Work Orientations of Older Workers:

1
Work Orientations of Older Workers Change from
1992 to 2000 WAM-Net Seminar Orientations to
and Experiences of Work Cardiff School of Social
Sciences Cardiff University 15 October
2008 Michael White
2
Overview

• Some theory about older workers
• Some theory about change its effects on work
  orientations
• Modelling change
• Methods
• Results
• Conclusions

3
Older workers commitment and the employment
relationship some theory
Etzioni (19611975) employees in commercial
organizations have calculative commitment and
their employers have remunerative power over
them. Edwards, R. (1979) theory of
bureaucratic control and case evidence of
deferred compensation practices. Lazear (1981
1995) principal-agent theory of deferred
compensation for employees under costly
monitoring ? overpayment of older employees
relative to productivity or effort.
4
Marketization a changing relative position for
older workers?

• Sørensen (2000 2001) increase in market
  competition and short-term market relations tends
  to eliminate employee rents, notably those
  based on class.
• Lazear (1998) employment practices based on
  employer reputation (trust) can no longer be
  relied upon.
• Such arguments also suggest a deterioration in
  the terms of the relational contract for older
  employees.
• - e.g., reduction in deferred compensation levels
  or in indulgency toward lower productivity/effort
  with ageing ? LOWER COMMITMENT

5
Declining rents of older (45) employees?

• 1992 2000
• f. m. f. m.
• any unempt last 5 yrs. 4.9 9.8 11.7 12.1
• mean weekly hours 29.3 41.8 32.0 44.2
• high-pressure HRM practices 3.1 4.3 3.7
  4.8
• Sources Employment in Britain survey (1992),
  Working in Britain survey (2000).

But these observations have ve as well as ve
interpretations!
6
Assessing change in the commitment of older
employees

• Assume that 1990s was a period of increasing
  marketization and competition. The hypothesis is
  that the commitment of older employees declines
  over this period relative to less-old employees
  because the former had expectations of seniority
  rents that have not been met (and these
  expectations do not apply to the less-old group).
• A relevant analytical model is differences in
  differences see treatment effects literature,
  e.g. Heckman and Robb (1985).

7
The Differences in Differences (DiD) Model

• Ignoring individual subscripts, for the present
  case the OLS framework is
• Yß0 ß1age1 ß2time1 d1age1time1 u
• where Y is the outcome and age and time are
  dummies.
• Note that d1(Ya1,t1-Ya1,t0) (Ya0,t1-Ya0,t0)
• (Ya1,t1-Ya0,t1) (Ya1,t0-Ya0,t0)
• where the subscripted Ys are (age x time)
  subgroups mean outcomes.

8
DiD some properties and limitations

• DiD removes unobserved fixed effects that are
  invariant by time, and unobserved time-varying
  effects that are invariant by group.
• BUT to deal with group compositional changes that
  are time-varying, DiD also requires inclusion of
  appropriate control variables (X).
• HERE, participation is likely to be affected by
  labour market conditions, possibly to varying
  degree by age group. E.g., those with low
  commitment may be more likely to exit or be
  excluded in recessionary periods ? selectivity
  problem.

9
OLS and the alternative method of matching

• Can use OLS to correct selectivity bias if (a)
  selection is on observables (b) no inference to
  excluded subpopulation.
• OLS involves extensive assumptions. Matching is
  related to OLS, but with fewer assumptions.
• Cases that cannot be matched are omitted ? only
  like cases are compared. (cf. sample trimming in
  OLS)
• Matching has some advantages and some
  disadvantages compared with OLS regression. HERE,
  matching is included alongside OLS in exploratory
  mode as sensitivity analysis.
• See Heckman et al. 1997.

10
Methods 1

• Data national British samples of employees aged
  20-60 from Employment in Britain survey 1992
  (EIB) and Working in Britain Year 2000 survey
  (WIB).
• Age groups 20-44, 45-60.
• Chief Y variables Organizational commitment
  (6-item summative scale), Employment attachment
  (binary).
• Additional Y variables Various dissatisfaction
  measures.
• Selectivity variables (Xs) From standard labour
  supply concept - see later.
• Analysis (a) OLS regression?DiD (b) Exact
  matching ?DiD.

11
Methods 1a OC scale

• I would work harder than I have to, in order to
  help this organisation succeed
• I feel very little loyalty to this organisation
  reversed
• Id take almost any job to stay with this
  organisation
• I find that my values and the organisations
  values are very similar
• Im proud to be working for this organisation
• Id turn down a job with better pay to stay with
  this organisation
• Responses strongly agree, agree, disagree,
  strongly disagree dont know
• Cronbach alpha 0.79.
• Sources Mowday-Porter OCQ (see Price 1997)
  Lincoln and Kalleberg (1990).

12
Methods 2 labour supply variables

• Age group
• Gender
• Highest educational qualification (4 categories)
• Class (3 categories)
• Any unemployment in last 5 years (dummy)
• No partner partner not employed partner
  lower occupation partner higher occupation
• Dependent child in household (dummy)

13
Methods 3 exact matching

• Form cross-classification of the labour supply
  variables in each survey.
• Weight each EIB cell such that its weighted entry
  is equal to the corresponding unweighted WIB
  entry.
• Drop any cell with zero entry in either survey
  from both surveys (losing 5.9 of EIB and 5.6 of
  WIB).
• Compute marginal means, differences etc.
• This procedure reconstructs EIB to correspond to
  WIB on the chosen variables, including all
  high-order interactions.

14
Results (1) Older workers organizational
commitment OLS ? DiD
Notes. OC is 6-item summative measure, with
higher score? more positive attitude. Robust
estimation estimated standard errors in
parentheses.
15
Results (2) Older workers employment
attachment (EA) Probits ? DiD
Notes. EA is binary, 1would continue working
even if financially independent. Probit models
with robust estimation marginal effects,
standard errors in parentheses.
16
Results (3) Dissatisfaction measures OLS for DiD
Basic controls each cell separate analysis.
Coeffs. are over44 x 2000, highmore dissat.
Interval measurement assumption.
17
Results (4) Older workers organizational
commitment DiD, exact matching
18
Results (5) DiD-exact match v. DiD-regression
19
Conclusions - theory

• Organizational commitment for older (45-plus)
  employees fell relative to those aged 20-44, over
  the 1992-2000 period.
• There was a relative increase in dissatisfaction
  with hours and workload for the older group, over
  the same period.
• These findings were common to women and men.
• This is evidence for a (perceived) decline in the
  rents obtainable by older employees.
• But there was no reduction in employment
  attachment, and no relative increase in
  dissatisfaction with pay or security, for older
  employees.
• ?Relative deterioration in contract, not labour
  market, and on the side of effort, not reward.

20
Conclusions - Methods

• Adding variables to the OLS specification made
  negligible differences to estimates.
• Estimates obtained by exact matching were often
  different in magnitude (but not in sign) from
  those produced by OLS.
• Exact matching for DiD may offer a useful
  alternative or supplement to OLS-DiD where the
  number of relevant variables is small.

21
References

• Edwards, R. (1979) Contested Terrain The
  transformation of the workforce in the twentieth
  century, Basic Books.
• Etzioni, A. (1975 1961) A Comparative Analysis
  of Complex Organizations, revised edition, Free
  Press.
• Heckman, J., Ichimura, H. and Todd, P. (1997)
  Matching as an econometric evaluation estimator,
  Review of Economic Studies, 65 261-294.
• Heckman, J. and Robb, R. (1985) Alternative
  methods for evaluating the impact of
  interventions, in Heckman, J. and Singer, B.
  (eds.) Longitudinal Analysis of Labor Market
  Data, Cambridge University Press.
• Lazear, E. (1981) Agency, earnings profiles,
  productivity, and hours restrictions, American
  Economic Review, 71 606-20.
• -- (1995) Personnel Economics, MIT Press.
• -- (1998) Personnel Economics for Managers,
  Wiley.
• Price, J.L. (1997) Handbook of organizational
  measurement, International Journal of Manpower,
  18 303-553.
• Sørensen, A.B. (2000) Toward a sounder basis for
  class analysis, American Journal of Sociology,
  105 1523-58.
• -- (2001) Careers and employment relations, in
  Berg, I. and Kalleberg, A.L. (eds.) Sourcebook of
  Labour Markets, Kluwer Academic/Plenum.
• For EIB survey, see Gallie, D., White, M., Cheng,
  Y. and Tomlinson, M. Restructuring the Employment
  Relationship, Oxford University Press, 1998.
• For WIB survey, see McGovern, P., Hill, S.,
  Mills, C. and White, M. Market, Class, and
  Employment, Oxford University Press, 2007.

22
www.psi.org.uk