Presentation is loading. Please wait.

Presentation is loading. Please wait.

Richard Disney* †, Carl Emmerson*, Gemma Tetlow* † University of Nottingham *Institute for Fiscal Studies, London September 24 th 2010 OME Total Remuneration.

Similar presentations


Presentation on theme: "Richard Disney* †, Carl Emmerson*, Gemma Tetlow* † University of Nottingham *Institute for Fiscal Studies, London September 24 th 2010 OME Total Remuneration."— Presentation transcript:

1 Richard Disney* †, Carl Emmerson*, Gemma Tetlow* † University of Nottingham *Institute for Fiscal Studies, London September 24 th 2010 OME Total Remuneration Conference Pension Rights and Job Mobility in Britain

2 Summary of talk How does membership of pension plan affect job mobility? –DB v DC plan incentives –Role of prospective pension rights The paper constructs measures of prospective pension rights calculated from wave 11 of British Household Panel Survey It then examines probability that workers leave jobs conditional on their coverage by, membership of, and prospective pension rights in, their pension plan. (Note: covered workers do not have to join their plan in the UK)

3 Summary of talk (continued) Main findings: –Members of DB pension plans have longer tenures than other workers. –Public sector workers have longer tenures than private sector workers even controlling for plan membership. –But workers with any kind of pension plan (including Personal Pension) having longer job tenures than workers without a private pension. Policy applications: –Labour market ‘flexibility’ (pension portability) – good or bad? –Implications of recent shift to DC plans.

4 Why do employers offer DB (or DC) plans? In a ‘spot’ market, workers are paid a wage reflecting their productivity at each point in the life cycle. But sometimes productivity is hard to measure (e.g. public sector, ‘teams’ etc). So employers can ‘defer’ (backload) pay to create incentives for workers not to shirk, or to quit. DB pensions (esp. ‘final salary’) are classic example – but even an employer-provided DC plan may provide ‘signal’ of ‘employer welfare policy’. Pensions may be provided for other reasons e.g. tax regime. In 1980s, idea arose that ‘labour market flexibility’ was desirable – ‘too long tenures’ were not ‘dynamic’ portable (Personal) pensions. ‘New’ ways of incentivising? e.g. measuring performance/bonuses?

5 Explaining association of job mobility to pension plan tenure: Incentive ‘story’ Why do workers in DB (defined benefit) plans face disincentives to quit? –‘backloading’ of pension accruals if pension related to final earnings and earnings are (broadly) rising through career –Costs of pension transfers (e.g. valuation in new plan lower than in old) –‘cliff vesting’ (not important in UK) –Protection from employer’s strategic firing of ‘high pension cost’ workers

6 A demonstration of ‘backloading’ (constant earnings growth, final salary basis, zero real growth of ‘deferred benefits’)

7 Predictions of ‘incentive’ story Quit/exit rates should be negatively related to prospective loss of pension benefits (not accrued rights, see Ippolito, 1997) –But these prospective losses not always measured carefully in previous studies Lifetime wages should compensate for deferred remuneration? (existing tests not convincing; loss of coverage associated with lower wages) Quit/exit rates from plans that don’t penalise (DC plans) should be higher than from those that do (DB plans). But this not always true e.g. Gustman & Steinmeier (1993) on US SIPP.

8 Source: BHPS 1992-1999

9 Data We utilise the British Household Panel Survey A special module in 2001 asked about tenure in pension plan and whether DB or DC plan Our core sample in this paper is all aged 20 to 59 in 2001 sample who are in employment in both waves 11 and 12 of the data (around 3560 individuals) We also know earnings, demographics, job changing behaviour… Note around 25% of workers covered by a pension plan may not be a member of it.

10 Pension coverage, by age, men

11 Pension coverage, by age, women

12 Our empirical strategy We calculate earnings profiles by age for educational groups and by gender We calculate prospective accruals of pension rights based on: –Whether individual is a member of a DB or a DC plan –Accrual fraction –Mortality assumptions/discount rate (2% real) etc We make alternative forecasts of earnings growth (but note our prospective earnings profiles need not be linear – for some people, growth of prospective pension rights from 1 more year’s work could be negative.) In illustrated results here, we don’t have different profiles for public and private sector workers, we assume no pensions are portable.

13 Simulated median log male earnings, by education, under two different specifications for age.

14 Simulated median log female earnings, by education, under two different specifications for age

15 Cumulative distribution of simulated expected one period pension loss among those in defined benefit pension arrangements, by simulated expected wage growth Notes: The density functions are estimated using an Epanechnikov kernel with a band width of £100. They are unweighted. Earnings growth assumptions: A = 2% per year B = 2% per year plus age, education and sex specific component C = 2% per year plus age and education specific component estimated off men only D = 2% per year plus age, lagged wage and twice lagged wage component E = As D but those individuals with estimated wage changes> +/–30% dropped.

16 Cumulative distribution of simulated expected pension loss at the maximum value among those in defined benefit pension arrangements, by simulated expected wage growth

17 Impact of prospective pension losses on job mobility: Alternative specifications of wage growth Estimation method: probit; marginal effects are cited. LHS variable: Probability of moving job between waves 11 and 12 of BHPS (2001/2002). **=1%, *=5% significance. Additional controls: Full-time, gender, quadratic in age, cubic in pension tenure, public sector, quadratic in time to travel to work, age-gender interactions, working status of partner, ethnicity, full sets of dummies for education, industry, firm size and region. Construction of earnings growth and pension loss variables as explained in test: this specification, one year loss if individual moves from one pension-covered job to another.

18 Mean pension loss by whether or not changed jobs: All individuals, and by members of DB plans only

19 Evidence on portability: testing for differences in pension losses on coefficients between public and private sector workers

20 What is the counterfactual? These results show strong association between prospective pension accruals and DB plan membership on one hand, and job mobility on the other. Suppose DB plans were increasingly replaced by DC plans, or no plans at all (other than Pension Accounts), could we conclude that job mobility rates would increase? E.g. if we got rid of Civil Service Plan (with or without compensating increase in civil service pay?) what effect on quit rates? It’s not that simple to ‘read off’ the result – people may ‘select in’ to certain jobs, and employers may select certain workers. Differences in exit/quit rates may not wholly reflect incentives, but also these selection processes.

21 Estimates of pension losses with selection into pension plan coverage One year loss with pension coverage in new job, Model A of earnings growth Notes: Probit coefficients cited (not marginal effects). Standard errors not cited: **=1% significance, *=5% significance. Estimation: column (1) probit; columns (2) and (3) Heckman probit. ‘Industry firm size mean offer rate’ = mean offer rate in industry of respondent and firm size demarcated as ‘large’ or ‘small’ if <>25 employees. N=2053 = number of uncensored observations. Additional controls: quadratic in time to travel to work, age-gender interactions, working status of partner, ethnicity, full sets of dummies for education, industry, firm size and region.

22 Conclusions New features vis-à-vis previous studies: –Careful measures of heterogeneity of pension losses among plan members in testing ‘incentive’ effect. –Exploiting UK choice to join/not join employer-provided pension plan gives new test of factors that cause individuals to self-select into plans There’s strong evidence of an incentive effect But also of selection (e.g. coverage matters whether you join or not + DC plan members job tenures also affected Policy implications –Reform of pensions e.g. public sector pensions –Should people be allowed not to join employer’s plan?


Download ppt "Richard Disney* †, Carl Emmerson*, Gemma Tetlow* † University of Nottingham *Institute for Fiscal Studies, London September 24 th 2010 OME Total Remuneration."

Similar presentations


Ads by Google