سیاست های پرداخت کارفرما ، نقل و انتقال عمومی و تصمیم گیری های بازنشستگی مردان و زنان در دانمارک

The empirical retirement literature measures individual responses to variations in income flows due to public transfers, private individual or employer-provided pensions. We estimate a model accounting for the incentive effects from these sources. A dynamic structural model is extended to allow both individual and employer heterogeneity. This is applied to a Danish matched panel of workers and establishments, spanning a period of reforms to a public early retirement programme. Employer-specific compensation is found to be an important determinant of work and retirement income flows. Employer effects on retirement age are only found among sub-samples where access to public transfers is limited.

An ageing population, longer life expectancy and earlier exit from the labour force motivate interest in the determinants of retirement. Private pensions and public transfers are of increasing importance since they finance the future consumption of a growing number of retirees for a greater proportion of their lifetime. Policy interest in early retirement is almost always motivated by the large consequences for public financing of social security (Lumsdaine and Mitchell, 1999). It is less often remarked that early retirement is important to employers because of the human capital endowment that older workers represent. Pensions create incentives which have consequences for retirement behaviour. The extensive retirement literature has focused largely on the incentive effects of public transfer programmes (Gruber and Wise, 1999), somewhat less on employer-provided pensions (Gustman and Steinmeier, 1998) and more recently on personal pensions (Mitchell, 1999). The novelty of this paper is the coherent modelling of retirement incentives due to each source of pension income: personal-, employer- and publicly provided. The emphasis is on the compensation policy of different employers as a retirement determinant, while coherently modelling public transfer incentive effects and allowing individual income and retirement heterogeneity. An employer may wish to determine the retirement age of workers for several reasons. It would allow the writing of efficient long-term labour contracts and facilitate labour force adjustments. Theories of efficient labour contracts address the agency problem of asymmetric information regarding worker effort in the presence of monitoring costs (Lazear, 1979). An efficient contract is shown to involve a back-loaded or tilted tenure-remuneration profile. This increases the expected cost of shirking because alternative employment offers lower future remuneration. An important feature of back-loaded profiles is that low-tenure workers earn less than their productivity and high-tenure workers earn more than their productivity. The employer needs to limit the length of time during which older workers are able to earn above their productivity. This can be achieved through a mandatory retirement age or the accrual profile of a defined benefit employer-provided pension plan. Alternatively, employers may wish to induce early retirement rather than impose lay-offs when facing adverse demand shocks. Employers can carry this out by altering defined benefit pension accrual, or offering retirement plans which provide higher accrual for certain age groups for a limited time (Lumsdaine et al., 1992). Also, unexpected positive technology shocks may accelerate older workers’ skill obsolescence (Bartel and Sicherman, 1993). While retraining is costly in general, older workers are at a disadvantage relative to younger cohorts, since such an investment has to be recouped over a shorter period of time. These theories suggest that the distribution of productivity between establishments should be reflected in the distribution of retirement age between establishments. Hence establishment-specific policies are important determinants of individual retirement decisions. In particular, we expect a strong within-establishment correlation between retirement dates once other factors have been controlled for such as individual occupation, education and demographic characteristics. Nevertheless, where early retirement plans are financed by the State we would expect establishment-specific policies to have effects on individual decisions which are less pronounced or even non-existent. Indeed, if most workers are eligible for such programmes, it may be very costly for a single establishment to design incentive-compatible contracts which exhibit the desired features and survive the effect of a State funded scheme.1 Our objective is to estimate the contributions of publicly funded programmes and employer wage policies to individual retirement incentives. A conventional structural dynamic programming model of retirement is presented. This is extended to account for heterogeneity in retirement behaviour between individuals as well as between employers. The model is estimated on a matched panel of small-to-medium sized establishments and all their employees. Information on co-workers is used to account for establishment-specific wage policies which may contribute to workers decisions to retire early. Programme effects are identified by exploiting changes in the structure of a Danish publicly funded early retirement programme. We argue that eligibility to the programme is exogenous to the retirement decision. This provides income variation between otherwise similar individuals that can be used to measure public transfer incentive effects. The remainder of the paper is organised as follows: The next section sets up the economic model and describes the estimation method. Section 3 discusses the relevant background on retirement and pensions to motivate Denmark as of interest because of the institutional variation it provides which can inform measurement. Section 4 presents the dataset used and defines and describes the two variables of primary interest, retirement age and income. Section 5 presents, interprets and discusses the results. A final section summarises and concludes.

We have analysed the effect of between-establishment variation in remuneration policy on individual worker retirement ages. A conventional structural dynamic programming model is extended to allow employer unobserved heterogeneity to influence retirement decisions through income streams in-work and out-of-work. These employer effects are identified by exploiting a matched panel of Danish small-to-medium sized establishments and all their employees. Exogenous eligibility to a public early retirement programme identifies the effect of individual income flows on retirement. Several strong assumptions have been imposed in order to obtain our results, and it is important to restate them as caveats to any conclusions drawn. Exogeneity of programme eligibility at the time of the reform cannot be tested within sample since both programme and dataset began in 1979/80. Measurement of eligibility is indirectly through mandatory pension contributions and subject to errors to the extent that workers may change insurance fund membership status before 1980. We have ignored potentially important retirement determinants (for example, household decisions, health and saving) in order to focus on adding employer heterogeneity to a relatively simple model of individual decisions. Notwithstanding these caveats, pay policy is a potentially important instrument through which employers may influence workers retirement ages, together with mandatory retirement rules. The potential for this instrument is made clear since employer effects on income in-work are found to carry over strongly into out-of-work income. In the presence of efficient labour contracts, tilted tenure-remuneration profiles should be accompanied by employer-specific retirement policies, according to the between-employer productivity distribution. However, only for women do we find evidence of a direct establishment effect on retirement age, after controlling for observables. The relatively attractive public early retirement programme, which is available to the majority of male workers from the age of 60, may disguise or reduce the importance of establishment retirement policies. This transfer is not means-tested against private unearned income, and so employer-effects on non-work income remain, though they are not important enough to generate employer-effects on retirement age. The finding that employer effects on retirement age are only found among sub-samples where access to a public programme is relatively limited is consistent with the idea that generous public benefits make it too expensive for employers to write optimal incentive contracts