Guaranteed Income Supplement (GIS) Status Amongst the Retired Population

1
Guaranteed Income Supplement (GIS) Status Amongst
the Retired Population An Analysis of the
Incidence and the Dynamics

• Ross Finnie
• David Gray
• University of Ottawa
• Yan Zhang
• September 2011

2
We gratefully acknowledge the generous financial
assistance provided by the Canadian Labour Market
and Skills Research Network and Human Resources
and Skills Development Canada.

• We have benefitted from suggestions from Tammy
  Schirle, Kevin Milligan, and Alex Grey

3
Homage to Charlie!
4
Policy context

• The head of the baby-boom is retiring
• into a hodge-podge of public and private pension
  schemes
• Both types are under financial duress
• Nonetheless, official poverty rates among senior
  citizens are relatively low
• We are most concerned with adequacy of resources
  of retirees

5
GIS program

• Last layer of the social safety net for senior
  citizens
• Demogrant
• Totally means-tested
• Sharp clawback mechanism, giving rise to strong
  work disincentives
• Benefits are modest, but overall expenditures
  high
• 652 / month for singles, 862 / month for
  couples
• 6.8 billion in 2006

6
GIS program (cont.)

• Given the requirement to renew annually and the
  very strict means testing, one can expect
  transitions across the threshold of eligibility

7
Outcomes of interest

• Receipt of GIS benefit during a calendar year
• Dynamics of GIS receipt
• Transitions into GIS receipt status after the
  youngest age of eligibility
• late retirement
• Exhaustion of savings
• Loss of spouse
• Transitions out of GIS receipt status (not nearly
  as common)
• Re-entry into the labour force
• Pension fund withdrawals
• Gain of spouse

8
Objectives

• Calculate descriptive statistics pertaining to
  the dynamics
• Incidence of GIS
• Descriptive analysis with cross-tabs
• Multi-variate models
• Include proxies of retrospective permanent income
  received at peak of earnings profile

9
Key role for permanent income

• To the extent that workers are able and willing
  to save for retirement, measures of permanent
  income during their working lives should have
  high explanatory power
• Ideal policy scenario
• Most of those relying on GIS in retirement would
  have been low-income workers

10
Potential influences other than permanent income

• Failure to save adequately for retirement
• Myopia
• low financial literacy
• Subjective rate of time preference
• Bad luck and/or flawed investment strategy
• Negative shocks to earnings in final third of
  career
• Disability
• Divorce / death of spouse
• Family responsibilities (undue expenses)

11
antecedents

• Abbott et al. (2008)
• policy issues surrounding retirement
• especially adequacy of old-age benefits
• Milligan (2005), Milligan and Schirle (2008),
  Baker et al. (2003)
• focus on labour market disincentives for the
  public pension schemes
• Luong (2009) and Poon (2005)
• deal with take-up and application rates for GIS

12
Antecedents (cont.)

• Uppal et al. (2009) is closest to our work
• Pathways into the GIS
• Investigate incidence of GIS receipt and its
  determinants
• They include proxies for permanent income and
  changes in income leading up to retirement
• We model receipt at annual frequency with
  repeated observations
• We include certain contemporaneous traits

13
Data

• Longitudinal administrative data base (LAD) 20
  file
• until 2008
• GIS receipt reported at individual level.
• but entitlement determined at couple level
• Precise record is net federal supplements GIS
  or spouses allowance
• Not reported separately before 1992
• we can observe incomes back to 1982
• Certain reporting lags exist
• receipt for 65-year olds might not be well
  measured
• Structure of the cohorts for regression analysis

14
Year in which subject turned 65 (first year for receipt of GIS) Year in which subject turned 50 (for measuring permanent income)
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
15
Empirical approach for incidence

• Follow from age 65
• Unit of observation is person-year
• Follow some subjects well into their 80s
• Binary dependent variable
• LPM and logit specification
• Avoid including regressors that are highly
  correlated with contemporaneous measures of
  income
• would generate mechanical, tautological effects

16
Empirical approach for incidence (cont.)

• Seek proxy for permanent income that is not
  highly correlated with current income
• Market Income when 50-52 years old
• 13 years removed from the minimum age of
  eligibility of 65 years,
• Even longer elapsed time period for older
  subjects

17
Empirical approach for incidence (cont.)

• Regressors divided into thee types
• Year effects
• Demographics
• Cohort year
• Age
• Gender crossed with marital status
• Province
• SMA size
• Residency status (binary)
• Immigrant status (how long in Canada?)

18
Empirical approach for incidence (cont.)

• income-related variables at age 50-52
• Market income
• Flags for
• EI receipt
• self-employment status
• RRSP contributions
• RRP contributions
• union membership

19
Descriptive analysis of incidence

• Based on 27 cohorts
• defined by calendar year in which the individual
  turns 65 (1982-2008)
• Figure 1 shows incidence rate by calendar year
  for 67-year olds
• for all cohorts pooled together
• Snapshot 34 in 1992, 30.5 in 2006
• No evidence that outreach program launched in
  2002 had the expected impact

20
Within full scale there is not a large change
21
Magnifying there is an uneven downward trend
since early 1990s
22
Descriptive analysis of incidence (cont.)

• Figure 2 shows the incidence rate as a fraction
  of the cohort that becomes age eligible
• All cohorts pooled together
• What happens to these cohorts as they age beyond
  65 years old?
• Incidence rate rises
• Implying that entries dominate exits

23
Incidence rises monotonically with age(all
cohorts combined)
24
Same figure rescaled
25
Descriptive analysis of incidence (cont.)

• Figure 3 shows these incidence profiles for
  selected cohorts
• They tend to rise with age

26
Rising incidence with age within earlier cohorts
27
Rising incidence with age within later cohorts
28
Results from multi-variate analysis

• Reference category has following attributes
• Single male
• Native Canadian
• Ontario
• Resident of Canada
• Speaker of majority language
• ASR gt 500,000
• Permanent income 30-40 K

29
Results from multi-variate analysis (cont.)

• Not previously self-employed
• Not previously on EI
• Not previously unionized
• Not previously contributed to RRP
• Not previously contributed to RRSP
• Constant term about 0.39
• Selected results shown as deviations

30
Results from multi-variate analysis (cont.)

• R-bar2 0.265
• there is thus a huge wedge between permanent
  income and contemporaneous income upon which
  entitlement for GIS is based
• Diagrams show incidence
• Including only the demographics as controls
• Including both the demographics and
  income-related variables
• All specifications include age, cohort, and (age
  cohort) indicators

31
East-West divide very clear Nfld. has the
highest rate
32
Immigrants much more likely to claim
33
Having a partner matters much more than gender
34
Negative, monotonic effect of permanent income
35
Effect of self-employment negativeRSPs and
RRSPs not substitutes
36
Other multi-variate findings

• Non-residents of Canada unlikely to claim
• Recent immigrants to Canada unlikely to claim
• Area size of residence mostly monotonic pattern
  
• Incidence decreases with density
• Role of population density important for EI as
  well

37
Dynamics

• Follow all subjects for five-year period between
  the ages of 66-70 years
• Calculate relative frequencies for discrete
  distribution of 0,1,2,3, or 5 years of receipt
• 71.3 of beneficiaries persistent
• Over interval from 66-70 years of age, 4 or 5
  years of receipt out of potentially 5 years
• 28.7 are intermittent, so dynamics cannot be
  ignored

38
Only 42 never receive benefits those who do
tend to be persistent
39
Among those entering at age 66, over 80 of them
claim every year
40
Dynamics (cont.)

• Cross-tabulated results
• Same empirical patterns as the case for receipt
  during any given single year
• Any factor that raised the likelihood of that
  outcome also applied to the outcome of persistent
  use

41
Conclusions

• Rough approximation of incidence rate 1 in 3
• Little evidence of administrative anomalies
• Having a partner has strong negative impact on
  receipt
• matters much more than gender
• No evidence that outreach program of 2002 had
  expected impact
• Incidence tends to rise monotonically with age
• Entries outweigh exits

42
Conclusions (cont.)

• Mild downward time trend over cohorts
• east-west divide amongst incidence rates very
  sharp
• Not unlike EI and CPPD
• immigrants more likely to claim
• Provided that they have met the residency
  requirements

43
Conclusions (cont.)

• Proxy for permanent income explains only a minor
  proportion of the variation
• Further research warranted into whether
• Individuals who could save for retirement are not
• Negative shocks during the final third of the
  career are frequent
• Certainly has the right sign
• Next stage is to estimate hazard models of
  entries into and exits from the GIS regime