Demographic change, immigration, and the labour market: A European perspective* by Juan F. Jimeno** DOCUMENTO DE TRABAJO 2004-18

September 2004

* This paper was written while I was visiting DG ECFIN under its Visiting Fellowship Programme. I am grateful for the hospitality. I also acknowledge financial support from the Spanish Ministry of Science and Technology (grant SEC 2001-0061). ** Universidad de Alcala, FEDEA, CEPR and IZA

Los Documentos de trabajo se distribuyen gratuitamente a las Universidades e Instituciones de Investigación que lo solicitan. No obstante están disponibles en texto completo a través de Internet: http://www.fedea.es/hojas/publicaciones.html#Documentos de Trabajo

Abstract

After a long period of high unemployment, the EU is about to face a significant change in the demographic structure of its labour force, due to a reduction in fertility rates in the past, and increasing immigration flows. There is a long standing literature of empirical studies aiming at measuring the effects of cohort sizes and of immigration flows on employment and unemployment rates and on the wage profiles of several population groups. And there are some reasons to think that these effects depend on the institutions determining the functioning of the labour market.

This paper argues that population ageing may produce a reduction of employment rates in the EU15 over the next two decades, as the share of the older workers in the labour force increase. Then it discusses the reasons why, despite this direct composition effects, there may be another indirect effects of changing composition of the labour supply on population specific employment and unemployment rates. Finally, it uses cross-country data to find how the interaction between the age composition of the labour force and the share of foreign workers in the labour force, on the one hand, and labour market institutions, on the other hand, contribute to explaining international differences in age and gender-specific employment and unemployment rates.

JEL Codes: J11, J64

Keywords: demographics, unemployment, immigration

1. Introduction

During the next decades the age population structure of the EU is going to change drastically. Usually, this demographic change is highlighted using dependency ratios (the inverse of the ratio of the population in working age respect to the population not in working age). Less mentioned is the fact that the age composition of the population in working age is also about to experience a significant change, with likely consequences on many aspects of the labour market, such as employment and unemployment rates, the structure of wages and productivity growth.

Another relevant demographic fact in this respect is that immigration flows to the EU have accelerated during the last decade, and, thus, the weight of foreign population has significantly increased. In fact, as shown in Figure 1, in 2001 70% of population growth in the EU is accounted for by immigration flows, a proportion which has followed an increasing trend in many of the Member States. Immigrants are relatively younger than the national EU population, have lower levels of education, and seem to display higher mobility rates, both in terms of geography and in terms of the sectors and occupations where they fill jobs.

Thus, both the size and the age composition of the EU labour supply are rapidly changing. Usually, economists work with labour market models in which labour supply is neutral in the long-run with respect to equilibrium unemployment. It is also usual to work under the assumption that changes in the composition of labour supply are arbitraged away by changes in relative wages, so that the differences between group-specific employment/unemployment rates are also invariant in the long run.

There is however a long standing literature, going back to Easterlin (1968), suggesting that individuals’ labour market prospects may be affected by the cohort size to which they belong. According to some results in this literature, larger cohorts enjoy relatively lower lifetime earnings and lower employment probabilities than smaller cohorts. In the US, this hypothesis has been widely tested with regard to labour earnings. For instance, Welch (1979) and Johnson (1980) found that cohort size depresses wages, especially for highly educated workers. Katz and Murphy (1992) find that, from 1965 to 1980, wages tended to decrease for a group of workers when the size of that group increased. Card and Lemieux (2001) study the factor behind earnings inequality in US, UK and Canada, concluding that cohort effects have good potential to fit the data showing an increasing education premium for the recent smaller cohorts of young graduates.

Another branch of this literature has looked at age-specific employment and unemployment rates. In a cross-country study Korenman and Neumark (2000) found that the elasticity of youth unemployment rates to the cohort relative size is around 0.5. However, using regional data for the US, Shimer (2001) estimated that an increase in the youth share of the working age population reduces the youth unemployment rate, with an elasticity of about – 1.5, while the elasticity of the prime age unemployment rate with respect to the youth share of the working age population is around –2. Jimeno and Rodriguez-Palenzuela (2002) use a sample of OECD countries to estimate the relative relevance of demographics, labour market institutions and macroeconomic shocks at explaining youth unemployment, finding that both in the short run and in the medium run youth unemployment increases with the relative size of the youth population. Similarly, Bertola, Blau, and Kahn (2002a) show that demographic shocks interacted with labour market institutions contribute to explaining the difference in the aggregate unemployment rate and in the relative employment rates of young and female workers between the US and some EU countries.

As for the impact of immigration on the labour market, there is another large literature that searches for the effects of immigrants on the labour market prospects of native-born workers. The overall conclusion is that these effects, as far as wages and employment rates are concerned, are not very significant. One explanation for this finding is that immigrants compete for jobs with low skilled native workers but, since they are more mobile and willing to accept jobs which the natives are not willing to accept, the employment rates and wages of low skilled native workers are not very much affected.²

The impact of demographics on the labour market is not only an issue of interest for labour economists. It is also a policy relevant issue for, at least, two reasons. First, from a European perspective, the EU has given itself very ambitious targets for employment rates in 2010.³ The most intense (and predictable) change about to occur in EU labour markets is the ageing of the working-age population. Hence, were this to have an effect on aggregate unemployment/employment rates, the policies needed to achieve this target should take into account the effects of ageing. Secondly, the effects of population ageing are bound to produce fundamental pension reforms across many EU countries. To the extent that pension benefits are, in some manner or another, related to employment and wage profiles, the evaluation of alternative pension reforms should take into account the effects of population ageing on the labour market performance of alternative population groups.

¹ See, for instance, Card (1990), Borjas (1994), and Borjas, Freeman and Katz (1996), for the US evidence, and Friedberg and Hunt (1995) and Bauer and Zimmermann (1999), for the evidence on Europe.

² This possibility is stressed by Card (2001).

³ These targets were established in the Lisbon’s and the Stockholm summits. The aggregate employment rate should as closer as possible to 70%, while the female employment rate should be higher than 60%. Moreover the target for the employment rate of workers aged 55-64 is 50%.

This paper focuses, from a European perspective, on one dimension of the impact of demographics on the labour market, namely age-specific employment and unemployment. Its structure is as follows. Section 2 comments on simple calculations showing the consequences of the ageing of working age population in regard to aggregate employment/unemployment rates under the assumption that age-specific employment/unemployment rates remain constant at the levels observed in 2001. Section 3 moves to theoretical arguments on why it is unlikely that this assumption will hold. Section 4 presents empirical evidence on the relationship between age-specific employment/unemployment rates, the age composition of the labour force, immigration flows, and labour market institutions. Finally, Section 5 concludes with some closing remarks.

2. Demographics, employment, and unemployment: Composition effects

The aggregate employment/unemployment rate is a weighted average of the employment/unemployment rates of several population groups. To the extent that there is substantial variation of employment/unemployment rates across different population groups, changes in the weights of each group in the working age population and in the labour force, respectively, will yield changes in the aggregate rates, even though the group-specific rates remain constant.

My goal here is to assess the changes in the aggregate employment and unemployment rates which would take place due to the future evolution of the age-composition of the population, under the assumption of constant population group specific employment and unemployment rates. I consider 18 population groups distinguishing gender, age (15-24, 25-54, and 55-64), and three educational levels (low, medium, and high).

For the weights of each group in the total population I rely on EUROSTAT’s data regarding the latest national population forecasts by five years age groups.⁵ As can be seen in Figure 2, in the whole EU15 the shares of youth population in the working age population will fall by about 1.5 percentage points in the next two decades. The shares of older persons (55-64) in the working age population will rise by about 5 percentage points. According to these forecasts, the Member States in which the working age population will age more rapidly are Belgium, France, Spain, and Finland.

⁴ The educational levels are defined according to ISCED97 classification : ISCED0_2 (Pre-primary) : low, ISCED3-4 (Upper secondary and post-secondary, non-tertiary education): medium, ISCED5_6 (Tertiary education) : high.

⁵ See EUROSTAT (2001).

Since unemployment and employment rates differ markedly by gender, age and education, a change in the relative weights of each population group may have a significant composition impact on the aggregate rates. As seen in Figure 3, for the whole EU15, in 2001 the unemployment rate of older male workers was about 2 percentage points higher than for prime age male workers, while they are roughly similar for women of both age groups. Youth unemployment rates were about 8 percentage points higher than for prime age workers, both in the case of men and in the case of women. These differences were significantly higher in Southern Europe (excluding Portugal), France, Belgium, and Finland. As for the employment rates, the differences across population groups were even larger. In the EU15, the employment rates of youth and older individuals were about 40 percentage points lower than for prime age individuals, in the case of men, and about 30 percentage points, in the case of women. The countries with largest difference between the prime age employment rate and that of older workers were Belgium, Italy, Luxembourg, France, and Austria.

Thus, assuming constant age-specific employment and unemployment rates, as the share of youth population in the labour force decreases, the aggregate unemployment rate will fall. Similarly, as the share of youth population in the working age population decreases, the aggregate employment rate will rise. But at the other extreme of the age distribution the composition effects have the opposite sign. As the shares of older individuals in the working age population and in the labour force increase, the aggregate unemployment rate will rise and the aggregate employment rate will fall. In principle, the facts that the educational levels of the younger cohorts of Europeans are higher than those of previous generations, and that unemployment (employment) rates are decreasing (increasing) in education may compensate for the rise of the relative weights in the population and in the labour force of the older individuals.

The sum of all these composition effects can be easily quantified. Let the weights in working-age population and in the labour force, respectively, of group I in 2001. Let and be the unemployment and employment rates, respectively, of group I in 2001. Then, the aggregate unemployment and employment rates are respectively given by

\[u _ {2 0 0 1} = \sum_ {I} \alpha_ {I, 2 0 0 1} ^ {L} u _ {I, 2 0 0 1} \quad \mathrm{e} _ {2 0 0 1} = \sum_ {I} \alpha_ {I, 2 0 0 1} ^ {P} e _ {I, 2 0 0 1}\]

Under the assumption that in the future each group-specific unemployment/employment rates remain constant at their 2001 level, then the corresponding rates in year T would be

\[u _ {T} = \sum_ {I} \alpha_ {I, T} ^ {L} u _ {I, 2 0 0 1} \quad \mathrm{e} _ {T} = \sum_ {I} \alpha_ {I, T} ^ {P} e _ {I, 2 0 0 1}\]

The two panels of Table 1 give the results of these calculations for the EU Member States in 2010 and 2020. The first horizon is relevant since that is the deadline for the fulfilment of the Lisbon’s summit targets. The second horizon is also relevant since at that time population ageing will start to accelerate even more in many EU countries.

As for the unemployment rate, the changing age composition of the labour force would have a rather minor effect on the aggregate unemployment rate, which would remain at about 7.4% in the EU15. In some countries, like Spain, where the baby boom took place later than in the other EU15 Member States, this composition effect would produce a negative trend in the aggregate unemployment rate that would fall by about 1 percentage point in both countries between 2000 and 2020. With regard to the aggregate employment rate, the conclusion is more negative. As a result of population ageing, the aggregate employment rate would fall by about 1 percentage points, between 2001 and 2010, and 1.4 percentage points, between 2001 and 2020. Looking at the longer horizon, Germany, France, Italy, Luxembourg, Austria, Netherlands and Portugal are the Member States where the fall in the employment rate would be higher as a result of the changing composition of the labour supply.

As indicated several times in the text, these calculations are made under the simplistic assumption that age-specific employment and unemployment rates do not change. But they will change as a result of business cycle fluctuations, in the short run, as a result of and changing labour market institutions, in the medium run.⁶ Moreover, demographics may also have an impact on the employment prospects of different population groups. We now turn to discuss the theoretical arguments underpinning such a relationship.

⁶ For some empirical estimates of the effects of labour market institutions on the relative employment and unemployment rates of different population groups, see Jimeno and Rodriguez-Palenzuela (2002) and Bertola, Blau and Kahn (2002b).

3. Composition of labour supply and unemployment : A look at the theory

3.1. A traditional supply-demand model

The first reason why unemployment and employment rates may vary across workers of different ages is imperfect substitution. There are jobs which can only be filled by workers of some specific ages. For instance, soccer coaches are relatively old, while young workers are typically found in clerk positions in the service sector. And this imperfect substitution holds even if workers of different ages have the same educational attainments. In a flexible labour market in which relative wages adjust to relative supply and demand of each specific type of labour, a reduction in the size of younger cohorts yields a rise of the youth relative wage. In a rigid labour market in which relative wages do not fully adjust, at least in the short run, a reduction in the size of younger cohorts yields a reduction of the youth relative unemployment rate.

More formally, let me assume that there are two types of workers (1: young, 2: adult) and the production function is given by

\[Y = \left(N _ {1} ^ {\rho} + \delta N _ {2} ^ {\rho}\right) ^ {\frac {1}{\rho}}, \quad 0 \leq \rho \leq 1\]

where Y is output, is employment of young workers, is employment of adult workers, and δ is an indicator of the relative efficiency of the two types of workers. For instance, if education is improving across cohorts, then . The elasticity of substitution between both types of labour is given by Being and the population size of each group, cost minimisation implies

\[u _ {1} - u _ {2} = \sigma \ln \delta + \sigma (\ln w _ {1} - \ln w _ {2}) + \ln L _ {1} - \ln L _ {2}\tag{1}\]

where and are the unemployment rates of young and adult workers, respectively.

There are two ways of reading this equation. First, suppose that relative wages are flexible while unemployment rates are given exogeneously by structural factors (frictions in the labour market, etc.) Then the previous equation establishes a negative relationship between the relative size of the youth population and the relative youth wage. This is, for instance, the intuition behind the analysis in Card and Lemieux (2001) which investigates to what extent the rising education premium is the US, UK and Canada can be explained by the shortfall of highly educated young workers caused by smaller youth cohorts that stopped to increase their education levels relative to their elders, as happened in these countries during the 1980s and 1990s.

There is a second, «more European», way of reading equation (1). Suppose that wages are determined by some wage setting procedure in which workers of different ages have different reservation wages and different bargaining power. The results of wage setting imply some relative wage which plugged into equation (1) gives the unemployment rate differential for workers of different ages. This is, for instance, the intuition behind the analysis in Jimeno and Rodriguez-Palenzuela (2002). Under this view, any labour market institutions which compress the wage structure will produce a higher unemployment differential, while tenure-related institutions favouring adult workers’ bargaining power will reduce this differential. Apart from this, labour market institutions may have other differential effects on labour demand and labour supply across population groups, and, therefore, affect this differential through other channels.

A similar intuition has been applied to the analysis of immigration. Here the most usual model assumes three « types » of labour (native-born skill workers, native-born unskilled workers, and immigrants). The assumption on the production function is that, while immigrants and native-born unskilled workers are perfectly substitutes, immigrants increase the productivity of native-born skill workers. Under this assumption, it can be theoretically shown that the effects of immigration on the wages and employment rates of native-born unskilled workers are ambiguous depending on the characteristics of the wage setting process.⁷ On the empirical front, Angrist and Kugler (2003) show evidence supporting the view that reduced flexibility, especially restrictions on product market flexibility, increases the negative impact of immigration on the labour market prospects of native-born workers.

3.2. The flow approach

The supply and demand framework above does not capture other dimensions along which young and older workers differ. It has been stressed that the main difference between young and adult workers in the labour market is, obviously, work experience. And, since young workers are relatively inexperienced, they have trouble keeping their jobs. There may also be more mobile, a characteristic that they seem to share with immigrants. Thus, in sclerotic labour markets, young and immigrants « grease the wheels of the labour market » by making it attractive to firm to open new vacancies.

⁷ On this, see Dolado, Jimeno and Duce (1997).

⁸ See, for instance, Shimer (1998).

In the standard model of equilibrium unemployment, the more mobile groups, which have higher separation rates, will have higher unemployment rates, as their job matches are destroyed more often. A change in the relative size of a population group has an impact on job creation and job destruction, so that equilibrium unemployment varies. But there is another feature of young workers and immigrants which may be relevant. They are often « mismatched » and continue searching for a job while employed. On-the-job search may create a positive externality by which the unemployment rates of both young and adult workers decreases as the share of the mobile, on-the-job searchers young/immigrants in the labour force increases. The reason is that firms find it profitable to open more vacancies as the share of mobile workers in the population increases. This is the type of model that Shimer (2001) uses to rationalise his findings on the positive relationship between the share of youth population and employment and participation rates across US states. In the remaining of this section I present some variations of a matching model, which closely follows Pissarides (2001), with some of the features highlighted above as relevant to understand the impact of populations ageing on equilibrium unemployment. The overall conclusion is that a rise in the share of young workers in the labour force is likely to increase both equilibrium unemployment and the unemployment rate of older workers, while it is likely to reduce the unemployment rate of young workers. The main assumption behind this result is the higher separation rates for young workers. Allowing for on-the-job search and for productivity upgrading of young workers with employment experience do not change this conclusion.

A. A simple model

To formalise this discussion about the relationship between the age composition of the labour force and equilibrium unemployment, let me assume that labour supply is composed of two groups. Group 1 (« young/immigrants ») are workers with less work experience and subject to higher job destruction. Their jobs are destroyed at rate (exogeneously given). Group 2 (« adults ») are workers with a lower job destruction rate, . The population is normalised to 1, and the relative weights of group are respectively and . Thus, I think of population ageing as a fall in while immigration flows increase

⁹ The less technically oriented reader may want to skip the following sections and go directly to section 4.

¹⁰ This differs from Shimer (2001), where population ageing is modelled as a fall in population growth. Moreover, in Shimer’s (2001) model wages are exogenous and, hence, he derives the consequences of a fall in population growth for short-term unemployment, not for equilibrium unemployment.

Workers and firms discount the future at a rate r and meet according to a standard matching function

\[m (v, u _ {1} + u _ {2}) = \bar {m} (\theta), \quad m _ {1}, m _ {2} > 0\]

where v is the vacancy rate, measured relative to the labour force, is the number of unemployed workers of group I , and is a measure of market tightness. Thus, the probability of a firm meeting a worker is , while the probability of a worker meeting a firm is , which is increasing in θ. For the simulation results I will present below, it is assumed that

\[\boxed {m} (\theta) = \theta^ {1 / 2}\]

so that

Wages )are determined by symmetric Nash bargaining so that the surplus of the match is equally divided between the firm and the worker. Each period the productivity of the match is given by , with plausibly The cost of keeping a vacancy open is . Under these assumptions, the Bellman equations given the value of an unfilled vacancy and of a filled vacancy by a worker of type are, respectively

\[\begin{array}{c} r V = - c + \alpha q (\theta) (J _ {1} - V) + (1 - \alpha) q (\theta) (J _ {2} - V) \\ r J _ {I} = x _ {I} - w _ {I} + s _ {I} (V - J _ {I}) \qquad I = 1, 2 \end{array}\]

being the proportion of workers of type 1 in the population of job searchers. for the value of a worker of type I of being employed and unemployed , assuming that the flow utility of being unemployed is normalized to 0, the corresponding Bellman equations are

\[r W _ {I} = w _ {I} + s \left(U _ {I} - W _ {I}\right) \quad r U _ {I} = \theta q (\theta) \frac {u _ {I}}{u _ {1} + u _ {2}} \quad I = 1, 2\]

Finally, in equilibrium, free entry drives the value of an unfilled vacancy to zero , and in steady state the flows in and out of unemployment are equal for workers of each type, so that

\[s _ {1} (\mu - u _ {1}) = \theta q (\theta) \alpha u _ {1} \quad s _ {2} (1 - \mu - u _ {2}) = \theta q (\theta) (1 - \alpha) u _ {2}\]

After obtaining wages from the bargaining condition and the free-entry condition , and substituting into the equation given the value of an unfilled vacancy, it can be obtained that

\[\frac {c}{q (\theta)} = \frac {\alpha x _ {1}}{2 (r + s _ {1}) + \alpha \theta q (\theta)} + \frac {(1 - \alpha) x _ {2}}{2 (r + s _ {2}) + \theta q (\theta) (1 - \alpha)}\]

while the two flow equations can be used to obtain

\[\frac {1 - \alpha}{\alpha} = \frac {s _ {2} (1 - \mu) [ s _ {1} + \alpha \theta q (\theta) ]}{s _ {1} \mu [ s _ {2} + (1 - \alpha) \theta q (\theta) ]}\]

Given some functional form for the matching function and parameter values for r and m, these two equations can be solved for a and q. For instance, I assume and . For these parameter values, the resulting unemployment rates in this economy for different values of the weight of Group 1 workers are presented in Table 2a. As the proportion of Group 1 workers increases, the unemployment rate of group 1 workers falls while the unemployment rate of group II workers rises, contrary to the intuition obtained in the supply-demand analysis in the previous section. Even allowing for imperfect substitution between both groups by assuming that firms’ profits differ by workers’ types, for instance =2 and with a higher flow cost of keeping a vacancy unfilled (c=3), the same conclusion applies.

B. Introducing on-the-job search

As an extension of the simple model above I now introduce on-the-job search by Group I workers. In this case, the ratio of vacancies to job searchers is , being the number of employed workers of group I who are engaged in on-the-job search. Now the Bellman equations giving the values of an unfilled vacancy (V) and of filled vacancies by workers of both types are:

\[\begin{array}{c} r V = - c + \alpha q (\theta) (J _ {1} - V) + (1 - \alpha) q (\theta) (J _ {2} - V) \\ r J _ {1} = x _ {1} - w _ {1} + [ s _ {1} + \beta \theta q (\theta) ] (V - J _ {1}) \\ r J _ {2} = x _ {2} - w _ {2} + s _ {2} (V - J _ {2}) \end{array}\]

where and . Here on the-job-search only creates more frictions in the labour market without generating any improving in the productivity of the job matches. In this sense, on-job search is “non-productive”. This assumption will be relaxed below.

The probability of an employment worker of type I receiving a job offer and leaving to another job is then . In this case, the flow equations are:

\[s _ {1} (\mu - u _ {1}) = \theta q (\theta) (\bar {\alpha} - \beta) u _ {1} \quad s _ {2} (1 - \mu - u _ {2}) = \theta q (\theta) (1 - \bar {\alpha}) u _ {2}\]

After some algebra, these equations can be reduced to

\[\begin{array}{r l} \frac {c}{q (\theta)} = & \frac {\bar {\alpha} x _ {1}}{2 (r + s _ {1}) + \beta \theta q (\theta)} + \frac {(1 - \bar {\alpha}) x _ {2}}{2 (r + s _ {2}) + \theta q (\theta) (1 - \bar {\alpha})} \\ & s _ {1} \beta = \theta q (\theta) (\bar {\alpha} - \beta) ^ {2} \\ & s _ {2} (\bar {\alpha} - \mu) = \mu \theta q (\theta) (1 - \bar {\alpha}) ^ {2} \end{array}\]

Table 2b presents the simulation results for this case, under similar parameter values as those used previously. Non-productive on-the-job search creates more frictions in the labour market, and, hence, labour market tightness is (slightly) lower and equilibrium unemployment is higher than in the simpler model without on-the-job search. But as in the previous case, the unemployment rate of workers of Group 1 decreases as its proportion in the labour force rises, while the unemployment rate of workers of Group 2 increases.

C. Productivity upgrading

I now consider the case in which there is on-the-job-search implies some productivity gains. The way I model this is by assuming that Group 1 workers after one period of employment have the same productivity as Group 2 workers (x₂). I keep assuming that the separation rate of this worker remains high (s₁) even after their productivity has increased, but that they quit search. In this case, the values of an unfilled vacancy (V) and of filled vacancies are:

\[\begin{array}{c} r V = - c + (\bar {\alpha} - \beta) q (\theta) (J _ {1} - V) + \beta q (\theta) (J _ {1} ^ {2} - V) + (1 - \bar {\alpha}) q (\theta) (J _ {2} - V) \\ r J _ {1} = x _ {1} - w _ {1} + [ s _ {1} + \beta \theta q (\theta) ] (V - J _ {1}) + [ 1 - s _ {1} - \beta \theta q (\theta) ] (J _ {1} ^ {2} - J _ {1}) \\ r J _ {1} ^ {2} = x _ {2} - w _ {1} ^ {2} + s _ {1} (V - J _ {1} ^ {2}) \\ r J _ {2} = x _ {2} - w _ {2} + s _ {2} (V - J _ {2}) \end{array}\]

where is the value of a job of high productivity (x₂) filled by a worker of Group 1 with a previous employment spell.

The determination of wages by symmetric Nash bargaining in this case is very cumbersome and algebra gets too complicated. Thus, I take a shortcut. I assume that the wage of Group I workers in low productivity jobs is as in the case of non-productive search. As for the wage of Group I workers in high productivity jobs, I assume that they get their previous wage (w₁) and then they share the increase in productivity with their employer . This assumption amounts to making young workers more attractive to hire as in their first job they do not get their share of the expected future productivity increase produced by job experience.

Under these assumptions, the Bellman equations and the flow equilibrium conditions can be written as follows

\[\begin{array}{l} \frac {c}{q (\theta)} = \frac {(\bar {\alpha} - \beta) [ 1 + r - \beta \theta q (\theta) ] [ r + s _ {1} + \beta \theta q (\theta) ] x _ {1}}{(1 + r) (r + s _ {1}) [ 2 (r + s _ {1}) + \bar {\alpha} \theta q (\theta) ]} + \frac {(\bar {\alpha} - \beta) [ 1 - s _ {1} - \beta \theta q (\theta) ] (x _ {2} - x _ {1})}{2 (1 + r) (r + s _ {1})} + \\ + \frac {\beta (x _ {2} - x _ {1})}{2 (r + s _ {2}) + \theta q (\theta) (1 - \bar {\alpha})} + \frac {\beta [ r + s _ {1} + \beta \theta q (\theta) ] (x _ {2} - x _ {1})}{2 (r + s _ {1}) + \bar {\alpha} \theta q (\theta)} + \frac {(1 - \bar {\alpha}) x _ {2}}{2 (r + s _ {2}) + \theta q (\theta) (1 - \bar {\alpha})} \end{array}\]

\[\begin{array}{c} s _ {1} \beta = \theta q (\theta) (\bar {\alpha} - \beta) ^ {2} \\ s _ {2} (\bar {\alpha} - \mu) = \mu \theta q (\theta) (1 - \bar {\alpha}) ^ {2} \end{array}\]

Table 2c presents the simulation results for this case. As the share of workers of Group 1 rises, labour market tightness, the equilibrium unemployment rate and the unemployment rate of workers of Group 2 increase, while the unemployment rate of workers of Group 1 falls.

4. Some empirical evidence

The theoretical models sketched above are not totally conclusive about the effects of changing composition of labour supply on aggregate unemployment and employment rates. These effects will depend on the degree of substitution across workers, labour market institutions, and, in equilibrium models, the assumptions about the difference behaviour of workers of different types regarding search and about the relationship between the cost of creating job vacancies and the size and composition of labour supply.

In this section I turn to the empirics. By using OECD cross-country data I try to assess to what extent the relationship between demographics and age and gender specific unemployment and employment rates are affected by labour market institutions. The data on age-specific unemployment rates are from OECD. The data on the youth population shares are from EUROSTAT. And the data on labour market institutions are from Blanchard and Wolfers (2000) and the LSE/CEP OECD data set (see the data appendix for more details). The sample is composed of 19 countries with annual observations through 1968- 1996.¹² To this sample I also add data on the proportion of foreign (extra EU) workers in the labour force for EU15 countries. 13

I estimate simple OLS regressions of the form:

¹¹ See the discussion in Shimer (2001), section VI.

¹² This panel is unbalanced as there are missing observations for some countries in several years. See the data appendix for the observations available for each country.

¹³ For more information on the composition of the sample and the definition of variables, see the data appendix.

\[y _ {i j t} = \mu_ {i} + \delta_ {t} + \beta_ {j} + \alpha_ {i} s _ {j t} + \gamma_ {i} l m _ {j t} + \lambda_ {i} (s _ {j t} * l m _ {j t}) + \varepsilon_ {i j t}\tag{2}\]

for demographic group i country j and time t. The model includes country and year effects, and and separate regressions are estimated for each population group. The dependent variables are unemployment and employment rates for males and females aged 15-24 and 25-54. The regressors and are, respectively, the (logarithm of the) share of the youth population (15-24) to the population aged 25-54, and vectors of the usual labour market institutions (timeinvariant and time-varying), such as employment protection legislation, characteristics of the wage setting process, and indicators of the generosity of unemployment benefits, among others. Following Angrist and Kugler (2003) I also consider an alternative model in which the unemployment rate of each demographic group also depends on the proportion of immigrants on the labour force (data from EUROSTAT) and its interaction with labour market institutions, so that the estimated model becomes:

\[y _ {i j t} = \mu_ {i} + \delta_ {t} + \beta_ {j} + \alpha_ {i} s _ {j t} + \gamma_ {1 i} l m _ {j t} + \lambda_ {1 i} (s _ {j t} * l m _ {j t}) + \gamma_ {2 i} m _ {j t} + \lambda_ {2 i} (m _ {j t} * l m _ {j t}) + \varepsilon_ {i j t} (2 ^ {\prime})\]

being the proportion of immigrants in the labour force in country j at time t. There are several cautions at interpreting the results form this regression. First, the sample size is too small due to data limitations on the stocks of foreign workers in EU countries. Secondly, we are using stock rather than flows, assuming that all foreign workers have the same impact on the labour market regardless of the date of their entry. As already mentioned, there are good reasons to believe that this may not be the case (see Card, 2001). Finally, immigration may be endogenous so that foreign workers arrive to the labour markets with higher employment rates and lower unemployment. Nevertheless, I think that the results from such a regression are illustrative and I will take into account these possibilities when discussing the results below.

The results are presented in Table 3, for age and gender specific employment rates, and Table 4, for age and gender specific unemployment rates. As for the employment rates of males aged 25-54, higher generosity of benefits, more restrictive employment protection legislation, a higher tax wedge, and a lower degree of co-ordination in wage setting seem to produce lower employment rates for this group when the relative weight of the young population is higher. This also applies to the immigration variable, although in this case the estimated coefficients have less statistical significance. Similar results are obtained for the employment rates of women aged 25-54, although in this case the interaction of the tax wedge and the demographic variable is not statistically significant and it is even positive when using time-varying measures of labour market institutions. As for young male workers, employment rates are lower when its relative weight in the labour force is higher and in “more rigid” labour markets (more generous unemployment benefits, more restrictive employment protection legislation, higher tax wedge and less co-ordination at wage setting). However, not a consistent pattern was found regarding the effects of the interaction of the demographic variables with labour market institutions. Similar comments apply to the results regarding young female workers.

As for unemployment rates of male workers aged 25-54, more generous unemployment benefits and a lower degree of co-ordination at wage setting seem to imply higher unemployment for this group when the weight of young workers in the labour force is higher. As for women in the same age group, employment protection legislation also produces higher unemployment when the weight of young workers in the labour force is higher, in addition to the other institutions found relevant for men. Regarding young workers, both for males and females, stricter employment protection legislation, less co-ordination at wage setting, and a lower tax wedge are conducive to higher unemployment when the relative weight of young worker in the labour force and the proportion of foreign workers are higher.

5. Concluding remarks

This paper suggests that projected demographic changes in the EU may have significant impact on employment and unemployment rates of several population groups, and that this impact may depend on the labour market institutions regulating the functioning of the labour market. First, the age structure of labour supply is changing very quickly, with diminishing the weight of young workers and increasing the weight of older workers. The net effect of this compositional change, under the maintained assumption that age-specific employment/unemployment rates do not vary, is that the EU15 aggregate employment rate will fall in the next two decades by about 1.5 percentage points, while the EU15 aggregate unemployment rate do not change, despite the reduction of the weight of younger workers with relatively high unemployment rate in the labour force.

Secondly, there are good theoretical reasons to believe that the changing composition of the labour supply will have differential effect on the employment/unemployment rates of specific population groups. This effect depend on the degree of substitution between workers of different characteristics, their relative levels of education, labour market institutions determining the wage structure, and frictions in the labour market with regard to the job destruction rates and job search activity of workers of different ages. In principle, as the weight of younger workers in the labour force decreases, their relative unemployment rate and the aggregate equilibrium unemployment rate also decrease.

Finally, the experience of the last three decades, with the arrival of the baby boomers to the labour market and the subsequent fall in the relative weight of young workers in the labour force, suggests that the impact of demographic change in the employment prospects of different population groups depends very much upon the labour market institutions determining unemployment benefits, employment protection legislation and wage setting. Th evidence presented in this paper is still preliminary and further work along this line is in order.

To conclude it should be noted that there are additional likely labour market effects of demographic change, particularly on productivity and, hence, on wages. The models sketched in Section 3 of this paper also have implications regarding wage inequality among population groups. Unfortunately, in the EU there are less good empirical evidence on these effects than in the US, mainly because the lack of availability of time-series data on wages of a cohort of individuals with controlled characteristics. But as time goes by, this should become an important issue in the European research agenda about the labour market consequences of demographic change.

Data Appendix

The following table gives the countries and the period included in the sample

When the variable regarding the proportion of Foreign labour force (extra EU15) over Total Labour Force is used, the sample is restricted to the EU15 countries appearing in the above table for the following periods:

Table A2. Sample composition (restricted)

References

1. The definitions and sources of the variables used in the regression analysis are the following:

References

1. Unemployment and employment rates. Source: OECD, Labour Force Statistics and Employment Outlook (several years).

References

1. Pop: Ln(population aged 15-24/population aged 25-54). Source: EUROSTAT. Database: New Chronos.

References

1. Immig: % Foreign labour force (extra EU15) over Total Labour Force. Source: EUROSTAT. Database: New Chronos.

References

1. Rrate: Replacement rate of unemployment benefits. Source: Blanchard and Wolfers (2000), (time-invariant), and LSE/CEP OECD Database (time-14 varying).

References

1. Benefit: Index of duration of unemployment benefits. Source: Blanchard and Wolfers (2000), (time-invariant), and LSE/CEP OECD Database (timevarying).

References

1. Empro: Degree of strictness of employment protection legislation. Source: Blanchard and Wolfers (2000), and (time-invariant) LSE/CEP OECD Database (time-varying).

References

1. Tw: Tax wedge. Source: Blanchard and Wolfers (2000), (time-invariant), and LSE/CEP OECD Database (time-varying).

References

1. Coord: Degree of co-ordination in wage setting. Source: Blanchard and Wolfers (2000), (time-invariant), and LSE/CEP OECD Database(time-varying).

¹⁴ See the data appendix in Nickell, Nunziata and Ochel (2002) for more details.

References

1. Angrist, J. and A. Kugler (2003): “Productive or Counter-Productive : Labour Market Institutions and the Effect of Immigration on EU natives”, Economic Journal (forthcoming).
2. Bauer, T. and K.F. Zimermann (1999): “Assessment of Possible Migration Pressure and its Labour Market Impact Following EU Enlargement to Central and Eastern Europe”, IZA, Research Report, 3.
3. Bertola, G., F. Blau and L. Kahn (2002a): “Comparative Analysis of Labour Market Outcomes: Lessons for the US from the International Long-Run Evidence” in The Roaring Nineties: Can Full Employment Be Sustained?, edited by A. Krueger and R. Solow. New York: Russell Sage.
4. Bertola, G., F. Blau and L. Kahn (2002b): “Evolving Institutions and Demographic Employment Patterns”, mimeo.
5. Blanchard, O. and J. Wolfers (2000): “The Role of Shocks and Institutions in the Rise of European Unemployment” Economic Journal, 110 (March), C1- C33.
6. Borjas, G. (1994): “The Economics of Immigration”, Journal of Economic Literature, 32, 1667-1717.
7. Borjas, G., R. Freeman, and L. Katz (1996): “Searching for the Effects of Immigration on the Labour Market”, American Economic Review, LXXXVI, 246-251.
8. Card, D. (1990): “The Impact of the Mariel Boatlift on the Miami Labour Market”, Industrial and Labour Relations Review, XLIII, 245-247.
9. Card, D. (2001): “Immigrants Inflows, Native Outflows, and the Local Labour Market Impacts of Higher Immigration”, Journal of Labour Economics, 19, 22-64.
10. Card, D. and T. Lemioux (2001): “Can Falling Supply Explain the Rising Return to College for Younger Men? A Cohort Based Analysis” The Quarterly Journal of Economics, vol. 116, 705-746.
11. Dolado, J.J., J.F. Jimeno, and R. Duce (1997): “Los efectos de la inmigración sobre la demanda de trabajo cualificado y no cualificado: Evidencia para España”, Cuadernos Económicos de ICE, 63.
12. Easterlin, R. (1968), Population, Labour Force, and Long Swings in Economic Growth: The American Experience, National Bureau of Economic Research. New York: Columbia University Press.
13. EUROSTAT (2001), Statistiques Sociales Européennes. Démographie. European Commission and EUROSTAT: Brussels.
14. Friedberg, R. and J. Hunt (1995): “The Impact of Immigrants on Host Country Wages, Employment, and Growth”, Journal of Economic Perspectives, 9, 23-44.
15. Jimeno, J.F. and Rodriguez-Palenzuela (2002): “Youth Unemployment in the OECD: Demographic Shifts, Labour Market Institutions, and Macroeconomic Shocks”, FEDEA, working paper 2002-15.
16. Johnson, W.R. (1980): “Vintage Effects in the Earnings of White American Men”, The Review of Economic and Statistics, vol 62, 3, 399-407.
17. Katz, L. and K. Murphy (1992): “Changes in Relative Wages, 1963-1987: Supply and Demand Factors”, The Quarterly Journal of Economics, vol. 107, 35-78.
18. Korenman, S. and D. Neumark (2000): “Cohort Crowding and Youth Labour Markets: A Cross-National Analysis”, in Youth Unemployment and Joblessness in Advanced Countries, edited by D.G. Blanchflower and R.B. Freeman. Chicago: Chicago University Press.
19. Nickell, S., L. Nunziata and W. Ochel (2002): “Unemployment in the OECD since the 1960s: What do we know?”, Bank of England, mimeo.
20. Pissarides, C. (2001), Equilibrium unemployment theory, MIT Press: Cambridge Ma.
21. Shimer, R. (1998): “Why Is the US Unemployment Rate So Much Lower?”, NBER Macroeconomics Annual.
22. Shimer R. (2001): “The Impact of Young Workers on the Aggregate Labour Market”, The Quarterly Journal of Economics, vol. 116, 969-1007.
23. Sneddon Little, J. and R.K. Triest (2002): “The Impact of Demographic Changes on US Labour Markets”, New England Economic Review, First Quarter, 47-68.
24. Welch, F. (1979): “Effects of Cohort Size on Earnings: The Baby Boom Babies Financial Bust”, Journal of Political Economy, 87, 565-597.

Table 1. Forecasted unemployment rates and employment rates in EU Member States in 2010 and 2020 under the assumption that population group specific rates remain constant at the 2001’s levels¹⁵ Unemployment rates

Employment rates

¹⁵ The number of population groups considered is 18, defined over three dimesions: gender, age (15- 24, 25-54, 54-65) and education attaintments (low, medium and high). The educational attaintments are coded according to ISCED97 classification : ISCED0_2 (Pre-primary) : low, ISCED3-4 (Upper secondary and post-secondary, non-tertiary education): medium, ISCED5_6 (Tertiary education) : high.

Table 2a. Simulation results Simplest model

Table 2b. Simulation results Model with non-productive on-the-job search

Table 2c. Simulation results Model with on-the-job search and productivity upgrading

Table 3.a. Regression results Employment rate of males aged 25-54

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Table 3.b. Regression results Employment rate of females aged 25-54

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Table 3.c. Regression results Employment rate of males aged 15-24

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Dependent variable:

Table 3.d. Regression results Employment rate of females aged 15-24

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Dependent variable:

Table 4.a. Regression results Unemployment rate of males aged 25-54

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Table 4.b. Regression results

Dependent variable:

Unemployment rate of females aged 25-54

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Dependent variable:

Table 4.c. Regression results Unemployment rate of males aged 15-24

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Table 4.d. Regression results Dependent variable: unemployment rate of females aged 15-24

Notes: Country and time fixed-effects included. t-statistics in parenthesis.

Figure 1. Share of immigration in EU population growth

2001 1991-2000 1981-1990 1971-1980 1961-1970

2000 2010 2020

Figure 2. Weights of the population of different age-groups in the working-age population (15-64) Population aged 15-24 Population aged 25-54

Population aged 55-64

Figure 3.a. Unemployment rates by age-specific population groups, 2001

Figure 3.b. Employment rates by age-specific population groups, 2001

RELACIÓN DE DOCUMENTOS DE FEDEA

DOCUMENTOS DE TRABAJO

References

1. 2004-18: “Demographic change, immigration, and the labour market: A European perspective”, Juan F. Jimeno.

References

1. 2004-17: “The Effect of Immigration on the Employment Opportunities of Native-Born Workers: Some Evidence for Spain”, Raquel Carrasco, Juan F. Jimeno y Ana Carolina Ortega.

References

1. 2004-16: “Job Satisfaction in Europe”, Namkee Ahn y Juan Ramón García.

References

1. 2004-15: “Non-Catastrophic Endogenous Growth and the Environmental Kuznets Curve”, J. Aznar-Márquez y J. R. Ruiz-Tamarit.

References

1. 2204-14: “Proyecciones del sistema educativo español ante el boom inmigratorio”, Javier Alonso y Simón Sosvilla-Rivero.

References

1. 2004-13: “Millian Efficiency with Endogenous Fertility”, J. Ignacio Conde-Ruiz. Eduardo L. Giménez y Mikel Pérez-Nievas.

References

1. 2004-12: “Inflation in open economies with complete markets”, Marco Celentani, J. Ignacio Conde Ruiz y Klaus Desmet.

References

1. 2004-11: “Well-being Consequences of Unemployment in Europe”, Namkee Ahn, Juan Ramón García López y Juan F. Jimeno.

References

1. 2004-10: “Regímenes cambiarios de facto y de iure. Una aplicación al tipo de cambio yen/dólar”, Francisco Ledesma-Rodríguez, Manuel Navarro-Ibáñez, Jorge Pérez-Rodríguez y Simón Sosvilla-Rivero.

References

1. 2004-09: “Could this ever happen in Spain? Economic and policy aspects of a SARS-like episode”, José A. Herce.

References

1. 2004-08: “Capital humano en España: Una estimación del nivel de estudios alcanzado”, Javier Alonso y Simón Sosvilla-Rivero.

References

1. 2004-07: “Modelling vintage structures with DDEs: Principles and applications”, Raouf Boucekkine, David de la Croix y Omar Licandro.

References

1. 2004-06: “Substitutability and Competition in the Dixit-Stiglitz Model”, Winfried Koeniger y Omar Licandro.

References

1. 2004-05: “The short-run dynamics of optimal growth models with delays”, Fabrice Collard, Omar Licandro y Luis A. Puch.

References

1. 2004-04: “Currency Crises and Political Factors: Drawing Lessons from the EMS Experience”, Francisco Pérez-Bermejo y Simón Sosvilla-Rivero

References

1. 2004-03: “El futuro de las pensiones en España: Perspectivas y lecciones”, J. Ignacio Conde-Ruiz y Javier Alonso.

References

1. 2004-02: “Do temporary contracts increase work accidents? A microeconometric comparison between Italy and Spain”, Virginia Hernanz y Luis Toharia.

References

1. 2004-01: “Job Match Quality throughout the Business Cycle in the Spanish Labour Market”, Cristina Fernández.

References

1. 2003-30: “Innovation, Investment and Productivity: Evidence from Spanish Firms”, Omar Licandro, Reyes Maroto y Luis A. Puch.

References

1. 2003-29: “ Efectos de las ayudas europeas sobre la economía madrileña, 1990-2006: Un análisis basado en el modelo Hermin”, Simón Sosvilla-Rivero y José A. Herce.

References

1. 2003-28: “Canarias y los Fondos Estructurales europeos”, Simón Sosvilla-Rivero.

TEXTOS EXPRESS

References

1. 2004-01: “The Spanish economy through the recent slowdown. Current situation and issues for the immediate future”, José A. Herce y Juan F. Jimeno.

References

1. 2003-01: “12+1 Reflexiones sobre 12+1 años de Gasto Farmacéutico”, José-Luis Perona Larraz.