Transcript Labour force participation and retirement of older New Zealanders

Labour force participation and retirement of older
New Zealanders
Emma Gorman, Grant M. Scobie
(NZ Treasury)
Andy Towers
(Massey University)
A presentation to the Population Ageing and Labour Market International Research Workshop
University of Waikato
© The Treasury
02 Feb 2012
Disclaimer
The views, opinions, findings, and conclusions or recommendations
expressed in this presentation are strictly those of the author(s). They do
not necessarily reflect the views of the New Zealand Treasury or the New
Zealand Government. The New Zealand Treasury and the New Zealand
Government take no responsibility for any errors or omissions in, or for the
correctness of, the information contained in these working papers. The
paper is presented not as policy, but with a view to inform and stimulate
wider debate.
2
© The Treasury
Overview
1.
2.
3.
4.
5.
6.
7.
3
Introduction
Motivations
Data
Descriptive statistics and definitions
Methods and results
Conclusions and discussion
Further work?
© The Treasury
Introduction
• New Zealand’s population is ageing
• Health costs + superannuation = fiscal pressures
• Increasing labour force participation
– Alleviate fiscal pressures
– Health and well-being
– Growth and productivity
• Our primary research question: what factors are
associated with labour force participation, in particular
the role of health.
4
© The Treasury
Keep working –
your country needs you!
Rapid increase in
participation of 65+
(aggregate data)
• Participation rates of older
New Zealanders are high
and increasing (relative to
other OECD countries)
- Increasing participation of
females
- Increase in age of eligibility
for NZS
Source: HLFS and StatsNZ medium projections
• Upward trend projected to
continue.
© The Treasury
2. Motivations
• Literature identifies poor health, financial incentives
(disability insurance, pensions), financial security to
encourage exit from labour force; continued
employment of spouse associated with further
participation.
• Enright and Scobie (2010) found health and marital
status to be important factors, using first wave of
Health, Work and Retirement survey (cross-section).
• Extends current literature by utilising longitudinal
data.
6
© The Treasury
Three approaches
1. Transitions to retirement
• Full-time to part-time
• Full-time to retirement
• (relative to remaining in full-time employment)
2. Retirement vs. participation
3. Effects of chronic disease on participation
7
© The Treasury
Data
• The datasets used are the Health, Work and Retirement (HWR)
survey (run by Massey University, funded by Health Research
Council).
• Aged 54-70 in baseline survey in 2006, followed up in 2008.
• Baseline sample randomly selected from electoral roll, with
oversampling of Maori to combat expected attrition
• ...and the first wave of its successor study, the New Zealand
Longitudinal Study of Ageing (NZLSA) run by Massey University,
and Enhancing Wellbeing in an Ageing Society (EWAS): a
research collaboration between the the Family Centre Social
Policy Unit and Waikato University.
8
© The Treasury
Data
• NZLSA is funded by Ministry of Science and Innovation
for 2010 and 2012.
• Three wave panel: 2006, 2008 (HWR) and 2010
(NZLSA) of approximately 1,800 New Zealanders
surveyed three times on their health, wealth and
social well-being.
9
© The Treasury
4. DESCRIPTIVE STATISTICS AND
DEFINITIONS
10
© The Treasury
0
.05
.1
Density
.15
.2
Expected ages of complete
retirement increase over time
55
60
65
70
75
Age expect to retire completely
80
85
2006
2008
2010
11
© The Treasury
Approximately 20% of retired individuals indicate they would
prefer to be in part-time employment. Barriers to participation?
Actual
2006
Full-time
Part time
Retired, no paid
work
2008
Full-time
Part-time
Retired, no paid
work
Full-time
Preferred
Part time
Retired
Other
62
12
3
26
76
23
9
7
70
3
5
4
100
100
100
Full-time
55
6
s
Part time
35
87
19
Retired
8
5
75
Other
2
s
6
100
100
100
s indicates an underlying sample size too small for meaningful inference.
12
Demand – mismatch of skills with industry demand, lack
of suitably flexible or part-time jobs, unwillingness to retrain or up skill older individuals, age-based
discrimination?
Supply – health and disability constraints, care giving
responsibilities, leisure preferences, social norms?
13
© The Treasury
Reasons for reducing or stopping work
...health and NZ Superannuation
0.300
0.250
0.200
0.150
0.100
0.050
0.000
Notes:
14
1. Respondents can indicate multiple categories.
© The Treasury
PARTICIPATION AND RETIREMENT
15
© The Treasury
Participation and retirement
• We construct two mutually exclusive categories for
analysis:
– 1. Participating (working or seeking work)
– 2. Retired (retired, no paid work)
• Students, homemakers and those with no identifiable
labour force status are excluded from the analysis.
• Only consider those who respond to all three waves
(approx 1800).
16
© The Treasury
Measuring health
• Self-rated health (poor, fair, good, very good, excellent)
• SF-12. summarises physical and mental health into two
property name.
indices (meanError!
50, Unknown
SD 10, document
range 1-100).
General health (overall perception of physical health)
Physical functioning (degree of health-related functional limitation)
Physical Health
Role physical (degree physical health affects daily activities)
Bodily pain (degree of current bodily pain)
Mental health (overall perception of mental health)
Mental Health
Role emotional (degree emotional health affects daily activities)
Social functioning (degree of health-related social limitation)
Vitality (general degree of perceived energy)
• Chronic disease indicators.
17
© The Treasury
Participation and health
Retired
Female
Physical
(SF12)
Mental
(SF12)
•
18
Participating
Male Female
Male
45
46
50
50
55
54
54
55
Retirees of both gender have lower physical
health than those participating.
© The Treasury
Health transitions
When
participating
(t)
When retired
(t+2)
Always
participating
Always
retired
Male
(SF12)
48
45
51
46
Female
(SF12)
46
45
51
46
• Those who move into retirement in next period are in
worse physical health initially than those who are always
participating
19
© The Treasury
Participation and health
• Justification bias: rationalisation of retirement.
– Changing social norms, no empirical consensus
• Measuring health – differing reference points,
relevance for participation?
• Reverse causality – effect of retirement on health
• Unobservables factors, may affect both health and
participation
• Health stock
- (Grossman, 1972)
20
© The Treasury
New Zealand Superannuation (NZS)
• Paid to all residents aged 65 or over
• Flat rate indexed to average weekly earnings
• No income, asset or work test
• Aims to prevent poverty in old age
• Spending on NZS projected to increase from 4% of GDP
in 2009 to 8% in 2050.
21
© The Treasury
METHODS AND RESULTS
© The Treasury
Methods
1. Transitions to retirement
• Full-time to part-time
• Full-time to retirement
• (relative to remaining in full-time employment)
2. Retirement vs. participation
3. Effects of chronic disease on participation
23
© The Treasury
1. Transitions
Full-time(t) – retirement(t+2)
• Y=
Full-time(t) – part-time(t+2)
Full-time(t) – full-time(t+2)
• = f ( health, net wealth, other household income, ethnicity,
region, migrant status, highest qualification, gender, financial
incentives (NZS), number of financial dependents, marital
status, participation status of spouse, attitude toward
retirement, age, survey year)
• Measured at time t
24
© The Treasury
1. Results
• Females more likely than males to move into part-time
rather than remain in full-time work.
• Better physical and mental health associated with
lower probability of direct transition to retirement,
measured by both SF-12 and self-rated.
• Age – older individuals more likely to move out of fulltime work.
• Examine role of health using more data, i.e. including
those who are always retired or always in work.
25
© The Treasury
2. Retirement vs. participation
1 if participating
• Y=
26
0 if retired
© The Treasury
Pr(labour force participation)
2006
2008
2010
Low health
Medium health
Health
27
High health
2. Retirement vs. participation
• Aim to account for unobservable individual specific
effects, using logit with random effects.
Yit  1( Xit `  Zi `  i  uit  0)
Explanatory variables
Individual specific
effect
Random error
• Likely to be correlation between unobservables and covariates
(e.g health), so bias from omitted variables remains.
28
© The Treasury
2. Retirement vs. participation
• Parameterise this correlation between covariates and unobservables
(Mundlak, 1978).
• i  X i `  it
• The final specification includes terms intended to capture time
invariant differences between people (unobserved heterogeneity); as
well as a term which captures the effect of changes within individuals
over time.
• i.e. Identify direct effect of health through within person variation.
• Estimate models separately for males and females.
29
© The Treasury
2. Results
Participation (base=retired)
Males
Females
Pooled logit
Logit with random effects
Average marginal effects (percentage points)
Net wealth quartiles (base=Q1)
Q2
Q3
Q4
Education (base=none)
Secondary
Tertiary
On NZS
Financial dependents
Health status (SF12)
Physical
Mental
Marital status (base=married with a non-working spouse)
Separated / divorced
Widow / er
Never married
Married with working spouse
2
2
-3
2
5
-2
Pooled logit
Logit with random effects
Average marginal effects (percentage points)
-3
-8 ***
-8 ***
-2
-7 **
-7 **
7 ***
9 **
-10 ***
3*
3
5
-13 ***
3 **
3
4
-8 ***
1
7 ***
9 ***
-29 ***
8 ***
4 ***
1 **
2 ***
1
3 ***
1 **
5
4
-7
7 ***
3
1
-7
4*
13 ***
6*
6
2
0
1
11 ***
4
7
1
- Other controls are age, region, ethnicity, other superannuation, benefit receipt, migrant status, attitude toward retirement, log of other household income.
- (***) 10% (**) 5% (*) 1%
30
Results
• After accounting for unobserved heterogeneity, physical health remains a
significant determinant of participation for males, however for females
this effect disappears.
- Direct effect of physical health on male labour supply?
- Females less likely to be employed in physically demanding
occupations?
• Individual specific heterogeneity explains a substantial proportion of
variation.
- Decisions by households and individuals are complex, and influenced
by factors unobserved and unreported in a survey.
31
© The Treasury
Results
• Continued employment of spouse – higher participation for males.
- No effect for females – more likely to be secondary earner.
• Dissolution of marriage – separation, divorce, widow – higher
participation for females.
• Large effects of New Zealand Superannuation and financial dependents.
• Positive effects of education found for females only.
32
© The Treasury
3. Chronic illness
•
•
•
•
•
•
•
•
33
Epilepsy
Diabetes
Blood pressure
Heart
Asthma
Respiratory
Ulcer
Stroke
•
•
•
•
•
•
•
•
•
•
Liver
Bowel
Hernia
Kidney
Skin
Arthritis
Sight
Hearing
Cancer
Hepatitis
© The Treasury
Prevalence of chronic disease by
labour force status
34
3. Chronic illness
1 if participating
• Y=
0 if retired
• =f ( indicators for each chronic disease, demographics and
economics controls as in previous models)
• Estimated separately for males and females.
• Marginal effects weighted by prevalence, implied effect on
aggregate participation.
35
© The Treasury
A.
Females
High blood pressure
Heart problems
Arthritis
B. Males
High blood pressure
Heart problems
Kidney
Cancer
36
Initial probability
After the change
62.6
66.4
67.9
67.5
59.6
62.0
Average marginal effect
(percentage point change in
probability)
-4.9
-6.8
-5.9
69.0
66.6
62.8
64.7
73.4
72.9
72.1
72.8
-4.4
-6.3
-9.3
-8.1
In summary
• Poor physical and mental health associated
with retirement, and transitions to retirement.
• After accounting for unobserved heterogeneity,
physical health remains important determinant
of retirement for males only.
• Heart problems and high blood pressure most
critical chronic diseases for participation.
• Other factors: NZS, marital status.
37
© The Treasury
Further work
• Insight into retirement transitions
• Labour market returns to health
investment
• Further realism of assumptions around
retirement in age in modelling
38
© The Treasury
Thank you
...Questions?
39
© The Treasury