Do Working Conditions at Older Ages Shape the Health Gradient? 17

advertisement
Do Working Conditions at Older
Ages Shape the Health Gradient?
17th Annual RRC Conference
August 6-7, 2015
Lauren L. Schmitz
New School for Social Research
Support for this project has been provided by the Center for Retirement Research
(project BC14-D3) pursuant to a grant from the U.S. Social Security Administration
and from the National Science Foundation (grant no. 1356857). All findings and
conclusions are those of the author and do not represent the views of the
National Science Foundation, the Social Security Administration, or the Center for
Retirement Research.
Workers in lower paid or manual occupations
have worse health outcomes
Occupational health gradient, full time workers ages 50-64, HRS (1992-2010)
0.2
0.1
0.08
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.18
-0.23
-0.35
-0.49
-0.53
-Coefficients are from a pooled ordered probit regression of SRHS on occupation
-Includes controls for age, sex, race, marital status, veteran status, year and cohort FE
-Omitted occupation=Executive & Managerial
Why do differences in health persist across
occupations at older ages?
• Do working conditions in the years leading up to retirement
contribute to health transitions?
• Or, does the gradient stabilize at older ages?
– Prior labor market experiences and other early life factors play a
bigger role in shaping the health disparities we see between
occupations
• (Gueorguieva et al. 2009, Fletcher et al. 2011)
The impact of occupation on health in the years
leading up to retirement is not well understood
• Difficult to estimate causality  workers do not randomly sort into jobs
– Workers sort into jobs based on observable and unobservable characteristics
that are highly correlated with health
»
Childhood health, childhood SES, adult health, earnings, risk preferences
influence both health and job selection across the life span (M. Marmot et al.
1997, Smith 1999, Korpi 2001, Ferrie et al. 2002, Schur 2003, Burgard & Lin 2013)
– Prior working conditions have a cumulative or durable impact on health
(Gupta & Kristensen 2008, Gueorguieva et al. 2009, Fletcher et al. 2011,
Robone, Jones, & Rice 2011, Fletcher 2012)
• Majority of longitudinal studies have only examined the impact of physical
job demands on health
– Combination of low decision latitude or job control and high job demands
that is bad for health (Karasek 1979, Karasek & Theorell 1990, Siegrist 1996,
Bosma et al. 1998), particularly at older ages (Ravesteijn et al. 2013)
Use two novel strategies to estimate effect of job
demands on the health of male workers ages 5064 in the HRS
1.
Dynamic panel model takes a life course perspective to minimize
confounding and selection bias
•
•
•
1.
Control for measures of health and SES over the life course
Include trends in earnings history from SSA earnings records for jobs
held before workers enter the HRS
Incorporate physical and psychosocial demands in the same framework
Instrumental variables (IV) specification exploits random sample of
men from the Vietnam-era draft to estimate the impact of working
conditions on the health of aging Vietnam veterans
•
Verify findings from the panel model using a different technology
Broader impacts and policy relevance
• Evaluate efficacy of policies that advocate delaying retirement
to increase labor force participation at older ages
• Healthy jobs increase productivity, extend the working life,
and boost Social Security benefits and retirement savings
• Implications for Medicare claims and the cost of retiree health
care programs
Summary of data sources used in analysis
Health &
Retirement
Study (HRS)
(1992-2010)
Health status (SRHS, doctor diagnosed
conditions, cognition, & depression),
health behaviors, employment
information, socioeconomic and
demographic characteristics
Earnings history
(Average level, volatility, & growth)
SSA Master
Earnings File
(MEF)
(1980-Baseline)
✓
✓
Expert ratings of working conditions
Three digit occupation codes
Occupational
Information
Network (O*NET)
(1992-2010)
✓
✓
Measurement and selection of working
conditions from the O*NET
• O*NET has information on over 200 characteristics of the work
environment
• To keep the model parsimonious and capture the most salient pathways
between work and health at older ages, use “work ability” model to
inform selection of O*NET variables
• Construct five indicators that capture aspects of the physical and
psychosocial work environment
• Factor analysis is used to confirm that each indicator captures a unique
portion of the variance in the HRS-O*NET data and corresponds to the
dimension hypothesized in the work ability model
The Work Ability Model
(Tuomi, Ilmarinen, et al. 1991, 1997, 2001)
Promotion of Work Ability
(age 45+)
Work demands &
the physical
environment
•
•
Promotion of
professional
competence
•
Physical & mental demands
Environmental hazards
Work organization &
work community
•
Psychosocial work
environment
Job retraining and
updating of skills
Support for a
healthy lifestyle
•
•
Time for regular exercise
Limits smoking and drinking
Working Conditions Derived from the Work Ability Model, O*NET (1992-2010)
Physical demands
(mean=0.55; Std. Dev.=0.21)
Use of arms and legs and moving your whole
body, such as climbing, lifting, balancing,
walking, and stooping
Environmental hazards
(alpha=0.86; mean=0.38; Std. Dev. =0.14)
Exposure to weather, extreme temperatures,
light, noise, contaminants, or cramped spaces
Impact and frequency of decision making
(alpha=0.92; mean=0.78; Std. Dev. =0.10)
Worker’s decisions have an impact on the
image, reputation or financial resources of
their employer
Supportive management practices
(alpha=0.88; mean=0.63; Std. Dev. =0.08)
Management is supportive, treats workers
fairly, and trains their workers well
Degree of control and influence
(alpha=0.97; mean=0.59; Std. Dev. =0.14)
Job allows worker to use their abilities, gives
them a sense of achievement, independence,
variety, authority, creativity and status
Estimate a dynamic model of work and health
• Dynamic random effects model with unobserved heterogeneity
(Wooldridge 2005; Gupta & Kristensen 2008; Robone et al. 2011)
healthit = workingconditionsit-2g + Xit b + rhealthit-2 + zi + uit
• Model unobserved heterogeneity as conditional on initial health at baseline
and factors that influence health and job selection over the life course
30
0
50
64
Analysis period
Childhood demographics and
educa onal a ainment
Job history
Earnings history
Effect of working
condi ons
on health transi ons
82
IV (2SLS) approach estimates impact of work on
aging Vietnam veteran health (ages 51-62)
5
healthi = aVETi + å(b jWCij ´VETi ) + Xi¢g + ei
j=1
• The VETi and WCij×VETi terms are treated as endogenous
• Draft eligibility (DRAFTi) and a set of five WCij×DRAFTi terms are used
as excluded instruments
•
DRAFTi is equal to “1” if the respondent’s draft lottery number was
called in the corresponding draft year and “0” otherwise
• Consistent estimation of the interaction terms requires that the
instrument (draft eligibility) be independent of working conditions
and any omitted variables (Nizalova & Murtazashvili, 2012)
 Independent tests confirm selective mortality or selection into the
sample by demographic and socioeconomic characteristics is not
operant in the aging Vietnam veteran sample
Results
• Panel model: Degree of control and influence is associated with
improved SRHS and cognition
– A one standard deviation increase in control and influence on the job
in the past wave increases the probability of reporting “very good” or
“excellent” health by 2.2 percent in the current wave—an effect
comparable in magnitude to exercising three or more times per week
– Effect persists even after controlling for childhood health, childhood
SES, education, mid-career earnings, health behaviors and
occupational status
• IV estimates of working conditions on aging Vietnam veteran health
confirm these findings
– Degree of control and influence Improved SRHS, cognition, ADL
summary scores and depression
– Environmental hazards Increased risk of heart disease and high
blood pressure
Discussion
• Overall, findings confirm psychosocial stressors in the workplace
continue to shape the health gradient at older ages
• What are the pathways between degree of control and influence
and health?
– Higher levels of control and influence on the job may insulate workers
from stress (e.g. Karasek 1979)
– Prolonged exposure to stress has been shown to damage brain function,
disrupt sleeping patterns, change how the body processes food, and lower
immunity, among others  allostatic load
• McEwen & Stellar 1993, McEwen 1999, 2000; McEwen & Gianaros 2010; Juster &
McEwen 2014
• Need more research that pinpoints health-enhancing aspects of the
work environment that may offset adverse conditions, reduce stress
over time and promote healthier aging
Thank you!
Download