 meal with meat, chicken or could not pay utilities on help with food, clothes or fish at least each 2nd day time in last 12 mnths [MORE money from NGO welfare [EL] than ONCE] agency in last 12 mnths [MORE than ONCE] Bristol Bryan Perry, Jan 2011 Non-ELSI items: negative association with ELSI 80 75 Population percentage 68 60 55 53 42 40 37 32 28 20 26 20 16 15 12 9 5 0 8 4 0 books, music CDs or downloads [cut back / went without A LOT] 2 crime / vandalism in the neighbourhood [MAJOR problem] Bristol 14 2 quality of accommodation paid for on holidays [N/A] Bryan Perry, Jan 2011 Non-ELSI items: positive association with ELSI 98 100 94 93 89 83 Population percentage 78 78 80 75 67 63 60 56 54 52 49 47 42 41 39 40 29 25 20 17 0 current health self-rating [VG / EX] overall condition of home (GOOD / VG] Bristol overall satisfaction with life [SATISFIED / V SATIS] Bryan Perry, Jan 2011 Non-ELSI items: positive association with ELSI 99 100 Population percentage 89 80 72 60 72 71 57 53 41 38 40 41 30 24 23 20 16 15 11 11 5 0 0 0 2 delayed repair / spot purchase of nonquality of kitchen and replacement of appliances essential but desirable $250 kitchen appliances [TOP of not working [NOT AT ALL] item [NO RESTRICTION] RANGE / GOOD] Bristol Bryan Perry, Jan 2011 Using ELSI for point in time comparisons Bristol Bryan Perry, Jan 2011 ELSI levels by age group, 2008 50 46 Population percentage 40 31 30 27 25 23 23 23 22 22 22 20 26 20 17 15 14 14 12 11 10 10 10 10 9 9 7 8 6 8 8 6 6 6 2 2 2 3 0 0-17 years 18-24 years 25-44 years Bristol 45-64 years 65+ years Bryan Perry, Jan 2011 ELSI levels by family income source, 2008 50 46 Population percentage 40 30 26 25 26 23 21 20 18 20 19 14 10 10 10 10 10 6 5 4 2 0 2 3 0 Income-tested benefit Market <65 Bristol 65+ Bryan Perry, Jan 2011 ELSI levels by family type (under 65), 2008 40 Population percentage 34 30 24 25 26 25 23 22 22 21 20 20 19 15 14 13 13 11 10 8 9 8 8 6 3 8 7 7 4 3 4 0 One person Couple only Bristol Sole parent Two parent Bryan Perry, Jan 2011 ELSI levels by tenure, 2008 50 44 Population percentage 40 32 31 30 27 26 25 24 23 21 21 21 20 20 19 16 14 13 10 10 16 13 12 10 9 7 9 6 6 5 4 3 4 4 4 1 1 1 0 Owned with mortgage Owned mortgage free Family trust Bristol Private Landlord HNZC Bryan Perry, Jan 2011 Children’s items by family ELSI level, 2008 Proportion of children (%) 60 44 42 40 40 38 23 20 16 13 11 9 9 5 4 3 0 13 0 0 0 0 0 0 4 1 0 0 0 4 0 0 2 pr strong / sturdy a waterproof coat for have children's children went without shoes for each child each child [EL] friends to a birthday music, art, dance, [EL] party [EL] swimming lessons [A LOT] Bristol Bryan Perry, Jan 2011 Children’s items by family ELSI level, 2008 Proportion of children (%) 80 61 60 47 40 40 39 33 31 26 20 19 17 14 12 12 9 7 3 1 0 house difficult to keep warm in winter [MAJOR problem] dampness or mould [MAJOR problem] Bristol 4 1 4 0 0 enough bedrooms for children of opposite sex (aged 10+) to be not sharing [EL] Bryan Perry, Jan 2011 Different worlds … 75 Proportion of children (%) 62 60 50 47 33 33 32 25 15 12 12 7 6 2 0 0 1 1 2 4 borrowed money from family spot purchase of nonand friends for everyday essential but desirable $250 needs [MORE than ONCE item [NO RESTRICTION] in last 12 mnths] Bristol 1 1 3 3 employed cleaner on a regular basis Bryan Perry, Jan 2011 Children: cumulative disadvantage • ELs – – – – – friends to birthday party waterproof coat separate bed separate bedrooms for children of opp sex (10+) all school uniform required • Economising ‘a lot’ – – – – – – continued with worn out clothes/shoes postponed visit to doctor did not pick up prescription unable to pay for school trip went without music, dance, kapa haka, art, swimming, etc involvement in sport had to be limited lower to higher LS Distribn of children across the LS levels (%) 12 10 18 18 41 % with 4+ deps 35 11 2 0 0 Bristol Bryan Perry, Jan 2011 ELSI is very useful as is, but it is still a work in progress • Respondent burden – 40 items – have a ‘short-form’ of 25 items, but need to reduce further • Compression in upper half of the scale – need for more items for modest to high living standards – potential with some ‘quality’ items, two new economising items, and one about ‘no restrictions’ for making an ‘unplanned purchase’ of around GBP125 – would like to reduce reliance on global self-ratings • Underlying conceptualisation of living standards – what does ELSI actually measure? Bristol Bryan Perry, Jan 2011 What does ELSI actually measure? Representative list of basics and non-basics (O & P) Don’t have Score Have Want, don’t have because of cost (= enforced lack) Want, don’t have for other reasons Don’t want 1 0 1 1 • ELSI scores reflect the degree of financial restriction (freedom) that respondents report about consumption relative to desired consumption from a list of basics and non-basics … • … that have been shown to reflect a common underlying or latent variable Bristol Bryan Perry, Jan 2011 Implications for monitoring changes in material living standards over time • ‘Our living standards are higher today than 20 years ago’ usually means that consumption is higher / more people have access to more of the goods and services considered out of reach 20 years ago. • ELSI is about changes in consumption relative to desired consumption, not changes in consumption per se • If expectations about consumption rise in step with rising living standards, then the average ELSI score is likely to show little or no change over time. Bristol Bryan Perry, Jan 2011 Implications for monitoring changes in material living standards over time (cont’d) • This is what we found from 2000 to 2004 to 2008 – no change in avg ELSI scores • The way people respond to global self-ratings of living standards is consistent with ELSI’s ‘relative to desired consumption’ conceptualisation – viz if a person’s living standards rise in rough step with those of their ‘reference group’ then their self-rating will remain much the same • The ELSI conceptualisation of living standards is internally consistent … but there is a marketing challenge ! Bristol Bryan Perry, Jan 2011 FRILS? • We have experimented with a ‘Fixed Reference Index of Living Standards’ (FRILS) – O and P items: have it / don’t have it – E items: economise a lot ≡ don’t have – SR items: omitted Representative list of O, P and E basics and non-basics Don’t have Have Want, don’t have because of cost (= enforced lack) Want, don’t have for other reasons Don’t want ELSI 1 0 1 1 FRILS 1 0 0 0 Bristol Bryan Perry, Jan 2011 ELSI and FRILS compared: point-in-time rankings • rankings for the usual population groups are much the same for ELSI and FRILS • main exception is for those aged 65+, esp singles … FRILS produces similar hardship rates but a much lower proportion with higher living standards • ‘adaptive preferences’ for the 65+, esp the singles? compared with younger adults they report fewer ELs for a given number of lacks, and also satisfaction with less: – to maintain dignity and/or sanity by saying they’re OK even when just getting by and ‘overstating’ self-ratings? or – having found that that is a ‘better’ way to live? Bristol Bryan Perry, Jan 2011 ELSI & FRILS compared: hardship rates ELSI FRILS 13 13 0-17 19 19 18-24 11 11 25-44 14 13 45-64 12 11 65+ 4 3 European 10 9 Maori 26 27 Pacific 33 31 Other 14 14 Overall Age group Ethnicity Bristol Bryan Perry, Jan 2011 Avg ELSI & FRILS scores compared difference from overall mean 10 ELSI 5 FRILS 0 ben with dep sole parent two parent single <65 couple <65 single 65+ couple 65+ -5 -10 -15 -20 Bristol Bryan Perry, Jan 2011 ELSI & FRILS compared: ‘doing OK’ rates ELSI FRILS 50 51 0-17 36 43 18-24 45 46 25-44 47 51 45-64 62 63 65+ 71 50 Overall Age group Ethnicity European Maori Pacific Other Bristol Bryan Perry, Jan 2011 ELSI & FRILS compared: ‘doing OK’ rates ELSI FRILS 50 51 0-17 36 43 18-24 45 46 25-44 47 51 45-64 62 63 65+ 71 50 Overall Age group Ethnicity European Maori Pacific Other Bristol Bryan Perry, Jan 2011 ELSI and FRILS compared: tracking changes over time • major difference for 2000 to 2004 to 2008 – no change in ELSI mean – noticeable rise in FRILS mean • using FRILS – there was an increasing proportion with higher scores (better living standards) … not so using ELSI % with higher LS index: 2000 = 100 ELSI FRILS ELSI FRILS 2000 31 29 100 100 2004 32 35 104 120 2008 32 41 104 140 Bristol Bryan Perry, Jan 2011 The current model Ownership ELs Participation ELs Economising Material living standards SR material LS SR satisfaction with LS SR income adequacy Bristol Bryan Perry, Jan 2011 Possible next model Ownership ELs Participation ELs Economising – few new ones Material living standards SR material LS Quality of kitchen, furniture, other durables ? Ability to purchase nonnecessities ‘at will’ Bristol Bryan Perry, Jan 2011 International comparisons • Return to deprivation indices • Great value in having an internationally comparative NMI scale – EU 9 items, with all its limitations Bristol Bryan Perry, Jan 2011 Hardship rate for population, EU index EU + NO, IS and NZ, 2008 Proportion with 3+ enforecd lacks (%) 60 50 40 30 NZ 20 10 0 LU NO NL SE IS DK AT FI UK IE ES BE FR DE NZ SI Bristol IT EE CZ PT GR CY SK HU LT PL LV Bryan Perry, Jan 2011 Hardship rate for those aged 65+, EU index EU + NO, IS and NZ, 2008 Proportion with 3+ enforecd lacks (%) 60 50 40 30 20 NZ 10 0 LU NO NL SE NZ DK IE IS UK DE FI FR AT BE ES IT CZ SI EE PT GR HU LT PL SK CY LV Bristol Bryan Perry, Jan 2011 Hardship rate for children (0-17), EU index EU + NO, IS and NZ, 2008 Proportion with 3+ enforecd lacks (%) 60 50 40 30 NZ 20 10 0 LU NL NO SE DK ES FI IS AT DE SI EE IE BE FR UK MT NZ IT CZ GR PT CY LT SK PL HU Bristol Bryan Perry, Jan 2011 1.8 This quadrant: above median deprivation rate and above median risk ratio 1.6 Risk ratio for children UK 1.4 IE SE LU BE NL ES DE IS 1.2 CZ AT DK NO NZ FR IT EE FI CY 1.0 SK PT GR LT SI 0.8 0.6 0 5 10 15 20 25 30 35 Deprivation rate for children (%) Bristol Bryan Perry, Jan 2011 Next steps • further index development – improving ELSI for middle to high living standards – more thorough development of FRILS – deprivation indices – dimensions – understanding drivers of low living standards? • revised item list for SNZ surveys using the 25 item budget • an official set of measures of poverty and hardship for NZ Bristol Bryan Perry, Jan 2011