Training, technology and retirement behavior

1
Training, technology and retirement behavior

• Raymond Montizaan, Frank Cörvers and Andries de
  Grip
• PhD-Colloquium 10-01-2007

2
Introduction

• Aim of the paper
• Analyzing the effects of workers training
  history and technological change on
• retirement behavior

3
Literature /1

• The paper is related to two strands of
  literature
• Life cycle literature
• The existence of a productivity-wage gap for
  older workers
• Lazear (1979), Medoff and Abraham (1981),
  Kotlikoff and Gokhale (1992),
• Literature on technological change
• Gradual technological change and technological
  shocks
• Bartel and Sicherman (1993), Friedberg (2003)

4
Literature / 2

• Contribution of the paper to the existing
  literature
• Discerning between the effects of firm-specific
  training and general training history on
  retirement behavior
• Incorporating technological change at industrial
  level, while accounting for clustering in
  standard errors and unobserved heterogeneity

5
Theoretical framework / 1

• Firm-specific training leads to a
  productivity-wage gap for older workers
• Blinder (1982)
• Distinction between general and firm-specific
  training
• General training is paid by employee wage
  equals marginal productivity
• Firm-specific training costs shared by employer
  and employee
• Workers can still leave the firm after training
  investments
• Upward sloping earning profiles as an instrument
  for retaining workers with firm-specific training
• Upward sloping earning profiles lead to
  productivity-wage gap for older workers.

6
Theoretical framework / 2

• Technological change will affect the
  productivity-wage gap
• Bartel en Sicherman (1993)
• Gradual technological change may lead to later
  retirement
• Technological shocks lead to earlier retirement

7
Theoretical framework / 3

• Two testable hypotheses
• Workers with firm-specific training will retire
  earlier, because of a productivity-wage gap
  induced by backward loading earning profiles
• Gradual technological change induces later
  retirement, while technological shocks leads to
  earlier retirement

8
Data / 1

• National Longitudinal Survey of Older Men (NLSOM)
  and the Jorgenson KLEM database
• NLSOM
• Panel data of 5,020 men, aged between 45 and 59
  in 1966
• 12 waves until 1983
• Data incorporates
• Employment history (no firm characteristics)
• Personal and demographic characteristics
• 2,591 retirement transitions
• 1,582 death

9
Data / 2

• Dependent variable
• Retirement status based on main activity during
  the survey week
• Independent variables
• Training history indicators
• 1966 question if respondent had ever trained
• Firm-specific training history workers who ever
  received company sponsored training and still
  benefit from this training in their present work
• General training history workers who ever
  received training (other than company sponsored
  training, excluding training in armed forces)

10
Data \ 3

• Technological change indicators
• Matched data on rates of industry productivity
  (Jorgenson KLEM database)
• Technological change measured as rate of change
  not accounted for by growth in quantity and
  quality of physical and human capital (Solow
  residual)
• Two indicators based on Bartel and Sicherman
  (1993)
• Mean annual technological change over 10-period
  before period t (long run technological change)
• Technological shocks (Z-score)
• Control variables Individual characteristics,
  business cycle effects, retirement preferences

11
Data / 4

• Selected general descriptive statistics

12
Method / 1

• Estimating the effects of training and
  technological change on
• retirement behavior with
• A parameterized proportional hazard model with
• Weibull distribution
• Correction for unobserved heterogeneity (shared
  frailty term at industrial level)

13
Results / 1
Weibull regression results and robustness
(coefficients and robust standard errors in
parentheses )
14
Results / 2

• Robustness analyses
• Alternative training indicator including
  training participation after 1966 of workers
  younger than 53
• Correction training history indicators for job
  mobility.

15
Conclusion / 1

• Main findings
• Hypothesis 1
• Older workers with a firm-specific training
  history retire earlier than workers with a
  general training background or workers without
  training
• Hypothesis 2
• Opposed to the earlier findings of Bartel and
  Sicherman, a high pace of gradual technological
  change leads to earlier retirement

16
Conclusion / 2

• Implication of our analysis
• As our study demonstrates
• The effectiveness of institutional arrangements
• to postpone retirement also depends on the
  training policies of
• firms.