Principles and practicalities in measuring child poverty Miles Corak Measuring child benefits: measu

1
Principles and practicalities in measuring child
poverty - Miles CorakMeasuring child benefits
measuring child poverty - Michael Mendelson Sid
Frankel

• Lars Osberg
• Economics Department
• Dalhousie University
• Low Income, Poverty, and Deprivation
• A Workshop sponsored and hosted by
• Statistics Canada and the University of Toronto
• June 5/6 2006

2
Different Focus Agreements, Disagreements
Omissions

• Agreements
• Poverty is relative to social norms of
  deprivation
• Focus
• Identify Poor Administration or Evaluation ?
• Type I Type II errors of classification
• Disagreements
• Market Basket or of Median Income ?
• Word Picture to motivate criterion ?
• Summarization rate gap ?
• Omissions
• Time Risk, Human Rights Link

3
Poverty economic deprivation - want of the
necessities of life

• not only those things which nature, but those
  things which the established rules of decency
  have rendered necessary to the lowest rank of
  people
• Adam Smith (1776)
• poverty is an absolute notion in the space of
  capabilities but very often it will take a
  relative form in the space of commodities or
  characteristics
• Amartya Sen (1983)
• MM MC agree

4
Capabilitiest Fit (commoditiest)

• Poor - If Capabilities lt Minimum Necessary
• Social Standards of time gt Minimum _at_ time t
• Fit F (household size composition, location,
  health, disability, life skills, etc.)
• Commodities alternative measures
• Actual consumption or potential consumption
  (income) or specific necessities
• Minimum Capabilities / Deprivation ??
• Ambiguity of citizenship

5
Issue - to identify the poor or the number
deprivation of the poor ?

• MM What is an adequate maximum child benefit
  for families with low incomes?
• MMs issue administration of a particular
  program Canada Child Tax Benefit - implies
  specific information set (Xi)
• Type I and Type II errors both matter
• Good Predictor of need at individual level
  needed for effectiveness
• Other Programs supposed to take care of
  special needs, adults

6
MC aim to define measure child poverty,
compare internationally define credible poverty
targets for society

• Issue given all programs, macro cycle,
  demography, etc., how big is the problem?
• Rate, average gap inequality of gaps
• Note If Type I error Type II error, total is
  OK
• Identification of deprivation of specific person
  not the issue
• Statistical agency information set ?
  Administrative sets
• Different criteria for success
• MC objective to Minimize Type I Type II
• MM objective to Minimize Type I Type II
• MM Much discussion of classification errors

7
Practical Issues in Poverty Measurement

• MM MC agree on dimensionality
• accounting period year,
• recipient unit family / household
• measure after tax transfer money income
• supplementation by specific item deprivation
• MM MC disagree on poverty line
• Relative Poverty or Budget based / Absolute
  ?
• Implications for Initial Level and for Updating
• LO Missing Issues
• Summarization
• Spell Durations
• Human Rights link

8
MM1 Market Basket Budget

• MM all relevant measures of low income or
  poverty are relative to contemporary living
  standards
• Historical evidence - absolute poverty lines
  market basket budgets increase over time with
  average incomes
• Non-transparent but appearance of objectivity?
• Details of construction buried deep
• wide range e.g. 18,342 to 30,402 for 4 in
  Toronto 1992
• Major political plus iff source is credible
• Major minus if change is suspiciously timed
• Implicit link to norms, occasional, idiosyncratic
• Not directly comparable across nations, time

9
MM2 reverse regression ???

• MM proposes regressing the sample of families,
  with income as the dependent variable and the
  deprivation indicators as the independent
  variables, for families of different structures
• Yi a0 a1D1 a2D2 ß1X1 ß2X2 ei
• But complains With current techniques, we know
  only that renters have lower incomes than owners
• Better estimate Prob(D1,D2 Yi ,X1,, X2 )
• BUT no escape from arbitrary choice of minimum,
  necessary Probability of (D1,D2)

10
MC Explicitly Relative Poverty Line

• Ho Established rules of decency change as
  society changes
• Transparent link to current distribution of
  income
• LIS standard ½ median equivalent income
• European Union 60 median equivalent income
• Explicitly transparent, standardized, comparable
  over time across countries
• huge plus for policy evaluation
• Two issues initial level updating
• Choice ½ or 60 - ?
• Rapid change in median can imply
  counter-intuitive changes
• Update for price changes OR Update for median
  real income changes ?
• Moving average can diminish volatility

11
When does the choice of updating method matter ?

• Polling data since 1946 - social norms track
  median earnings
• IF median real incomes change, updating method
  matters sometimes has been historically
  important
• Canada 1960s, 1970s, USA in late 1990s, Ireland
  1990s
• BUT since 1981, not much change in median real
  incomes in Canada

12
Absolute or Relative Poverty Line ? Not much
change in real median incomes !
13
How best to motivate choice of initial level of
poverty line ?

• MM general word picture of standard of living
  at poverty line should drive number
• How to change specific items over time?
• How to defend general criteria as non-arbitrary?
• MC ratio of potential consumption to median
  household income implies number
• less than half deprivation
• LO ½ has intuitive meaning
• 50/50 equity norm ( solution to ultimatum
  game)
• Individuals always apply Reality check
• number drives examples, which drive credibility
• Comparisons drive Concern
• Over time is poverty getting worse ?
• Across countries do other nations do better ?

14
Item Deprivations An Increasingly Important
Complement to povline

• Communicable to Public
• Human Rights Perspective gt specific goods
• Convention on the Rights of the Child - includes
  health and nutrition, clothing, housing, and
  other goods, services and opportunities necessary
  for normal physical, mental and social
  development
• Can see behind within family allocation of
  resources
• to children, elderly, women.
• Legal recognition of social norm of minimum
• literature now emphasizes multi-dimensionality
• Sens Capabilities Perspective
• Social Exclusion Agenda
• Social Wage or Cash purchase ? who cares?
• Comparable across differing public-private policy
  mix

15
LICO orphan methodology

• Relatively little discretionary income
• LICO based on 20 E (food clothing house
  share of expenditure income, family size,
  rural/urban)
• Year to year updating on CPI as per absolute
• Rebased occasionally to community norms as per
  relative
• Long time series useful !!!
• for research politics
• BIG MISTAKE to terminate until any replacement
  methodology has acquired credibility

16
MC - Principles for Measurement

• (1) use income based measure of resources
• (2) complement with small set of specific
  indicators
• (3) draw poverty lines with regard to social
  norms
• (4) establish a regular monitoring system and
  update poverty lines within a five year period
• (5) set a both a backstop and a target by using
  fixed and moving poverty lines
• (4) (5) less important if real growth in median
  is small
• (6) offer leadership and build public support for
  poverty reduction.
• Explicit Link to Human Rights Agenda would be
  useful

17
LO1 Annual Income ?

• Year compromise accounting period
• immediate needs are much more short term
• long term impacts may be multi year
• Implicit Assumption income is perfectly
  substitutable within period but zero between
  period substitutability
• Consumption Income
• smooth consumption despite low income ?
• Borrow?
• Beg from relatives ??
• BUT credit access for most low income households
  is very poor families may be poor /or alienated

18
Why does poverty duration matter?

• long-term poor are materially worse off than
  short-term poor
• Income-based measures imperfect indicator of
  resources
• Low current income but buffers access to
  credit, run down consumer durables, obtain
  transfers
• long spells exhaust buffers to consumption
• Psychological/Social impacts may increase with
• individual duration
• Social stereotypes may depend on the distribution
  of durations
• High turnover/short spell process means more of
  nonpoor have been there

19
Why does immediate need matter ?

• No groceries no credit no income immediate
  need in a week or less
• Transitory child poverty can imply events
• Events can have long run impacts
• Eviction from housing ?
• Forced mobility school attainment ?
• Family stress gt Family breakup? Abuse ?
• Traditional Welfare State agencies recognized
• But new HK paradigm has firmly long run
  perspective
• Much higher incidence (3x in SIPP) than in annual
  poverty measures

20
(No Transcript)
21
LO2 Poverty Gap matters too !!Why not measure
Poverty Intensity ?

• Sen-Shorrocks-Thon (SST) Index.
• P (RATE ) (GAP ) (1G (X )).
• Empirically inequality of poverty gaps 1G
  (X) is very nearly constant.
• SST volume of Box, one dimension nearly
  constant.
• Changes over time, differences across
  jurisdictions can be approximated by.
• RATE x GAP FGT index (a 1)
• Poverty Box in 2 dimensions.
• Poverty Intensity proportional to area RATE X
  GAP.

22
(No Transcript)
23
Variation over time mostly in rate, but misses
recent trend in poverty gap
24
Qualitative Conclusions Re Poverty Trends Often
Differ if consider gap

• Relative poverty in early 1990s ?
• Canada, UK, Sweden poverty rate gap disagree
• US Germany indices agree
• UK Absolute poverty line - poverty rate and
  poverty intensity change in opposite directions
  every time - 1974/79/86/91/95

25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
Communicability ?

• The whole point of measuring poverty is to
  influence the policies that might affect poverty
• A poverty index not used or understood by the
  public by policy makers is pointless
• Humans not very good at deriving accurate
  relative size information from large tables of
  numbers.
• Quite good when presented as area proportionate
  graphs

31
Equivalence scale - a matters