Between 1960 and 1990, the population of the European Community grew by about 17%, but it is expected to fall by some 2% by the year 2025, while that of Eastern and Western Europe taken together will rise by some 3-4%. This contrast reflects both the growing number of pensioners in the Community and the rapid ageing of its working population, which may call for policies to improve the quantity or quality of the labour force. These may include incentives to female participation, selective immigration policies, `pro-natalist' family policies, and improvements to human capital formation through schooling and further education. Many of these issues were discussed at a conference on `Labour Market Implications of European Ageing', held in Munich on 23/25 April. Demographers and labour economists investigated these policy options on the basis of European ageing's likely effects on productivity, wages, mobility, unemployment and educational activity. The conference was organized by Paul Johnson, Lecturer in Social History at the London School of Economics, and Klaus F Zimmermann, Professor of Economics and Director of the Seminar for Labor and Population Economics at the Universität München, respectively Research Fellow in and Co-Director of CEPR's Human Resources programme. CEPR gratefully acknowledges the financial support provided by Directorate-General V of the Commission of the European Communities (Employment, Industrial Relations and Social Affairs).
Paul Johnson opened the conference with his paper, `Ageing and European Economic Demography', which noted that the proportion of the total European population aged over 65 is expected to rise from roughly 13% today to some 20% in the year 2025. Total fertility has fallen dramatically since the `baby boom' of the 1950s to well below replacement levels; child and adult mortality rates have also declined, while average life expectancies have risen by up to ten years. The richer countries of Northern and Western Europe tended to lead their Southern and Eastern neighbours in these changes, but this distinction is much less clear-cut than that between Europe as a whole and its near neighbours. Turkey and Morocco have fertility rates well above replacement levels, with less than 4% of their populations over 65.

Johnson argued that traditional fertility models based on the need for security in old age are not relevant to post-war Europe, while economic explanations based on increased female participation cannot fully account for such non-marginal changes. Existing explanations of mortality trends are also inadequate: economic growth appears to raise life expectancy, but the relationship is weak and may operate through improvements in long-term environmental and nutritional conditions or better access to health care in old age. While these trends have major implications for social security systems and hence for public finance, general equilibrium attempts to model their interaction with labour markets depend critically on assumptions about savings behaviour that have little empirical support, and which even marginal changes in assumptions concerning technological progress can easily dominate over 40-50 years.

Rudolf Andorka (University of Economic Sciences, Budapest) noted that Hungary led the rest of Eastern Europe in the fertility decline. To offset labour force ageing, he recommended active policies to increase female labour market participation at all levels, increased but controlled immigration, and generous, `pro- natalist' child allowances. John Ermisch (University of Glasgow and CEPR) noted that a one-time migration impulse could not offset changes to age structure, which required a steady stream of migrants over several years. Heather Joshi (London School of Hygiene and Tropical Medicine and CEPR) stressed that children's education and socialization require social as well as private inputs which can also assimilate migrants; host country governments should not regard migrants as `inferior' substitutes for native labour but seek to reap the benefits of the multicultural contact that migration entails.

Ageing and Employment Trends

Stanislawa Golinowska (Institute of Labour and Social Studies, Warsaw) presented her paper, `Ageing of Labour Resources in Central and Eastern Europe: First Stage in Ageing of Societies in this Region', which focused in particular on Poland. Eastern Europe's socialist governments used a wide range of family policy instruments to promote fertility, including extensive child care provision and maternity leave, family allowances and subsidies to a variety of children's goods, which in Poland were supplemented by `pro-natalist' pressure from the church. UN projections indicate that Poland's population will grow from some 7.7% of the European total today to some 8.2% in 2010. It has one of the highest fertility rates in Europe (although its rate of growth is now falling), while life expectancy is some 6-7 years lower than the EC average. These trends pose potentially major migration problems for Europe: Poland has a long history of emigration and a diaspora of some 15 million; and there are now increased incentives to migrate westwards above all to Germany in response to economic crisis and rising unemployment, although the political incentives to do so have clearly diminished. Poland will be unable to absorb the expected growth of its productive labour force and also cope with the transformation to a market economy in the next 20 years. Golinowska concluded that foreign investment is essential both to reduce the pressure for migration and to preserve economic and hence political stability.

Walter Krämer (Universität Dortmund) maintained that the recent decline in older men's life expectancy does not indicate poor health care; German and Soviet evidence suggests that `healthier' male population segments of certain cohorts died during World War II, and a similar statistical artefact should be expected for Poland. Rudolf Andorka disagreed, noting that male life expectancy has recently fallen in Czechoslovakia, Hungary and Romania, which all suffered much smaller wartime population losses, while life expectancy in East Germany has fallen relative to the West. Naohiro Yasihiro (Japan Center for Economic Research, Tokyo) noted the strong correlation of the rise in life expectancy with increased income and improved nutrition and health care in Japan. Large-scale post-war population movements may account for these results in Europe if the healthier (surviving) population segments migrated.

Paul Johnson then presented the paper, `Ageing and Employment Trends', by Martin Rein and Klaus Jacobs, which examined employment trends for men aged 55-64 in Western Europe, the US and Japan. Retirement is not a dichotomous choice: some continue to work in the same jobs with no change in status, wages or working hours; some drop out of the labour market completely; still others turn to a transitional `bridging' phase. During 1970-90, inter-country differences tended to increase; France, Germany and the Netherlands were characterized by `high early exit', with some 20-25% of men aged 60-64 in paid employment; in Japan, Sweden, the UK and US such employment activity rates were around 50%.

Self-employment is much higher for men aged 60-64 than for those aged 55-59. This may reflect a cohort effect (if self-employment was more common when they began their careers), compulsory retirement at age 60, the exclusion of the self-employed from routes to retirement available to those in wage or salary positions, or their deliberate choice to continue working as a `bridge' to retirement from such positions. The latter form of `early retirement' is common in Japan, Sweden and the US but governed by three very different processes. Japan has low and strict mandatory retirement ages, but retirees are often rehired by the same or an affiliated firm on less secure employment contracts; ageing may place pressure, however, on the sense of `social obligation' that underpins this system. In Sweden, a system of partial pension arrangements allows retiring wage and salary recipients to recoup much of their `lost' income by continuing to work part-time. In the US, early retirees are left to their own initiative, but poor public pension provision obliges many to work to cushion the effects of abrupt exit on earnings; while legislation against age discrimination enhances their chances of finding such work.

Heather Joshi suggested using longitudinal data for a proper cohort analysis instead of the authors' cross-sectional approach. Further study of the effects of domestic commitments may indicate whether men retire early to care for spouses or other members of the extended family. This is also the age range in which `early death' has major differential effects on the data depending on employment histories; the underprivileged, unhealthy and unskilled often die before ever drawing a pension.

Age, Training and Productivity

Christopher Flinn (New York University and International Centre for Economic Research, Torino) presented his paper, `The Implications of Cohort Size for Human Capital Investment'. He examined the effects of cohort size changes on the on-the-job human capital investment of young labour market participants in the US over the past 50 years and the wealth levels achieved by the various cohorts. In his perfect foresight partial equilibrium model, all cohorts' human capital investment decisions are uniquely determined, labour market participants within each cohort are defined to be identical, they concentrate all their human capital investment in their first period of employment and have perfect information concerning demand shocks and cohort size. Flinn used this model to estimate the direct effects of cohort size changes on the distribution of welfare across cohorts for US white males during 1880-2010. He found that cohort wealth is quite insensitive to levels of human capital investment, which is itself quite responsive to changes in the cohort size sequence. Estimating a second model, in which workers assume (irrationally) that the rental rates on human capital will remain constant over their lifetimes, yielded a similar intergenerational wealth distribution, thus further emphasizing the latter's stability with respect to changes in investment levels.

Ken Burdett (University of Essex) pointed out that Flinn's assumption that only one type of human capital exists and the dynamic structure in which past and future investment decisions determine the returns on current investment implies the possibility of intergenerational conflict. Different cohorts' investments in human capital may be complementary, however, as technology changes: for example, if one generation invents electricity, a succeeding generation can become computer programmers.

In his paper, `Does an Ageing Labour Force Call for Large Adjustments in Training or Wage Policies?', Didier Blanchet (Institut National d'Etudes Démographiques, Paris) noted that on- the-job training increases productivity in the later stages of workers' careers, so ageing due to lower population growth makes schooling and on-the-job training more attractive. Schooling is costly and relatively unrewarding when younger cohorts dominate, but its collective cost diminishes in an ageing society and allows a longer initial period of human capital accumulation. Ageing can therefore be a positive factor rather than a problem for productivity, and the remaining issue is how to determine the `right' amount of human capital investment to maintain a relatively constant total life-cycle production. As a population ages, career profiles also change; if the labour hierarchy is linked to age and seniority, workers will reach the higher levels later in life than in a growing society. Blanchet's simulation of a reduction in the population's growth rate from +1% to 1% raised the mid-career age by four years, although there was a higher rate of wage progression at the end of their careers. A fall in the population growth rate also implies a slower wage progression. Strong resistance by workers to any such changes in their age-earnings profiles will tempt firms to encourage early retirement.

Paul Chapman (National Economic Development Office, London) questioned the absence of any discussion of labour force entry in the paper and noted the problems of determining whether some workers were active or unemployed and incorporating both full- and part-time work. Naohiro Yasihiro disputed Blanchet's assumption that the return to on-the-job training is flat across age groups. Firms find it less profitable to train older workers, who may also be less easy to train. The Japanese practice of employment rotation within the firm is much more conducive to on- the-job training, however, than the UK system, where workers tend to remain in the same jobs for extended periods.

Ageing, Wages and Unemployment

Coen Teulings (Universiteit van Amsterdam) opened the next session with his paper, `The Relations between Wages, Age and Education in the Netherlands', written jointly with Joop Hartog and Hessel Oosterbeek. During the last three decades, the average age and educational level of the Dutch labour force have increased substantially and both these trends have been stronger for females than for males. The observed increase in the demand for education was not associated with any substantial changes in the expected structure of earnings and cannot therefore be explained by human capital theory alone; it may be attributed to increasing parental incomes or government subsidies. Teulings presented estimated wage equations that displayed a declining return on human capital and a constant return on experience, while the variance of earnings within age and educational groups was quite large. Earnings also increased with a lagged variable for cohort size, contrary to intuition, but including experience and age changed the sign of this cohort size effect, which suggests that seniority is less important in the Netherlands than it appears.

Robert Wright (University of Glasgow) suggested extending the analysis to consider inexperienced-young and experienced-old workers as complements, with the latters' relative wages falling as the population ages, and also including the effects of unemployment. Winfried Schmähl (Zentrum für Sozialpolitik, Bremen) suggested extending the model to include variables to capture the effects of on-the-job training and retraining.

Anders Klevmarken (Gothenburg University) closed the morning session with his paper, `On Ageing and Earnings: A Review', which reviewed the existing empirical evidence on shifts in age- earnings profiles in response to changes in labour supply particularly cohort effects and demand. He then estimated an earnings model on three waves of panel data for Swedish males from the 1980s. Its results indicated that schooling, age and the market return to human capital investment significantly affected individual earnings, while work experience and cohort size effects proved insignificant, although it was unreasonable to expect precise results on cohort size from only three data waves. He then presented a second model which estimated `earnings mobility', or movements towards or away from an average earnings profile, on the same data set. This model excluded the cohort size variable but employed a complex error structure, with separate variables to capture `heterogeneity', or individuals' permanent deviations from the average earnings profile, and the residual `temporary' deviations from such a profile. The variance of heterogeneity increased with age, while temporary variance was largest at the beginning of workers' careers.

Axel Börsch-Supan (Universität Mannheim and CEPR) noted that Klevmarken's theoretical discussion allowed earnings to depend on career choice, participation and human capital returns, but his empirical work was relatively limited. He proposed enhancing the model's structure by incorporating seniority theory into the model and applying it to a fuller data set. Christopher Flinn suggested testing whether schooling varied with the sample period to see if it was endogenous.

Ageing, Migration and Unemployment

Klaus F Zimmermann opened the afternoon session with his paper, `Ageing, Migration and Labour Mobility', written with Rainer Winkelmann, which examined the role of migrants and their effects on population ageing in Germany. There are currently some 20% more younger than older workers in the German (and the European) labour force, but these proportions are expected to reverse in the next 30 years; the share of foreigners in the German labour market is around 8%, comparable to that of foreign-born workers in the US, so the German experience may serve as a useful case- study for Europe. Zimmermann examined labour mobility defined as frequency of job change and/or unemployment spells for Germans and foreigners separately and outlined a theoretical model of the demand for quantity and quality of labour and its implications for relative wages, unemployment and direct job changes.
Estimating this model on a large German micro data set, he showed that foreign workers are more flexible in changing jobs but more often unemployed at older age levels. Both groups exhibited U-shaped relationships between age and frequency of unemployment; foreigners faced a lower risk of unemployment until the age of around 30 and a higher risk in older age groups. German workers' direct job changes declined with age whereas foreigners' direct job changes declined with duration of stay. Increased migration was likely to raise unemployment but have only limited effects on direct job changes. Simulations of expected age structures for the period 1995-2020 indicated that both measures of labour mobility will initially decrease and then increase in Germany, while the age structure will lead to a much flatter pattern of labour mobility for the Community as a whole.

Gerd Ronning (Universität Konstanz) noted that Germany's post-war experience of immigration called for a precise definition of `foreigners' in Germany. He also questioned the accuracy of the retrospective questions about job changes and unemployment used in the questionnaire. There may now be substantial differences between the behaviour of foreigners from EC countries and those from elsewhere, such as Turks and Yugoslavs. Knut Gerlach (Universität Hannover) suggested that foreigners' apparently greater labour market flexibility may reflect the sorts of jobs they do; he suggested comparing the flexibility of foreigners and natives in comparable jobs or at least at the industry level.

Christoph Schmidt (Universität München) presented the final paper of the conference, `Ageing, Migration, and Unemployment', which assessed the effects of changes in age composition on the incidence of unemployment in different age groups in West Germany during 1950-89. An ageing population may overwhelm social security resources and lead to both unemployment and underemployment. Germany's younger cohorts may have to support not only the ever-growing pool of pensioners, but also the relatively high proportion of the expanding older cohort that is unemployed, even if the trend towards `early retirement' can be halted or reversed. Immigration is often cited as a remedy for population ageing, and indeed it has augmented the younger cohorts in Germany; but large-scale immigration could lead to relatively high unemployment for the younger cohorts if generational crowding is important.

In his estimations of age-specific unemployment rates as functions of overall trend and idiosyncratic components, however, Schmidt found no strong cohort size effects for specific age groups, except for the very young; migration will have only a small impact on these, equivalent to augmenting the younger cohorts with exceptionally high birth rates. Indeed, under the German system of centralized bargaining, with powerful unions and an unchanged wage structure, larger cohorts may even be able to reduce their relative unemployment rates, once they are established in the labour market.

John Ermisch welcomed Schmidt's incorporation of strong unions in the model, but he criticized the assumption that migration is exogenous; emigration from Ireland to the UK may perhaps be explained entirely by relative unemployment rates. He also suggested excluding the trend term, which is rarely significant, to allow easier cointegration testing. Walter Krämer suggested weighting the age-specific unemployment rates by the proportions of the labour force that actually run the risk of becoming unemployed since civil servants and the (predominantly young) military cannot become unemployed involuntarily.


The papers presented at this conference will be published early next year, in a volume to be edited by Paul Johnson and Klaus F Zimmermann.