Work Disability, Work, and Justification Bias in Europe and the US

1
Work Disability, Work, and Justification Bias
in Europe and the US
Arie Kapteyn (RAND) James P. Smith (RAND) Arthur
van Soest (Netspar, Tilburg University)
2
Overview

• Motivation
• Anchoring vignettes, justification bias
• Data (HRS SHARE)
• Model for work disability, justification bias,
  and employment
• Estimation and simulation results
• Conclusions

3
Motivation

• Labor force participation of older workers is
  policy relevant
• Health is a main reason for not participating
• Self-reported work disability is a summary
  measure of work-related health
• But has problems
• Comparability across countries
• Justification bias
• Anchoring vignettes can be used to tackle these
  problems

4
This paper

• Demonstrates that anchoring vignettes can be used
  to analyze justification bias
• Uses data on work disability self-reports, work
  disability vignette evaluations, and employment
  status from SHARE 2004 and HRS 2004
• Introduces an econometric model which extends the
  hopit model with an employment equation and
  accounts for
• an effect of employment status on response
  scales and reported work disability
• an effect of work disability on employment
• Presents estimates of this model for the US and 8
  EU countries

5
Self report of work disability

• Do you have any impairment or health problem
  that limits the kind or amount of work you can
  do?
• Response categories
• 1. None
• 2. Mild
• 3. Moderate
• 4. Severe
• 5. Extreme/Cannot Do

6
Vignettes

• Example (affect)
• Henriette generally enjoys her work. She gets
  depressed every 3 weeks for a day or two and
  loses interest in what she usually enjoys but is
  able to carry on with her day-to-day activities
  on the job.
• Does Henriette have any impairment or health
  problem that limits the kind or amount of work
  she can do?
• None, Mild, Moderate, Severe, or Extreme/Cannot
  Do?

7
More examples

• Pain Vignette
• Catherine suffers from back pain that causes
  stiffness in her back especially at work but is
  relieved with low doses of medication. She does
  not have any pains other than this generalized
  discomfort.
• Cardio Vascular Disease Vignette
• Norbert has had heart problems in the past and he
  has been told to watch his cholesterol level.
  Sometimes if he feels stressed at work he feels
  pain in his chest and occasionally in his arms.
• We work with nine vignettes three on pain, three
  on affect, three on CVD

8
How Do Vignettes Work?
None
John 1
Mary 1
Resp. 1

Liam 1
Extreme
9
How Do Vignettes Work Differential Item
Functioning (DIF)
None
None
John 1
Mary 1
Resp. 1
John 2
Resp. 2
Mary 2
Liam 1
Liam 2
Extreme
Extreme
10
How Do Vignettes Work Adjusting for DIF
None
None
None
John 1
John 2
Resp. 2
Mary 1
Mary 2
Resp. 1
John 2
Resp. 2
Mary 2
Liam 1
Liam 2
Liam 2
Extreme
Extreme
Extreme
11
The data

• For a subset of counties that agreed to
  participate, SHARE wave 1 in 2004 included a set
  of vignette questions on general health status
  and on work limiting disabilities as part of a
  drop-off questionnaire for a random subsample of
  the 50 population.
• The eight SHARE countries that agreed to
  participate in the drop-off containing vignette
  questions were Germany, France, Spain, Belgium,
  Greece, Italy, the Netherlands, and Sweden.
• HRS 2004 administered the same vignettes, in an
  experimental module administered to a random
  subsample of the 50 population in the US

12
Example pain vignette 1 Catherine
13
Red is tough Blue is soft
Rank correlations (1,2)
0.44 (1,3) 0.06 (2,3) -0.33
14
Correlations between rankings
15
A ranking of toughness
16
Toughness and employment protection
17
Self-reported Work Disability

• country none mild moderate severe
  extreme
• -------------------------------------------------
  --
• US 49.84 21.51 16.02 7.35
  5.28
• SHARE-EU 46.56 25.46 17.37 7.80
  2.81
• germany 40.19 30.04 21.34 7.02
  1.41
• sweden 53.62 15.11 14.86 12.17
  4.24
• netherl 52.05 30.66 9.27 4.92
  3.10
• spain 47.59 20.24 17.21 11.70
  3.26
• italy 46.01 27.04 15.47 7.20
  4.27
• france 49.13 23.02 18.60 6.60
  2.65
• greece 68.32 12.17 10.20 7.43
  1.89
• belgium 36.64 35.22 19.13 6.91
  2.11
• -------------------------------------------------
  -
• Total 47.00 24.92 17.18 7.74
  3.15

18
Simulation results Work disability in EU and
US(Hopit model)
19
Simulation results Italy and US(Hopit model)
20
Simulation results Germany and US(Hopit model)
21
Simulation results Spain and US(Hopit model)
22
Employment Rates

• US 50.16
• SHARE-EU 26.27
• germany 28.42
• sweden 40.82
• netherlands 32.32
• spain 26.34
• italy 20.41
• france 26.42
• greece 25.98
• belgium 21.19
• Total 29.54

23
Work Disability and Employment

• US SHARE-EU
  
• -----------------------------
• none 67.61 36.94
• mild 48.70 22.92
• moderate 30.38 13.78
• severe 14.58 8.84
• extreme 0.99 4.54
• -----------------------------
• Total 50.16 26.27

24
Work Disability and Employment
25
Dependent Variables

26
Work Disability Equation



27
Equation for Thresholds
28
Vignette Evaluations
29
Employment Equation
30
Important Assumptions

• Response consistency same thresholds in
  self-assessments and vignette evaluations
• Justification bias Shift in response scales a
  special form of DIF
• Vignette equivalence workers and non-workers and
  respondents in different countries interpret
  vignettes in the same way
• No causal effect of employment status on health
  (cf., e.g., Böckerman Ilmakunnas, Health
  Economics, 2009)

31
Work Disability EquationModel with DIF

• US EU-US Country
  dummies
• Female -0.017 -0.075 constant
  -0.189
• Married/LT -0.118 0.078 Germany
  -0.689
• Educyrs -0.039 0.034 Sweden
  -1.066
• Heart prob 0.463 -0.033 Netherlands
  -0.699
• Lung dis 0.421 -0.098 Spain
  -0.999
• High blood 0.112 -0.028 Italy
  -0.895
• Diabetes 0.256 -0.076 France
  -1.023
• Pain 0.411 0.048 Greece
  -1.690
• Arthritis 0.364 -0.038 Belgium
  -0.654
• Cancer 0.206 0.209
  
• Cesd score 0.148 0.126
  significant
• Obese 0.137 -0.020 at
  2-sided
• Age 58-64 0.166 0.004 5
  level
• Age 65-71 0.135 0.161
  
• Age 72 0.426 0.087

32
Work Disability EquationModel without DIF

• US EU-US Country dummies
• Female -0.106 -0.007 constant
  -0.126
• Married/LT -0.175 0.110 Germany
  -0.503
• Educyrs -0.046 0.042 Sweden
  -0.612
• Heart prob 0.484 -0.027 Netherlands
  -0.615
• Lung dis 0.423 -0.141 Spain
  -0.646
• High blood 0.139 -0.065 Italy
  -0.798
• Diabetes 0.321 -0.110 France
  -0.848
• Pain 0.426 0.006 Greece
  -1.260
• Arthritis 0.354 -0.019 Belgium
  -0.542
• Cancer 0.142 0.262
  
• Cesd score 0.170 0.114
  significant
• Obese 0.167 0.004 at
  2-sided
• Age 58-64 0.157 -0.018 5 level
  
• Age 65-71 0.138 0.122
  
• Age 72 0.463 0.010

33
Thresholds Equation(Model with DIF)

• 
  US EU-US
  
• Work dummy 0.097 -0.104
  
• Female 0.096 -0.082 Germany
  -0.187
• Married/LT 0.051 -0.031 Sweden
  -0.452
• Educyrs 0.003 -0.004 Netherlands
  -0.074
• Heart prob -0.016 -0.014 Spain
  -0.356
• Lung dis -0.002 0.037 Italy
  -0.083
• High blood -0.024 0.028 France
  -0.147
• Diabetes -0.065 0.069 Greece
  -0.417
• Pain -0.038 0.046 Belgium
  -0.114
• Arthritis -0.005 -0.008
  
• Cancer 0.066 -0.048 const thrh 1
  0
• Cesd score -0.022 0.008 thr2 - thr1
  0.722
• Obese -0.039 0.014 thr3 - thr2
  0.704
• Age 58-64 0.021 0.034 thr4 - thr3
  0.822
• Age 65-71 0.020 0.039
  
• Age 72 0.021 0.026 sigma u
  0.426
• , significant at 5 and 10 level,
  respectively

34
Employment Equation (Model with DIF)

• US EU-US
• work disab -0.464 0.272
  
• Female -0.246 -0.319
  
• Married/LT -0.034 -0.125 constant
  0.440
• Educyrs 0.034 -0.020 Germany
  0.291
• Heart prob -0.015 -0.002 Sweden
  0.784
• Lung dis -0.125 0.002 Netherlands
  0.241
• High blood -0.001 0.051 Spain
  0.208
• Diabetes -0.228 0.152 Italy
  -0.126
• Pain 0.102 -0.025 France
  0.300
• Arthritis 0.031 -0.058 Greece
  0.157
• Cancer 0.004 0.034 Belgium
  0.041
• Cesd score -0.051 -0.026
  
• Obese 0.204 -0.284 ,
  signif. at
• Age 58-64 -0.620 -0.406 5,10
  level
• Age 65-71 -1.268 -1.296
  
• Age 72 -1.813 -1.402

35
Employment Equation (Model without DIF)

• US EU-US
• Work disab. -0.516 0.319
  
• Female -0.289 -0.282
  
• Married/LT -0.069 -0.096 const work
  0.473
• Educyrs 0.030 -0.015 Germany
  0.310
• Heart prob 0.015 -0.021 Sweden
  0.852
• Lung dis -0.102 -0.032 Netherland
  0.239
• High blood 0.016 0.034 Spain
  0.252
• Diabetes -0.189 0.122 Italy
  -0.129
• Pain 0.131 -0.058 France
  0.310
• Arthritis 0.041 -0.066 Greece
  0.215
• Cancer -0.023 0.058 Belgium
  0.044
• Cesd score -0.035 -0.039
  
• Obese 0.223 -0.296 ,
  significant
• Age 58-64 -0.633 -0.399 at 5,10
  level
• Age 65-71 -1.285 -1.288
  
• Age 72 -1.815 -1.407

36
Simulated self-reported work limitations model
including employment equation
37
Employment rates
38
Percent working by disability category
39
Percent working by disability category
40
Conclusions 1

• Norms about what constitutes a work disability
  vary considerably across countries
• Elicitation of norms by vignettes is fairly
  noisy, but suggests some consistency across
  domains and the norms appear consistent with
  legal employment protection
• Differences in self-reports are at least partly a
  reflection of social norms, rather than of true
  disability differences
• Scale corrections make a difference for comparing
  work disability across countries

41
Conclusions 2

• Justification bias is significant in the US but
  not in SHARE-EU, reflecting different attitudes
  towards working
• Correcting for justification bias reduces the
  estimated effect of work disability on employment
  in the US, but not very much
• The relation between work disability and work is
  much stronger in the US than in the EU
• Other reasons than health reduce participation
  among older people in the EU more than in the US