-Income Support Status Amongst the Retired Population An Analysis of the

dinnerworkableUrban and Civil

Nov 16, 2013 (3 years and 11 months ago)

69 views

“Low
-
Income Support Status A
mongst the
Retired P
opulation


An Analysis of the
Incidence and the Dynamics”


Detailed Outline


Submitted by


Ross Finnie

School of International and Public Affairs

University of Ottawa

P.O. Box 450, Station A

Ottawa, ON

K1N 6N5

rfinnie@uottawa.ca


David Gray

Department of economics

University of Ottawa

P.O. Box 450, Station A

Ottawa, ON K1N 6N5

dmgray@uottawa.ca



Prepared for HRSDC/
C
L
S
RN project

Challenges for Canada’s Retireme
nt Income
System”





February 2011


1. Introduction
: Context, Motivation,

and
Policy Issues


Retirement Lost


Canada’s aging work force hasn’t saved enough to retire.
Pension benefits are being slashed, employees are working longer,
and the

elderly
are selling their homes and going back to work: Is this what your retirement will
look like?



Such was the headline of the Globe and Mail on 17 October 2009. Journalist
Jacquie McNish began a week
-
long investigation into Canada’s pe
nsion crisis with tha
t
article. The ‘vital statistics’ that she cites in order to adorn the article are the following:
84 % of public service workers have pensions, and 78 % o
f these plans are ‘gold
-
plated’,
defined
-
benefit pensions; 25 % of private sector workers have pensio
ns, and 16 % of
these plans are ‘gold
-
plated’ defined
-
benefit pensions; 11 million workers, or 60 %, of
Canada’s workers have no pension at all; 8 million workers, or 45 %, have neither
pensions nor registered retirement savings plans.
Those figures do
not reflect the under
-
funding and the actuarial unsoundness of many existing pension plans, such as the ones
for Air Canada and General Motors Employees.



It is against the backdrop illuminated by those stylized facts that our proposed
project fits. Gi
ven the scant labour market opportunities that are available for workers
that are considered to be of normal retirement age or older (i.e 65 years or older), and
given the shriveling safety net in the form of private pensions, both of the defined benefit
o
r the defined contribution type
, the income security of retired Canadians is of grave
concern.

Our focus is the state of
poverty amongst this segment of the population.
While it is well known that until recently, the three elements of Canada’s
public

pe
nsion
schemes, namely the CPP, the OAS,
the OAS allowance

(paid to
the
partner of the GIS

recipient
)
,
and the GIS (guaranteed income s
upplement
) regimes had succeeded in
drastically reducing the poverty rate among retired Canadians over the post war period
, it
seems reasonable to expect that these regimes will face great fisc
al strain within a few
years (Uppal et al. (2009))
.



The primary topics of our project are a) the incidence of
‘neediness’ (to be
defined presently)
amongst this population and b) the
dynamics of
entries and exits from
this state.
Our analysis is in the spirit of the poverty/low
-
income literature that is fairly
developed
in regards to
the working
-
age population, and we intend to borrow some of the
techniques that are applied in the lit
erature on poverty for our analysis. In a point of
departure from that literature, however, we link in retrospective fashion the labour market
histor
ies

of retirees during at least part of their working
lives to their post
-
retirement
‘neediness’
status.
As such,
we
include in our analysis several phases of the life cycle
.



There are a

number of metrics that one could employ in order to flag the state of
‘neediness’. For instance, there are the well
-
known measures of the ‘LICO’ (low
-
income
cutoff) and th
e ‘LIM’

(low income measure) whose calculations involve subjects’
incomes. In contrast, in this project our metric involves a direct measure of a
governmental intervention that is designed to address poverty amongst the retired
population
, namely the
GIS
(guaranteed income security) benefit.
The GIS r
egime is the
only
means
-
tested

public retirement benefit that is targeted to the
group of
retired
individuals and couples who are deemed to be the neediest.
Whereas the CPP
(Canada
Public Pension)
regime is
a pay
-
as
-
you
-
go system that bases benefits in part on prior
contributions, and the OAS
(Old Age Security)
benefit takes the form of a universal
demogrant, the GIS is the last resort for income support.
1

It is
thus
a quintessential
‘safety net’ type provis
ion. As such, it is the most directly targeted public policy
instrument for addressing poverty amongst this group, and is most likely to
encounter
fiscal
strain with the fraying of the
patchwork of
private pension schemes.



2. Literature Review



Our
paper is in the spirit of a recent piece by Uppal et al. (2009). Using
essentially the same data set that we propose to exploit, namely the LAD file, those
authors conduct a credible and interesting empirical analysis of the incidence of GIS
receipt. Mor
e specifically, they model the event of a subject receiving full or partial GIS
benefits for three consecutive years for those aged 66 to 68

years
. In a similar vein to
our project, their primary explanatory variables are indicators for the subjects’ pri
or
income levels (including labour market income, non
-
labour market income, and other
family income), indicators for major changes in prior income levels, and proxies for the
degree of stability of labour market income. We plan to update their analysis (
the latest
year for which they have data is 2006), and we will include a wider range of explanatory
variables, some of which measure contemporaneous attributes. We plan to expand the
scope by analyzing transitions in and out of GIS
-
receipt status. As tho
se authors note,
“…some individuals will have income near the boundaries of GIS eligibility and cycle in
and out of receipt regularly, while others may drop into or out of receipt because of one
-
time factors such as RRSP withdrawals or investment gains” (p
. 11). Uppal et al. (2009)
deliberately abstract from such variability by modeling only the incidence of the event
described above.



While we intend to keep our analysis focused as described above, we will relate
our findings to other issues that hav
e been treated in the burgeoning retirement literature.
For instance, there is the goal of supporting flexibility in transitions to retirement, as
outlined in 2008 report of the expert

panel on older workers to the Minister of Human
R
esources. A recent c
ollection of papers (Abbott
et al.

2008) stemming from a
colloquium will serve as our source for conducting our survey of the literature and
situating our results within it. We will make explicit the policy ramifications of our
findings, paying particular

attention to policy instruments such as the CPP.
In the
interests of universality, the literature
survey
component of our project
will relate our
work to selected pieces from the US literature and
OECD surveys of retirement issues.



The methodological

blueprint
for our study
is
borrowed from
a number of fairly
recent studies from the sub
-
strand dealing with poverty dynamics (Finnie and Irvine
(2008), Finnie, Irvine, and Sceviour (2005), Finnie, Irvine, Sceviour (2004), Finnie, R.



1

The OAS benefit is clawed back for very high
-
income recipients.

and A. Sweetman (2003)
).
Our project thus involves
a synthesis of a sub
-
strand of the
literature on retirement issues (geared toward low
-
income status) and the literature on
poverty dynamics amongst the working
-
age population.


3.

Outline of
Empirical Approach





The focus

of our empirical analysis is on the event of a subj
ect being in a low
-
income state.
We will interpret the receipt of the GIS benefit as a flag for the state of
‘neediness’
among the retired population.
In the case of married or common
-
law
couples, the
GIS benefit is sometimes combined with the OAS allowance benefit (if the
partner is between 60
-
64 years of age).
This
complementary
benefit is designed
expressly to assist couples who receiv
e
only one rather than two OAS pensions
.
We
intend to conduct a
very focused empirical investigation of the
GIS status
of such couples
once they are no longer age
-
eligible for OAS allowance benefit.



We will commence our analysis with a brief introduction based on the aggregated
data for
GIS
payments. The goal of t
his extremely simple analysis is only to examine the
basic trends and patterns over the past three decades. This
portion
would be based on
publicly available data.


The
bulk of our value
-
added is derived from micro
-
level data. The
major objectives
are
the following:




An analysis of the statistics describing the
trends and patterns
of the
incidence of GIS benefits from 1982 until
the latest year

for which data
are available (probably 2007).
C
ross
-
tab
ulated statistics of incidence,
such as means, varianc
es, and order statistics will be calculated according
to variables

that are listed below.



An analysis of the
incidence of

‘neediness’

amongst the elderly
, as
measured by the state of
GIS receipt

upon the transition into retirement
upon reaching the GIS
-
eli
gible age of 65 years
, based on
cross
-
sectional
identification.

This is modeled
as a function of those attributes and
variables listed
below
as well as the calendar year in a multi
-
variate
framework
.
This approach will exploit the longitud
inal dimension
of our
data set (described in more detail below).





An analysis of the
d
ynamics of
‘neediness’
amongst

the elderly, as
measured by
the state of
GIS recipiency
, based on
longitudinal
identification.


o


We first e
xamine
entrances

after
reaching the
age of

65
(i.e.
post
retirement,
the subject transits from the state of
non needy to needy
).

o

We next investigate the event of
exit
ing

from this state, which might be
relatively
rare. Possible correlates include re
-
entr
y

into the labour market
(and thus reversi
ng a retirement choice) or re
-
marriage.
Conditional on
this event, we plan to examine the subsequent income levels.


Given the main outcomes to be explained, namely the incidence of GIS receipt
upon
retirement at age 65
and
subsequent
transitions in an
d out of that state, t
he explanatory
variables are divided into three major groups.




Policy variables
:
We will conduct a statistical analysis of i
nteractions
between outcomes related to
GIS
receipt and outcomes regarding the two
other primary policy in
terventions that are
targeted at the retired
population:
OAS and CPP
benefits.
One policy outcome of interest is
early entry into the CPP regime before the age of 65 is attained.
In this
instance, the correlations and associations that are estimated can
not be
attributed to causality. These correlations will capture in part mechanical
effects, as eligibility for GIS benefits is means
-
tested.

We note that there
was a policy change that applied directly to the GIS regime in 2002, and
we will search for e
vidence of its effect.



Demographic effects
: Information is available for the following attributes:

age, sex, marital status, urban/rural residency, language (French versus
English), and immigrant status.

We can also observe events such as a
change in

marital status (i.e. a marriage or re
-
marriage, a divorce, or a
death of a spouse) that we expect has an impact on the dynamic aspects.

o

Income/wealth

effects:
This
group of variables is selected in order to
investigat
e
the long
-
term factors that are tied

to events occurring at
earlier

stages of the subject’s life
-
cycle. How well
-
paid was the subject, and how
stable was his/her labour market attachment? Examples of these
indicators of prior labour activity include levels of labour market income,
levels
of non
-
labour market income, the existence of self
-
employment
income, variability of income, coverage by a

private
pension plan, saving
for retirement, EI/UI recipiency, and CPPD recipien
cy before normal
retirement age.
This approach will allow us to exa
mine questions such as:
is it common for subjects to deplete their savings or to receive defined
benefit pensions that are eroded by inflation?



We not
e

that some of these variables will
capture in part
mechanical effects. For instance, the proxy that

we calculate for
permanent income over the working years is highly correlated with post
-
retirement income in the form of pension
s
upon which eligibility of receipt
of GIS benefits is based.


4.
Data
Source
and
Empirical Methodology


Our data set is th
e Longitudinal Administrative Database (
LAD
), which is based
on T1 tax data.

This data set has the advantage of having very large sample sizes, as it
now consists of a 20 % sample of the underlying population of adult Canadians.
It
contains detailed inf
ormation on the levels of income as well as a breakdown of the
sources of income. The record which is critical for our study is a flag for
‘neediness’

amongst the elderly. It is measured at the family level and is labeled
“Net Federal
Supplement
s


GIS
+ spouse’s allowance
” Since

1992, this record should contain
accurate information for a representative sample
of adult Canadians
due to the fact that
after 1991, almost all low
-
income individuals have an incentive to file a tax return in
order to benefit f
rom the GST
/HST

rebate. Before 1991 this record is unreliable

because
many low
-
income individuals had little incentive to file a tax return
,
and so we cannot
include anyone in our sample who retired before 1992. W
e can
, however,

make use of
pre
-
retiremen
t labour market histories of subjects betwe
en 1982 and 2006 and include
tem as explanatory variables.



The first part of our econometric analysis
consists of
the
incidence analysis
. This
has an analogue in the literature on post
-
secondary education

(
PSE)
, an application of
which is Finnie and Mueller (2009).
In that sub
-
strand of the literature, outcomes
occurring during the PSE period are linked to outcomes and variable
s

occurring during
the lengthy pre
-
PSE period.
We seek to model the incidence of

post
-
retirement poverty
as a function of contemporaneous attributes and pre
-
retirement variables.
We d
raw

from

a sample of subjects who retired after 1991,
which constitutes our risk set for the event of
entering retirement and receiving GIS benefits.
T
he underlying population is those
individuals who are retired and eligible to receive GIS benefits between 1992 and 2006.
The dependent variable is dichotomous and captures the event of
receipt of GIS benefits
during the
reference year. T
he longitudinal
feature of the LAD file allows as to
retrospectively ‘
peel
back their histories’
into the pre
-
retirement period. For an
individual retiring in 2006

(who are the youngest in our sample)
, we have retrospective
information going back to
1982, when they were
age
41 or so.
For all
other cohorts
, we
observe their labour market itineraries
starting at older ages
.
We will exploit the
longitudinal feature of the LAD file, so most of the
subjects will be observed over
different phases of their
own
life
-
cycles.
We

must t
ake account of fact that we do not
have obs
ervations

for
all subjects over the same pre
-
retirement intervals, and in this
sense, the structure of our estimating sample is
unbalanced
.


The event of formal retirement from the labour force does not h
ave a precise
definition, and it is not straightforward to pinpoint the exact timing. For the purposes of
this part of
our research, we are interested in those whose initial entry into retirement
is
accompanied by GIS receipt. Since one has to reach 65
years of age in order to be
eligible, our set of eligible workers is restricted to those who are either 65 or 66 years of
age. If they receive the GIS benefit, we treat them as both retired and a GIS recipient. If
they do not receive the GIS benefit, and

they do receive CPP benefits, we will treat them
as retired. This
s
tep could potentially miss individuals who retire but do not qualify for
CPP benefits due to an insufficient work history. To address this issue of identifying the
retired

population
, w
e
will
add a further check by
search
ing

for precipitous declines in
declared labour market earnings observed between the ages of 63 and 64 and 65 and 66.
By extending this window of observation for earnings back to ages 60 (or thereabouts),
and searching

for a precipitous decline in earnings, one can capture many of those who
retired early from the labour force.



For this component of our econometric analysis

(i.e. the incidence analysis)
, only
a few of the contemporaneous variables can be included as

explanatory variables. Most
notably it is not appropriate to include variables such as retirement income from other
sources, as one would
only
be capturing mechanical effect
s
.
We
seek to assess the
impact of
long
-
term factors
,

such as the subject’s perm
anent income when he/she was
active, or his/her propensity to save.
2

While these variables are quite relevant and are
pre
-
determined with respect to post
-
retirement poverty,
the
y tend to be
correlated with
the
contemporaneous variables. L
ow
-
income
status

ex post
is caused in large part by
low
-
income
status
ex ante
, and this is driven in part by relatively time
-
invariant,
unobservable

factors.
The exogenous variables that we intend to include, such as
coverage by a private pension plan, coverage by an ind
ividual retirement fund, and
variability of income over the working life, are likely to be s
omewhat
endogenous.
Ideally one would observe
truly exogenous variables

pertaining to their backgrounds
,
such as
tastes for
risk
, the discount rate for future inco
me, innate ability in the labour
market, myopia in regards to planning for retirement,
and
the
propensity to save
, but such
factors are u
nobservable. One might argue that evidence of p
reparation for retirement
in
the form of RRSP and/or

RRP
contributions,

while being
functions of income
,
should
also
exh
ibit some independent variation.

In summary, the indicators that we employ for
income over prior phases of the life
-
cycle are pre
-
determined, but they might not be
totally exogenous variables.



Despite t
he fact that we can probably not totally remedy the econometric
challenge discussed in the above paragraph,
and that certain estimates might be biased,
from a public policy perspective, our estimates could still be quite informative.

We will
assess the m
agnitude of the statistical relationships, which might be useful for forecasting
purposes and for reforming Canada’s retirement income apparatus with both its public
and its private components.



The second part of our econometric analysis pertains to th
e
transitions into and
out of the state of GIS receipt.
The first equation will model the event of transiting from
the state of non
-
receipt to the state of receipt. The initial
conditions for this event are

retirement at the age of 65 (or before), and n
ot receiving GIS benefits during the first two
years of potential eligibility: i.e. ages 65 and 66. The sample is expressly selected such
that it consists of those who commenced their retirement without receiving GIS benefits.
There are a number of flag
s that one could use in order to restrict the sample in this
fashion, such as including subjects who are 66 or 67 years old as our risk set and looking
for the presence of retirement income coupled with low (or zero) levels of earnings.

Although this is

a selected sample
,
it is well
-
defined
and is designed to address a
pointed
question
. Given the nature of the question being addressed, it is unlikely that the
selection of the sample will impart a bias on our estimated coefficients.



The estimating eq
uations will be set up as
hazard models based on transition data

at an annual frequency. They thus take a discrete choice form, and the
parametrization

will be the logit model.
As mentioned above, t
he methodological blueprint is set out in a
number of fa
irly recent studies
from the sub
-
strand dealing with poverty
dynamics.






2

By permanent income we mean essentially the maximum point (or near the peak) on the

worker’s age
-
earnings profile, which does not account for transitory deviations that can occur from year to year. This is
sometimes estimated by averaging actual earnings over an extended period (of perhaps 5 years in our
paper)



The second equation for the part of the econometric analysis that focuses on
transitions models the event of leaving the state of GIS receipt. This will be estimated as
a hazard
model that is also based on
annual transition data. The parametric form will
again be the logit model. In this instance the risk set and the event are very well
-
defined,
and so the initial conditions problem should not come into play.



All of our emp
irical analysis will be based on estimating samples that are
separated by gender.
As such, it will be straightforward to compare any set of descriptive
statistics or regression results across the genders. This topic is rife with policy issues
that have
a gender perspective. For instance, it is well known that the incidence of
poverty is higher among singles than it is amongst married couples. It will be interesting
to compare the dynamics of poverty in reference to a divorce or the death of a spouse


is
there a major difference between the case of the death of a

wife and the death of a
husband?

We will investigate closely the effects of family structure variables, which are
well reported in the LAD file.



5
. Results


6. Conclusion


References


Abbott, M., C. Beach, R. Boadway, and J. MacKinnon (2008), editors
Retirement Policy
Issues in Canada
,
John Deutsch Institute for the Study of Economic policy, McGill
-
Queens University Press


Baker, M. and D. Benjamin (1999) “Early Retirement Provisions
and the Labor Force
Behaviour of Older Men
: Evidence from Canada”
Journal of Labor Economics
part 1, 17
(4) , 724
-
56


Finnie, R. and R. Mueller (2009) “The Effects of Family Income, Parental Education, and
Other Background Factors on Access to Post
-
Seconda
ry Education in Canada” MESA
project Research paper


Finnie, R. and I. Irvine (2008) “The Welfare Enigma: Explaining the Dramatic Decline
in Canadians’ Use of Social Assistance, 1993
-
2005” Commentary on Social Policy
number 267, CD Howe Institute


Finni
e, R., I. Irvine, and R. Sceviour (2005) “Social Assistance Use in Canada: National
and Provincial Trends in Incidence, Entry, and Exit” Analytical Studies Research Paper
Series, Statistics Canada, working paper number 245


Finnie, R., I. Irvine, and R.

Sceviour (2004)

“Welfare Dynamics in Canada: The Role of
Individual Attributes and Economic Policy Variables” Analytical Studies Research Paper
Series, Statistics Canada, working paper number 231


Finnie, R. and A. Sweetman (2003) “Poverty Dynamics: Emp
irical Evidence for
Canada”
Canadian Journal of Economics

36, No. 2, pp. 291
-
325


Milligan, K.
(2008) “
The Evolution of Elderly Poverty in Canada


Canadian Public
Policy

34 S, S79
-
S94


Milligan, K. and T. Schirle (2008) “Improving the Labour Market Incen
tives of Canada’s
Public Pensions”
Canadian Public Policy

34 (3), 281
-
304


Myles, J. (2000) “The

Maturation of Canada’s Retirement Income System: Income
Levels, Income
Inequality
, and Low
-
Income Among the Elderly”
Statistics

Canada,
Catalogue

No.
11F0019MP
E issue No. 147


Schirle, T. (2009) “Income Inequality Among Seniors in Canada: The Role of Women’s
Labour Market Experience
Canadian Labour Market and Skills Researcher Network

WP
No. 51


Uppal, S., T. Wannell, and
E. Imbeau (2009) “Pathways into the GIS

Perspectives on
Labour and Income
Statistics

Canada


Catalogue No. 75
-
001
-
X, August pp. 5
-
14