Fundación de Estudios de Economía Aplicada

Business Cycle Effects on Labour Force Transitions for Older People in Spain* by Sergi Jiménez-Martín** Judit Vall Castello*** Documento de Trabajo 2009-25

Economía de la Salud y Hábitos de Vida FEDEA – “la Caixa”

July 2009

* We thanks Nacho Garcia-Perez and Yolanda Rebollo for helpful comments. Financial help from project ECO2008-06395-C05-01 is gratefully acknowledged.

** Department of Economics, Universitat Pompeu Fabra, and FEDEA. Email: sergi.jimenez@upf.edu

*** Marie Curie PhD Research Fellow, Maastricht Graduate School of Governance, University of Maastricht, The Netherlands. Email: judit.vallcastello@maastrichtuniversity.nl

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. These Working Paper are distributed free of charge to University Department and other Research Centres. They are also available through Internet: http://www.fedea.es. ISSN:1696-750X

Jorge Juan, 46 28001 Madrid -España Tel.: +34 914 359 020 Fax: +34 915 779 575 infpub@fedea.es

Abstract

This paper analyses the determinants of observed exits from employment for people aged 45-59 years old in the context of the Spanish labour market in 1981-2006. The main aim of the paper is to identify the effect of the business cycle (BC) on the timing and the type of exit route out of the labour force.

We proceed in two stages. In the first stage, we study the determinants of exits from employment to non-employment. In the second, we take into account the fact that there are several competing exit routes (unemployment, disability or inactivity) and estimate a competing risk model to evaluate how important BC conditions are in determining the respective exit probabilities. We make use of the recently released Muestra Continua de Vidas Laborales to estimate discrete time hazard regression models. We match this information with a number of variables constructed with macroeconomic data derived from the Instituto Nacional de Estadistica to measure growth and employability performance of different economic sectors and regions in Spain in order to capture the variation in the business cycle between times, sectors and regions. Time-varying covariates are also included in the analysis to model the monetary incentives provided by the system. We find that both BC conditions and a number of special schemes included in the unemployment and disability legislation affect the exit timing and also the choice of the route out of the labour market.

Keywords: duration models, exits from employment, business cycle conditions. JEL classification: J21, J22, R23

Resumen

Este artículo analiza los determinantes de las salidas del empleo de individuos con edades comprendidas entre 45 y 59 años en el contexto del Mercado laboral Español de 1981 a 2006. El principal objetivo del presente trabajo es la identificación de los efectos del ciclo económico (CE) en el momento y el tipo de salida del mercado laboral. En la primera parte del artículo estudiamos los determinantes de las salidas del empleo hacía una única situación de no-empleo, mientras que en la segunda parte incorporamos tres salidas potenciales del mercado laboral (desempleo, incapacidad y inactividad) y estimamos un modelo de duración de riesgos en competencia para evaluar la importancia de las condiciones del ciclo económico en la determinación de las respectivas probabilidades de salida del empleo. En nuestro trabajo utilizamos la Muestra Continua de Vidas Laborales para la estimación de dos modelos de duración en tiempo discreto. A esta información microeconómica de los individuos le añadimos variables macroeconómicas que hemos construido con datos del Instituto Nacional de Estadística para capturar la evolución del crecimiento y la empleabilidad de diferentes sectores económicos y regiones de la economía española. De esta manera, conseguimos incorporar en nuestros modelos las variaciones temporales, sectoriales y regionales del ciclo económico en España. Nuestros resultados sugieren que tanto las condiciones del ciclo económico como parte de los incentivos incluidos en ciertas áreas de la legislación española de desempleo e incapacidad tienen un efecto significativo en el momento y en la ruta de salida del mercado laboral.

Palabras clave: modelos de duración, salidas del empleo, efectos del ciclo económico. Clasificación JEL: J21, J22, R23

1. Introduction

Labour market policies and legislation are evolving over time in response to changing economic and demographic conditions around the world and, for a long time, European governments focused on promoting early retirement for older workers, particularly during the 1970s and 1980s, in order to generate job opportunities and reduce unemployment for younger workers.

However, because of the huge number of older workers taking advantage of the early retirement option and because of the consequent substantial decrease in both the activity and labour force participation rates of the older population, this policy soon proved to have a number of problematic and unforeseen effects. At the same time, society was ageing due to increased life expectancy and reduced fertility rates, and this only exacerbated the problem (Fischer et al. 2006).

As a result of these two parallel developments, there was an increase in the financial pressure of the social security system and on taxes of the working population in relation to the non-working one.

As the ageing of society is being regarded as one of the most important social and economic challenges for Europe in recent times, these developments have prompted a shift in policy focus towards ways of delaying retirement and of keeping older workers economically active in the labour market. To achieve this goal, governments introduced several policies such as incentives for the firms that hire older workers or limitations in the accession rules to these early retirement schemes. However, these efforts have not been translated into increases in activity rates of older workers because many of these workers are able to find alternative programs to leave the labour market, such as the disability or unemployment schemes, before reaching the retirement or early retirement age. Therefore, these other possibilities of leaving employment should be studied and understood so that they can be integrated in the analysis when designing policies to keep older workers active in the labour market.

The literature analyzing labour market transitions usually identifies a collection of factors as potential determinants of these transitions, ranging from individual-specific determinants, employer, firm and industry characteristics or business cycle conditions to incentives embedded in the Social Security institutions (See for example Garcia-Perez et al.2007, 2008, Blanco 2000). It is important to understand the incentives to withdraw from the labour force provided by each of these factors in order to devise effective labour market policy reforms.

Therefore, our aim in this paper is to identify the different incidence of each of these factors on labour force exit behaviour among Spanish workers aged 45-59 distinguishing between exits to unemployment, disability benefits or inactivity. We will pay special attention to the role of the business cycle on the inflow into disability benefits, the monetary incentives provided by the system and the relative power of agebased legislation schemes in inducing transitions between these labour force states.

1.1. Literature

The issue of the effect of the business cycle on labour force participation behaviour has not received much attention in the Spanish literature mainly because of the difficulty of combining macroeconomic and microeconomic data in a coherent way. However, workers’ participation decisions during expansionary or recessionary periods are crucial for understanding how labour markets adjust to macroeconomic fluctuations (Darby et al, 1998). At the same time, the economic environment also affects the performance of the firms operating in the labour market which make their decisions on labour demand needs partly based on the economic conditions of the country. Furthermore, the effect of the business cycle on firm performance is usually heterogeneous among economic sectors and regional units of a single country. In this paper, we try to capture these differences with a number of variables that describe labour market conditions and economic and industrial growth at the sector and regional level.

Most of the literature that analyzes the effect of the business cycle on labour market participation rates approaches the topic from a macroeconomic point of view. These studies usually choose the unemployment rate as a proxy for the business cycle and they evaluate its effects on participation rates in a panel data setting using a pool of different countries. Their findings usually suggest that participation rates are lower during recessionary times. However, macroeconomic time-series data are not enough to identify the causal effect behind an individual’s behaviour, as information on workers characteristics and choices is not available. At the same time, the problem with microeconomic surveys is that they are not long enough so as to capture the effect of the business cycle. Therefore, studies using this kind of microeconomic survey data usually include one additional regressor (calendar time) to control for the business cycle¹ (Van den Berg et al, 2000). In this paper, our data (administrative records) allow us to reconstruct the entire working history of the individuals since turning 45, which we combine with a number of business cycle indicators constructed with macroeconomic data. This process results in a better identification of the causal effects of the economic conditions on individual’s behaviour.

The effects of labour market institutions on labour market transitions at advanced ages have been widely studied in the literature (see Gruber and Wise, 1999, for a crosscountry comparison). In the case of Spain, Boldrin et al. (1999, 2004) show the strong incentives to retire early provided by the Spanish Social Security legislation for individuals with below average working histories and/or low wages. This finding explains the strong increase in the number of early retirees in Spain during the 1990s. Two more papers by Kugler et al. (2003) and Garcia-Perez et al. (2007) focus on the impact of a reform in 1997 of the labour market legislation on transitions into a permanent contract. In the spirit of Boldrin et al. (2004), Argimon et al. (2007) find that the Social Security Wealth variable has a positive impact on retirement decisions using a reduced-form duration model and Cairo (2007) employs a competing risk model to study transitions to early and partial retirement for men between 60 and 70 years old in 2005. She includes two variables, Social Security Wealth and Social Security Accrual, to model monetary incentives and finds evidence that Social Security incentives do affect retirement decisions.

The effects of the Spanish unemployment legislation on re-employment probabilities and on the duration of the spell of unemployment have been studied by Bover at al. (2002) and Jenkins et al. (2004). The paper by Bover et al. analyzes the effects of both unemployment benefits and the business cycle on unemployment durations for the period 1987-1994 using the Spanish Labour Force Survey (EPA) for men aged 20 to 64. Their conclusions suggest that the hazard rate of leaving unemployment is higher with favourable business cycle conditions and lower for individuals receiving unemployment benefits (as opposed to unemployed individuals who are not entitled to benefits). The paper by Jenkins et al. (2004) focuses on the effects of the unemployment benefit level on the re-employment probabilities of Spanish men aged 20 to 59 between 1987 and 1991 and finds that benefits do affect unemployment durations but that the level of these benefits has only a small disincentive effect on re-employment probabilities. Business cycle conditions are also found to have an impact on re-employment rates of unemployed workers.

¹ With microeconomic data, it is usually assumed that the duration of employment is independent of business cycle conditions when the number of periods analyzed is relatively small.

On the other hand, Garcia-Perez and Sanchez-Martin (2007) restrict their analysis to older people and their paper explores the monetary incentives embedded in the Social Security institutions to stay unemployed. They use Spanish data to estimate the parameters of a theoretical model of job search behaviour and conclude that both labour market conditions and institutional incentives play a very important role in explaining the search behaviour of unemployed individuals between 55-65 years old in Spain. For people in this age range, remaining unemployed is found to be the best strategy given the level of unemployment benefits and the penalties associated with the early retirement option.

Another strand of the literature has focused on the alternative of leaving employment and claiming a disability pension. The paper by Jimenez-Martin et al. (2006) investigates the effects of permanent disability benefits on labour market behaviour of people aged 45-59 in Spain. The results seem to prove that the disability option has been used as a form of exit from the labour market for individuals substantially close to the early retirement age who do not fulfil the requirements to access this route, as some of these individuals are found to receive disability benefits without deserving them.

Blanco (2000) estimates a competing risk duration model to study the decision of individuals between 50 and 64 years of age to leave the labour force and either take early retirement benefits or claim disability pensions. In her paper, Blanco does not include a variable to measure the monetary incentives of the Social Security System but she incorporates a number of labour demand variables describing the firm and sector in which the individual is employed. Her findings suggest that both the decision to retire and the decision to claim disability benefits are very much affected by the type of job and the level of income of the last job.

The current study contributes to this body of literature analyzing transitions from employment to unemployment, disability pensions and inactivity of Spanish workers aged 45-59 and incorporating as regressors a rich set of variables describing the firm and sector in which the individual is employed, business cycle indicators both at the sector and regional level and age-specific dummies in order to capture the incentives provided by some special schemes included in the Social Security legislation. The aim is to identify the effects of each of these factors on labour market transitions, paying special attention to the effect of business cycle variables on transitions into unemployment and disability pensions. The analysis is conducted separately for men and women due to the different evolution of the main labour market variables, such as participation rates, between the two genders.

Our results show the strategic use of disability pensions by older workers approaching retirement as an alternative means of leaving the labour market when the economic situation of the country deteriorates and economic growth slows down.

The paper proceeds as follows: Section 2 describes and analyzes the evolution of the main labour market and economic indicators in Spain during the last 30 years paying special attention to the group of workers with ages comprised between 45-59 years old. Section 3 describes the Spanish legislation on unemployment and disability benefits focusing on those schemes specifically applicable to older people. Section 4 presents the two econometric models that are used for the analysis, which are the single exit logistic hazard model and the competing risk model. Section 5 explains the database used for the analysis, the sample selection criteria and the variables included. In part 6, we present the results of both the non-parametric analysis, in terms of empirical hazard rates graphs, and the parametric analysis. Finally, some concluding remarks are included in the last section of the paper.

2. Determinants of labour force transitions for workers aged 45-59 in Spain

Figures 1 and 2 show Spanish labour force participation rates and unemployment rates for men and women aged 45-59 from 1987 to 2007. Figure 1 shows dramatic falls in the labour force participation rate for men during the late eighties and mid nineties following, to some extent, the business cycle and resulting in strong increases in the unemployment rate for this group of workers during this period. Since 1995 labour participation rates have recovered but participation has not been as high as it was in the late , while unemployment rates have been steadily decreasing since the mid nineties. In 2001, and for the first time since the , a decrease in participation rates is not followed by an increase in the unemployment rate (although, as can be seen in figure 3, there is an increase in the number of disability pensions granted to individuals aged 45-59 from 2001). Figure 2 shows a very different trend for women in the age bracket 45-59: in a comparable period of time the labour force participation rate has increased steadily whereas the unemployment rate has followed a similar pattern than the unemployment rate for men aged 45-59. This increase in the participation rate can be explained by the generational change that resulted in a replacement of low educated older women by more educated younger women in the same age bracket.

² The jump in 2002 is due to a methodological change in the Spanish labour force survey.

Figures 3 and 4 show the way in which both the unemployment rate and the inflow into disability pensions react to business cycle conditions. In figure 3, we can see that both the unemployment rate and the number of new disability pensions granted to workers aged 45-59 strongly increased during the first half of the nineties when the GDP growth dropped to negative numbers for the first time since the 1980’s. However, since 2000 these two variables have showed an opposite evolution with the unemployment rate remaining reasonable constant between 6% and 8% and the number of new disability pensions granted to this group of workers increasing steadily until 2007.

Figure 4 reflects the relation between the number of pensions granted to workers aged 45-59 and the business cycle. It can be observed that there is a quite surprising perfect negative relation between these two variables so that whenever the GDP increases, the number of disability pensions for this group of workers decreases and vice versa.

Therefore, it seems clear that, at the aggregate level, changes in the number of disability pensions granted to this group of pre-retirement workers and changes in their unemployment rate seem to respond to adjustments in the speed of economic growth. This relationship is also maintained if we broaden the analysis to include all working age individuals, as can be seen in figures 5 and 6 that plot both the GDP growth and the unemployment rate with the rate of new disability pensions to employment for all Spanish workers between 1977 and 2008.

Finally, if we take this aggregate data from 1977 until 2008 and perform a simple regression of the rate of new disability pensions to employment including as explanatory variables the GDP growth and yearly dummies, the results show a negative and significant relation between the GDP growth and the rate of new disability pensions to employment that remains significant when we include other explanatory variables such as the unemployment rate and the GDP per capita (see table 1).

Consequently, this seems to suggest that workers may be strongly affected by business cycle conditions and that their decisions on labour market transitions (particularly from employment to unemployment and disability) may be affected by the level of economic growth³.

3. The Spanish Social Security rules for older people: unemployment and disability

According to the current unemployment rules, there are two kinds of unemployment benefits, an unemployment insurance system (UI) for those workers that contributed while employed but that have been fired from the previous job⁴, and unemployment assistance (UA) for individuals who do not qualify for the unemployment insurance benefits. Individuals entering the UI scheme are entitled to receive 70% of the wages of the last job during the first six months and 60% after that. These quantities are subject to a minimum and a maximum amount. The minimum corresponds to 75% of the minimum wage and the maximum is not fixed and is proportional to the number of dependents. These benefits are paid for a period of onethird of the accumulated job tenure and are only paid for a maximum period of two years⁵.The UA benefits are only paid to individuals with dependents and with an average family income below 75% of the minimum wage. There is a fixed amount paid which corresponds to 75% of the minimum wage. This benefit is only paid for a maximum period of two years⁶ (Bover et al., 2002).

There is, however, an exception to this rule for people aged 52 and more. They can receive 75% of the minimum wage until they reach the official retirement age and the years spent under this scheme are counted as contributive years towards an old-age pension (as the public employment agency, INEM, pays the contributions of the unemployed)⁷.

³ These decisions on labour market transitions might be taken jointly by the employee and the employer or might be unilateral by one of the sides.

⁴ They must have held the job for at least one year.

⁵ After these two years of contributory unemployment benefits, unemployed individuals who are still not working can have access to social assistance benefits.

⁶ There is also a special scheme in Andalucía and Extremadura for agricultural workers who have been employed for 40 days during the year. They are entitled to receive 75% of the minimum wage for 90 to 300 days each year. The number of days depends on their age and number of dependents.

There are also two schemes of disability pensions, depending on whether the disability is temporary or permanent. If the disability is temporary and the risk covered is “common illness”, a period of 180 days of contribution to the system during the last 5 years is required and benefits are granted for a maximum period of 18 months. After that, the individual has to either return to work or enter the scheme of permanent disability⁸. No minimum contribution is required for disabilities resulting from “workrelated accidents”.

For permanent disability situations, three levels of disability are identified which depend on the severity of the injury/illness and which involve different eligibility requirements and pension amounts. Included in the first level (inability to perform the usual job), there is a special provision for which workers with few qualifications from disadvantaged socio-economic circumstances who are older than 55 years⁹ are eligible to receive disability benefits until retirement at age 65.

Between 5 and 15 years of contribution are required when the source of the disability is an ordinary illness and the benefit base is calculated in the same way as oldage pensions. No contributive requirement is needed if the disability is caused by a work related or unrelated accident or by a professional illness. In the case of a workrelated accident or professional illness, the benefit base corresponds to the average wage in the last year of work and for a work-unrelated accident the benefit is calculated as the average annual wage for a 24 month-period that the individual can choose from the last seven years of work. All three schemes are automatically converted to old-age pensions when the individual turns 65 (Jimenez-Martin et al., 2006).

⁷ This rule was introduced in 1989.

⁸ Provided on passing a medical examination that determines the “permanent disability” condition.

⁹ The argument behind this special arrangement is the fact that this group of workers is considered to be in a particularly difficult position for finding a new job.

4. Modelling framework: The hazard regression model

The analysis is divided into two stages. In the first one, we estimate a single exit logistic hazard model to study transitions from employment to non-employment with a particular focus on the duration of employment spells. In the second part of the paper, a competing risk model is estimated in order to distinguish between the various routes out of employment: disability pensions, unemployment or inactivity.

4.1. Single exit logistic hazard model

We apply the logistic model to analyze single exit transitions from employment to non-employment for Spanish workers aged 45-59. We have chosen this discrete time model because our data is recorded in monthly intervals and, as we are dealing with employment contracts, exits are usually observed at the beginning, the middle or the end of the interval. Therefore, the logistic model seemed more appropriate than the complementary log-log model (which is the discrete time representation of a continuous time hazard model). However, a number of tests conducted with the cloglog model provide very similar results to our logistic results so that, for our analysis, both models could be used.

The logistic hazard model estimates the probability that an individual will leave employment at time t given the fact that he has been employed for, at least, t periods (where t is the discrete time duration variable).

We observe a person’s spell until the end of the t month, at which point we observe that the spell is either complete or right censored. Therefore, for each individual in our sample, the probability of a spell being completed at time t given that it has continued, at least, until time t takes the form of:

\[h _ {i} (t, x _ {i} (t), z _ {i} (t)) = \operatorname * {P r} (T _ {i} = t \mid T _ {i} \geq t, x _ {i} (t), z _ {i} (t)) = F [ \alpha_ {0} + \alpha_ {1} (t) x _ {i} (t) + \alpha_ {2} (t) z _ {i} (t) + \theta_ {i} (t) ]\]

(1)

Where,

is the discrete time hazard rate for month t conditional on a vector of exogenous variables; describing personal characteristics of the individual such as age or education level and which include the variables that describe the economic situation of the sector and the region in which the individual works. are the variables that represent duration dependence and depend on the number of months spent in employment. For the purposes of this paper, we characterize duration dependence as a cubic polynomial function of time:

\[\theta_ {1} t + \theta_ {2} t ^ {2} + \theta_ {3} t ^ {3}\tag{2}\]

Where and are the shape parameters, which are estimated together with the slope parameters and the constant.

Therefore, the likelihood contribution of a censored spell is given by:

\[\mathrm{L} _ {\mathrm{i}} = \operatorname * {P r} (T _ {i} > t) = \prod_ {k = 1} ^ {t} \left(1 - h _ {i k}\right)\tag{3}\]

Whereas the contribution to the likelihood function of a complete spell is:

\[\mathrm{L} _ {\mathrm{i}} = \operatorname * {P r} (T _ {i} = t) = \frac {h _ {i t}}{1 - h _ {i t}} \prod_ {k = 1} ^ {t} (1 - h _ {i k})\tag{4}\]

And the likelihood function for the sample as a whole is represented by:

\[\log \mathrm{L} = \sum_ {i = 1} ^ {n} c _ {i} \log \left(\frac {h _ {i t}}{1 - h _ {i t}}\right) + \sum_ {i = 1} ^ {n} \sum_ {k = 1} ^ {t} \log \left(1 - h _ {i k}\right)\tag{5}\]

Where is the censoring indicator:

\[c _ {i} \left\{ \begin{array}{l} 1 \text { if spell is complete } \\ 0 \text { if spell is right censored } \end{array} \right.\]

If we define as a (0,1) variable that is 1 when the observed duration equals t:

\[Y _ {t i} = 1 (T _ {i} = t)\tag{6}\]

Then, the log-likelihood function of the sample for takes the same form as the likelihood function of a standard binary dependent variable model (Jenkins, 1995):

\[\begin{array}{l} \log \mathrm{L} = \sum_ {i = 1} ^ {n} \sum_ {k = 1} ^ {t} y _ {i k} \log \left(\frac {h _ {i k}}{1 - h _ {i k}}\right) + \sum_ {i = 1} ^ {n} \sum_ {k = 1} ^ {t} \log \left(1 - h _ {i k}\right) \\ = \sum_ {i = 1} ^ {n} \sum_ {k = 1} ^ {t} \left[ y _ {i k} \log h _ {i k} + (1 - y _ {i k}) \log (1 - h _ {i k}) \right] \end{array}\tag{7}\]

Thus, the model is estimated by Maximum Likelihood where the hazard rate corresponds to a logit probability; estimation is the same as estimating a sequence of logit models defined on the surviving population at each duration, where the dependent variable equals 1 whenever an exit from employment is observed and 0 otherwise.

4.2. Competing risk model

We further consider the possibility of exit from employment to one of the three following destination states: unemployment; disability pensions and inactivity. With the assumption of independence of the destination-specific hazard rates¹⁰, the discrete hazard rate for exit at time t to any of the three destinations is simply the sum of the destination-specific discrete hazard rates, and

\[h (t) = h _ {u} (t) + h _ {i} (t) + h _ {d} (t)\tag{8}\]

where

is the hazard rate of experiencing a transition from employment to unemployment is the hazard rate of a transition to inactivity and,

is the hazard rate to disability pensions.

The discrete time hazard into one of the m states is equal to the probability of making a transition in interval t, conditional upon surviving up to the beginning of the interval.

We assume a specific form for the conditional destination-specific hazard rate for individual i to destination m in interval t such that¹¹:

\[\mathrm{h} _ {\mathrm{it}} ^ {\mathrm{m}} = \frac {\exp \left(\beta_ {0} ^ {m} + \beta_ {1} ^ {m} X _ {i t} + \beta_ {2} ^ {m} Z _ {i t} + \theta_ {t} ^ {m}\right)}{1 + \sum_ {m = 1} ^ {3} \exp \left(\beta_ {0} ^ {m} + \beta_ {1} ^ {m} X _ {i t} + \beta_ {2} ^ {m} Z _ {i t} + \theta_ {t} ^ {m}\right)}\tag{9}\]

Where, as before, is a vector of personal characteristics, include the variables that describe the economic situation of the sector and the region in which the individual works and is the baseline hazard function which is modelled as a cubic polynomial.

For the given hazard rate described above, the individual worker’s likelihood contribution has the same form than the likelihood of a standard multinomial logit model (Allison, 1982).

\[L _ {i} = \left[ \prod_ {m = 1} ^ {3} h _ {i t} ^ {m} \right] ^ {c _ {t i} ^ {m}} \left[ h _ {i t} ^ {0} \right] ^ {1 - \sum_ {m = 1} ^ {3} c _ {t i} ^ {m}} \left[ \prod_ {\tau = 1} ^ {t - 1} h _ {i \tau} ^ {0} \right]\tag{10}\]

¹⁰ The assumption of independence may be questionable in the context of our analysis but it is still reasonable.

¹¹ We assume that the conditional destination-specific hazard rate is independent from the other competing risks and can be derived from a latent variable model in which the error term has a standard logistic distribution

where is a destination-specific censoring indicator which equals 1 if worker i exits to state m in interval t, and is the conditional probability of making no transition (reference category).

\[h _ {i t} ^ {0} = \frac {1}{1 + \sum_ {m = 1} ^ {3} \exp \left(\beta_ {0} ^ {m} + \beta_ {1} ^ {m} X _ {i t} + \theta_ {t} ^ {m}\right)}\tag{11}\]

\[c _ {i t} ^ {m} = \left\{ \begin{array}{l l} 1 \text { if } c _ {i t} = m \\ 0 \text { if } c _ {i t} = 0 \end{array} \right. \implies c ^ {m}\tag{12}\]

5. Data

The study will use the Continuous Sample of Working Lives (Muestra Continua de Vidas Laborales, MCVL) which is a microeconomic data set based on administrative records compiled from three sources; the Spanish Social Security Administration, the Tax Office and the National Census. There have been four waves until now, 2004, 2005, 2006, and 2007 and it is envisaged that the information will be updated annually. We have used the 2007 wave. It contains a random sample of 4% of all the individuals who, at some point during 2007, had contributed towards the social security system (either by working or being in an unemployment scheme) or had received a contributory pension. The random sample selected contains over one million people.

There is information available on the entire employment history of the workers, including the exact duration of employment, unemployment and pension spells, and for each spell, the firm’s sector and region of activity, the type of contract held, the wages measured as contribution bases, the level of pension and unemployment benefits, among others. There is also some information on personal characteristics such as age, gender, nationality and level of education.

The macroeconomic variables used to capture the economic business cycle are constructed with information derived from the Spanish Instituto Nacional de Estadistica. We measure growth and employability performance of different economic sectors and regions in Spain to capture the variation in the business cycle between times, sectors and regions.

5.1. Sample selection criteria and variables

We select employment spells of individuals in the sample between 45 and 59¹² years old and we follow them until their first exit from employment or until they are censored (last month of age 59 or 2006), whichever occurs earlier. We don’t consider spells of individuals who are unemployed, inactive or receiving disability pensions, as our focus is on transitions out of employment. The analysis is conducted separately for women and men.

There is a maximum of 180 periods of time (15 years*12months) for each individual and the sample selected contains 26.556 women, of which 9.451 leave the labour market (5.163 to unemployment, 653 to disability, 3.635 to inactivity) and 58.684 men, of which 24.883 leave the labour market (14.549 to unemployment, 1.814 to disability, 8.520 to inactivity). We also used a variant of this sample in which we applied the same sample selection criteria laid out above with the extra requirement that the individuals selected need to have been continuously working for the two years prior to entering the sample. With this restriction we exclude from our sample individuals with short temporary contracts (which can be signed for one to three years) and with a more unstable job situation who probably have a different labour market behaviour than workers with longer term contracts. At the same time, these workers may be affected in a different way by business cycle conditions.

5.2 Explanatory Variables

First, we have included three variables that capture duration dependence. The baseline hazard is modelled as a cubic polynomial function of duration. The level of education obtained is incorporated with three dummies: The first one captures individuals who cannot read or write or who dropped out of education before finishing high school. The second one includes individuals holding a high school diploma (or equivalent) whereas the third one is for people with a bachelor’s degree and other higher education. Regional and cohort dummies are also included as regressors as well as dummies for ages 52, 55 and 58 to capture the effect of the special unemployment and disability schemes available at these ages on transitions out of employment. The dummy variable at age 58 is introduced in order to capture the effect of individuals who have contributed for enough years to receive two years of unemployment benefits at the age of 58 before joining the early retirement scheme when they turn 60.

¹² We don’t include individuals aged 60 because, at that age, they can already access some early retirement schemes and this would constitute another possible exit route which is not the focus of our paper.

There are six dummy variables that represent the economic activity of the firm in which the individual is working, one for each sector of activity (primary, secondary and tertiary sectors, public administration, education and social services). There is also a variable that describes the number of workers in the firm and three dummy variables that account for partial/full time employment, temporary(short-term)/permanent contract of employment and voluntary/involuntary termination of the contract¹³. Two more variables describe the employment history of the worker. The first one captures the number of years that the individual worked from age 15 until the last month of age 44. The second one represents the tenure of the worker and is calculated as the number of years spent working in the employment spell with which the individual enters the sample.

Monetary variables:

The variable that we include in order to capture the monetary incentives of the worker is the contributions to the social security administration, which is a double censored (from below and above) version of the wages (see Boldrin et al., 2004, for a procedure to recover wages from contributions). We have information about this variable for each month that the individual is contributing to the Social Security Administration.

Macroeconomic variables:

In order to capture the macroeconomic conditions at each point in time, we construct a variable that describes the unemployment rate of each of the 18 autonomous communities in Spain and we assign this value to each individual according to the autonomous community in which they work. Depending on the economic sector in which the individual is reported to work, we also assign them the value of two variables that capture both the unemployment rate and the average number of hours worked in each of the 17 economic sectors in which the National Institute of Statistics divides the Spanish economy. These macroeconomic variables vary over time. We also include an interaction term of these variables with the logarithm of duration. Finally, we include the GDP growth in order to capture national business cycle trends that are not captured by the regional or sector variables.

¹³ Partial Contract: A dummy variable which is 1 if the contract of employment is a part-time contract.

Temporary Contract: Is a dummy variable constructed from a variable describing the type of contract and equals 0 if the contract is temporary or short-term and 1 if the contract is permanent.

6. Results

6.1.Non-parametric analysis

If we take a look at the empirical hazard rates derived from our sample, we can already see the strong impact on the hazard rate of some of the explanatory variables included in the analysis. The empirical hazard for a given number of months is calculated as the total number of exits from employment in each month divided by the population in employment at the beginning of that month. This summarizes the sample probability of leaving employment at each particular point in time. However, for simplicity purposes, we present this information compiled by age¹⁴ instead of months.

Figures 7 and 10, show the interval hazard rates for men and women respectively. Both figures show three peaks of the hazard rate at ages 52, 55 and 58, which are attributable to the special schemes on the unemployment and disability pension legislation. However, in the case of women, the effect at age 58 is considerably stronger than the two effects at ages 52 and 55 which are approximately of the same size. For Spanish men, the increase in the hazard rate is more progressive and higher at each of the three age intervals.

Figures 8 and 11 plot the empirical hazard rates according to temporary/permanent contract and it becomes clear that it is workers with a temporary contract (men and women) the ones leaving the labour force between ages 45 and 59. Individuals with a permanent contract tend to leave less the labour force before reaching the retirement age. This tendency for temporary workers to leave the labour market could be interpreted as a means of escaping from a very uncertain and unstable job situation, as it is probably very difficult for these older workers to find a job once their temporary contract expires. In fact, as stressed above, the special unemployment and disability schemes at ages 52 and 55 were already introduced in an attempt to provide a minimum degree of income stability for older workers who are unable to find a job after being fired due to industrial relocation or modernization plans. However, there is another possible interpretation; the choice to work under a temporary contract might already be a lifestyle choice indicating increased preference for leisure time. Unfortunately, we don’t have information on preferences or the difficulties of these workers to find a new job.

¹⁴ Considering the fact that each individual enters the sample on the month that he/she turns 45 years old, then after the first 12 months in the sample, everybody turns 46 (if they are still in the sample).

Finally, figures 9 and 12 show the hazard rates for men and women with different educational levels. For women, the hazard rate clearly decreases according to the level of education and the peak at age 58 is higher for those with a level of education below a bachelor degree. For men, there is not much difference in the hazard rates for individuals with or without a high-school diploma although hazard rates for those with bachelor’s degrees are slightly lower.

6.2. Results: Hazard Regression analysis

In this section we estimate the effect of personal and job-related characteristics, business cycle and age dummy variables on the hazard of leaving employment, while controlling for duration dependence through a cubic polynomial function of time.

The results are shown in tables 2, 3, 4 and 5 for the case of men and 6, 7, 8 and 9 for women.

We have estimated two models for each gender, a logistic hazard regression model for single exit from employment to a non-employment situation and a competing risk model to assess the differences between exits from employment to either unemployment, disability pensions or inactivity. As a result of the non-linearity of the model estimated and the inclusion of interaction terms in the regression, we only discuss the results of the regressions in terms of sign and statistical significance of the coefficients of the explanatory variables. The size of these effects is discussed in the subsequent section through calibration of the effects of changes in selected variables on the predicted hazard rates.

6.2.1 Duration dependence

We have modelled the baseline hazard as a cubic polynomial function of duration because, together with the age dummies, this specification reproduces the shape of the empirical interval hazard rate reasonably closely.

We have also tried to estimate the model with a piece-wise constant specification of the baseline hazard in order to allow for greater flexibility as we don’t have to assume any functional form for the baseline hazard. However, in the competing risk case, when duration was represented by time dummies (one for each month), several of these dummy variables were not identified as for certain months there was no exit observed to one of the exit routes¹⁵. Furthermore, in this specification, age dummies could not be included as they caused multicollinearity problems with the duration variables. On the other hand, if we increase the number of months included in each dummy variable to reduce the number of these variables and allow for a better identification of the model, we are unable to capture age effects.

In order to keep consistency between the two specifications that we use (logistic and competing risk models) and to be able to test the effects of some unemployment and disability schemes through age dummies, we decided to represent duration dependence with a cubic polynomial function of time.

In addition, the coefficients of the explanatory variables are very similar in the model in which duration dependence is modelled as a piece-wise constant and the model with a cubic polynomial baseline hazard so that the results that we report correspond to the model with the cubic polynomial baseline hazard.

6.2.2 Individual features and employer, firm and industry characteristics

Our findings confirm that the effect of personal characteristics is similar to what the literature on the topic suggests. Individuals with higher education and individuals with a part-time contract show lower hazard rates. Therefore, the idea of introducing partial retirement schemes in order to keep older workers active in the labour market would seem like a positive alternative as part-time workers leave employment less than full-time workers. On the other hand, temporary contracts are not a good policy to retain workers in employment, as the hazard rate increases substantially for individuals with a temporary contract.

¹⁵ The same problem occurred if we grouped the time dummies in small groups of 3 or 4 months for each time dummy representing duration.

In terms of regions, Madrid has the lowest hazard rate, followed by the Mediterranean region, Catalonia, Southern Spain and the Centre (Castilla), while the Northern region has the highest hazard rate.

Social security contributions (our proxy for wages) show a negative sign so that employees with higher salaries have a lower hazard rate as the monetary incentives to stay employed are stronger compared to receiving unemployment or disability benefits. This effect increases with elapsed duration.

The coefficient for the number of years contributed between ages 15 and 45 is positive and significant suggesting that individuals who have been contributing for a longer period to the social security administration have a higher hazard to leave the labour force possibly because they fulfil the contributory eligibility conditions to receive unemployment or disability benefits. On the other hand, individuals who have been working for longer in the current job at the time of entering the sample have a lower hazard rate, which could be interpreted as having a stronger attachment to the employer the longer you work for the same firm.

6.3 Business cycle conditions and age effects

Business cycle effects are measured in terms of the unemployment rate of the autonomous community in which the individual is working, the unemployment rate of that firm’s sector of activity, the average number of hours worked in the sector of activity and, finally, the GDP growth rate to capture national business cycle trends that are not captured by the regional or sector variables.

The first variable, the unemployment rate of the autonomous community shows the anticipated positive sign for both males and females, but it is stronger and more significant in the case of men. The negative sign of the interaction term with duration suggests that the effect of the unemployment rate is lower for individuals who remain employed for a longer period of time.

However, the coefficient of the unemployment rate of the sector of activity of the firm has a negative sign, which is unexpected and is very significant for both genders while the interaction term displays a positive and significant sign.

The higher the number of hours worked in the sector, the lower the hazard rate of leaving employment and this effect is stronger for individuals who have been employed for longer.

The GDP growth rate has the expected negative (and very significant) sign reinforcing the idea that workers tend to leave employment at times when the economy is at its lowest rate of growth.

For men, the three age peaks at 52, 55 and 58 have a higher and very significant hazard rate whereas for the case of women only the 52 variable turns out to be significant. However, all of them have a positive coefficient.

The results of the competing risk model present a very interesting finding for the disability option. If we take a look at table 4, the coefficient for the unemployment rate of the autonomous community is positive and significant and the GDP growth coefficient is negative and also very significant suggesting that exit to disability is very much influenced by the economic cycle of both the region and the country as a whole. The medical requirements needed to access this route are not as strict as to prevent workers from using this route as an alternative to claiming unemployment benefits during bad economic times. In this case, a higher sector unemployment rate is also reflected in an increased hazard rate of leaving employment to receive disability benefits.

For the women, the coefficient for age 55 is positive and significant which implies that many women are using the special scheme in the disability legislation for workers aged 55.

We have also estimated the same competing risk model using the sample of people observed as working continuously for the two years prior to entering the sample. The aim is to test whether the results are any different for workers with a more stable job situation (as we exclude workers with very short-term contracts).

The results of this estimation are shown in table 5. The sign of the coefficients do not change with respect to the original sample but the sector unemployment rate has now a stronger effect on hazard rates than the unemployment rate of the autonomous community and the GDP growth. Therefore, a deterioration of the economic conditions in a particular economic sector increases the hazard rate of making a transition to disability benefits for individuals employed in that sector. As a result, individuals with a more stable employment situation are more affected by the economic situation of the sector of activity and by the economy-wide business cycle conditions than by the economic situation of the autonomous community in which they work.

6.4 Effects on hazard rates

We finish by analyzing the relative size of the effects outlined in the previous section. Of all the variables, we have chosen to concentrate on the macroeconomic variables, as they are the main focus of our paper, and on the variable that describes the type of temporary/permanent contract, so as to illustrate the size of the differences on the hazard rates of these two groups of workers. Comparisons of size are complicated as they depend on the reference group of individuals and the values chosen for the evaluation. In figure 13 we compare the predicted hazard rates of a representative individual¹⁶ with either a temporary or a permanent contract of employment. For that comparison, we keep macroeconomic variables at their sample means: 15.02% unemployment rate for the autonomous community; 6.65% unemployment rate for the sector of activity; 40.24 average number of hours worked and 2.92% GDP growth.

According to our predictions, a person in our sample has a 1.3 percentage point higher hazard rate of leaving employment if he has a temporary contract of employment than if he has a permanent contract.

Figure 15 shows that when GDP growth is at the lowest point of our sample period (-1.8%) the hazard of a 58 year-old men leaving employment is 4 percentage points higher than at times of maximum GDP growth (6.7%). For a 59 year-old men the difference is of 2 percentage points.

The effect of changing the unemployment rate to the maximum and minimum value of the sample period is not as strong as the effect of the GDP growth but it is still important: for a 58 year-old men the hazard rate is 1 percentage point higher when the unemployment rate is at its maximum (34.9%) than when it is at its minimum rate (3.47%) (figure14).

¹⁶ The representative individual has the values of the reference category of the regressions for the explanatory variables that are dummies and the mean values of the sample for the other explanatory variables such as contributions (wages), number of workers in the firm, etc…

7. Concluding remarks

In this paper we have investigated the effect of business cycle conditions in Spain on labour market transitions from employment to disability pensions, unemployment and inactivity for men and women aged 45-59.

We have also described two special schemes in the unemployment and disability legislation which induce people to make the transitions when they reach a certain age.

For this analysis, we have estimated two kinds of survival analysis models, a logistic hazard model for single exit transitions from employment to all the other nonemployment situations and a competing risk model for transitions from employment to each of the other three states: unemployment, disability and inactivity.

We have used two databases for our analysis: on the one hand, the so-called “Muestra Continua de Vidas Laborales”, which is an administrative database, obtained from a random draw of the Spanish Social Security records and, on the other hand, from the Spanish National Institute of Statistics we have derived the macroeconomic information on the business cycle conditions.

Our main findings can be listed as following:

• Business cycle conditions are found to affect not only transitions from employment to unemployment, but also transitions from employment to disability. This effect holds both for men and women. Individuals are more likely to leave employment to receive disability pensions when the economic situation of the country is less favourable. Disability pensions are being used as an alternative means of leaving the labour market for individuals who find it difficult to get a new job (because of age) and who (most probably) do not fulfil the requirements to receive unemployment benefits.

• For exits to disability pensions, the effect of the economic environment in the sector of activity proves to be stronger than the effect of the economic environment in the region. However, the opposite can be said for the unemployment option, as a higher unemployment rate in the sector of activity in which the individual is working reduces the hazard of leaving employment to receive unemployment benefits. Furthermore, this effect is stronger than the effect of the unemployment rate of the autonomous community.

The special schemes in the unemployment and disability legislation mentioned in this paper that relax the conditions to access these benefits at ages 52, 55 and 58, trigger specific age effects in transitions.

REFERENCES

1. ALBA-RAMIREZ, A. (1997): “Labour Force Participation and Transitions of Older Workers in Spain”, Working paper 97-39 U. Carlos III Madrid.
2. ALLISON, P. D. (1982): "Discrete-Time Methods for the Analysis of Event Histories," in Sociological Methodology, ed. by S. Leinhardt. San Francisco, CA: Jossey-Bass Publishers.
3. ARGIMON, I., C. I. GONZALEZ, and R. VEGAS (2007): "Retirement Behaviour and Retirement Incentives in Spain," Banco de España.
4. BENITEZ-SILVA, H., J.I. GARCIA-PEREZ and S. JIMENEZ-MARTIN (2008): “A Cross Country Comparison of the Effects of Social Security Policies and Employment Uncertainty on the Labour Supply and Claiming Behaviour of Older Workers”, MIMEO.
5. BERG, G. J. V. D., and B. V. D. KLAAUW (2000): "Combining Micro and Macro Unemployment Duration Data," IZA Discussion Paper No. 109.
6. BLANCO, A. (2000): "The Decision of Early Retirement in Spain," FEDEA, EEE 76.
7. BLAU, D. M. (1994): "Labor Force Dynamics of Older Men," Econometrica, 62, 117- 156.
8. BLONDAL, S., and S. SCARPETTA (1998): "Falling Participation Rates among Older Workers in OECD Countries: The Role of Social Security Systems," OECD Economic Department Working Paper.
9. BOLDRIN, M., S. JIMENEZ-MARTIN, and F. PERACCHI (1999): "Social Security and Retirement in Spain," in Social Security Programs and Retirement around the World, ed. by J. Gruber, and D. Wise: Chicago University Press for the NBER.
10. — (2004): "Micro-Modelling of Retirement Behaviour in Spain," in Social Security and Retirement around the World: Micro-Estimation, ed. by J. Gruber, and D. Wise: Chicago University Press for the NBER.
11. BOVER, O., M. ARELLANO, and S. BENTOLILA (2002): "Unemployment Duration, Benefit Duration and the Business Cycle," Economic Journal, 112, 223-265.
12. BOVER, O., and R. GOMEZ (2004): "Another Look at Unemployment Duration: Exit to a Permanent Vs. A Temporary Job," Investigaciones Economicas, XXVIII, 285-314.
13. CAIRO, I. (2007): "An Empirical Analysis of Retirement Behaviour in Spain: Partial Vs. Full Retirement," Unpublished.
14. FISCHER, J. A. V., and A. SOUSA-POZA (2006): "The Institutional Determinants of Early Retirement in Europe," Universitat St.Gallen Discussion Paper No.2006- 08.
15. GARCIA-PEREZ, I. and N.AHN (2002): “Unemployment Duration and Workers Wage Aspirations in Spain”, Investigaciones Económicas.
16. GARCIA-PEREZ, I. and Y.REBOLLO (2005): “Wage Changes Through Job Mobility in Europe: A Multinomial Endogenous Switching Approach”, Labour Economics.
17. GARCIA-PEREZ, I. (2006): “Non-stationary job search when jobs do not last forever: a structural estimation to evaluate alternative unemployment insurance systems”, Journal of Applied Econometrics, 21(2), 245-272.
18. GARCIA-PEREZ, I. and Y.REBOLLO (2007): “The Use of Permanent and Temporary Jobs Across Spanish Regions: Do Unit Labour Cost Differentials Offer an Explanation?”, Moneda y Crédito.
19. GARCIA-PEREZ, I., and A. SANCHEZ-MARTIN (2008): "Social Security and the Search Behavior of Workers Approaching Retirement," Working Paper Universidad Pablo de Olavide Sevilla..
20. GARCIA-PEREZ, I., S. JIMENEZ MARTIN, and A.R. SANCHEZ MARTIN (2008): “Incentivos Implicitos en el Sistema de Proteccion Social en Espana y sus efectos sobre las pautas de retiro”, Informe Proyecto FIPROS 2007/17.
21. GARCIA-PEREZ, I. and Y.REBOLLO (2009): “The Use of Permanent Contracts Across Spanish Regions: Do Regional Wage Subsidies Work?”, Investigaciones Económicas.
22. GIELEN, A. C. (2007): "Working Hours Flexibility and Older Workers' Labor Supply," IZA Discussion Paper No. 2946.
23. GREENE, W. H. (2008): "Econometric Analysis," Sixth Edition, Prentice Hall, New Jersey.
24. GRUBER, J., and D. A. WISE (1999): "Social Security and Retirement around the World," Chicago University Press for the NBER.
25. GRUBER, J., and D. A. WISE (2004): “Social Security and Retirement around the World: Micro Estimation”, Chicago University Press for the NBER.

GRUBER, J., and D. A. WISE (2007): “Social Security and Retirement around the World: Fiscal Implications”, Chicago University Press for the NBER.

1. GUTIERREZ-DOMENECH, M. (2006): "The Employment of Older Workers," "La Caixa" Working Paper Series No.04/2006.

HAARDT, D. (2006): "Transitions out of and Back to Employment among Older Men and Women in the UK," ISER Working Paper. 2006-20. Colchester: University of Essex.

1. JENKINS, S. (2004): "Survival Analysis," Colchester: Institute for Social and Economic Research, University of Essex. .
2. JENKINS, S. P., and C. GARCIA-SERRANO (2004): "The Relationship between Unemployment Benefits and Re-Employment Probabilities: Evidence from Spain," Oxford Bulletin of Economics and Statistics, 66, 305-329.

JIMENEZ-MARTIN, S., J. M. LABEAGA, and C. VILAPLANA (2006): “A sequential model for older workers’ labor transitions after a health shock”, Health Economics, 15(9), 1033-1059.

JIMENEZ-MARTIN, A., and S. SANCHEZ-MARTIN (2004): "Social Security Rules, Monetary Incentives and Retirement Behaviour in Spain," Hacienda Publica Española, 2004.

1. JIMENEZ-MARTIN, A., and S. SANCHEZ-MARTIN (2007): “An evaluation of the life-cycle effects of minimum pensions on retirement behavior”, Journal of Applied Econometrics, 22(5), 923-950, 2007.

MEGHIR, C., and E. WHITEHOUSE (1997): "Labour Market Transitions and Retirement of Men in the UK," Journal of Econometrics, 79, 327-354.

NARENDRANATHAN, W., and M. STEWART (1993): "Modelling the Probability of Leaving Unemployment: Competing Risk Models with Flexible Base-Line Hazards," Applied Statistics, 42, 63-83.

Figure 1: Unemployment and labour force participation rates for men aged 45-59 in Spain, 1987-2007¹⁷

Figure 2: Unemployment and labour force participation rates for women aged 45- 59 in Spain, 1987-2007¹⁸

Source: Spanish National Institute of Statistics: “Encuesta Poblacion Activa” (EPA).

¹⁷ This is 3-month interval data.

¹⁸ Although the scale of the graph of labour force participation rate for women is greater than that for men, the evolution is much less erratic than it is for men. Even if we decrease the scale of the graph of women, there is no different pattern observed during the crisis years (as it is the case for men).

Figure 3: Unemployment rate and new disability pensions granted each year to individuals aged 45-59 in Spain, 1980-2008 Unemployment & New Disability Pensions: Individuals Aged 45-59 Source: Unemployment rate of individuals aged 45-59: Spanish National Institute of Statistics (EPA), Number disability pensions granted each year to individuals aged 45-59: Own calculations from MCVL.

Figure 4: GDP growth and new disability pensions granted each year to individuals aged 45-59 in Spain, 1980-2008 Source: GDP growth: Spanish National Institute of Statistics, Number of disability pensions granted each year to individuals aged 45-59: Own calculations from MCVL.

Figure 5: GDP growth and rate of new disability pensions to employment in Spain (individuals of all ages), 1978-2008 Source: Spanish National Institute of Statistics (EPA).

Figure 6: Unemployment rate and rate of new disability pensions to employment in Spain (individuals of all ages), 1978-2008

Source: Spanish National Institute of Statistics (EPA).

Table 1: Regressions results; Dependent variable: rate of new disability pensions to employment

Figure 7: Interval hazard rate men 45-59

Figure 8: Interval hazard rate men 45-59 by permanent/temporary contract of employment

Figure 9: Interval hazard rate men 45-59 by education level

Figure 10: Interval hazard rate women 45-59

Figure 11: Interval hazard rate women 45-59 by permanent/temporary contract of employment

Figure 12: Interval hazard rate women 45-59 by education level

Figure 13: Interval hazard rate exit to Unemployment; men & women 45-59. Figure 14: Interval hazard rate exit to Disability; men & women 45-59.

Figure 15: Interval hazard rate exit to Inactivity; men & women 45-59.

Table 2: Means and standard deviations. Sample of men 45-59

Table 3: Determinants of the probability of leaving employment, Spanish male aged 45-59 (logistic hazard regression model)

Table 4: Determinants of the probability of going to disability pensions, unemployment and inactivity Spanish male aged 45-59 (competing-risk hazard regression model)

Table 5: Business cycle coefficients, Spanish male 45-59 conditional on working two years before entering the sample (competing risk hazard regression model)

Table 6: Means and standard deviations. Sample of women 45-59

Table 7: Determinants of the probability to leave employment, Spanish female aged 45-59 (logistic hazard regression model)

Table 8: Determinants of the probability of going to disability pensions, unemployment and inactivity, Spanish female aged 45-59 (competing-risk hazard regression model)

Table 9: Business cycle coefficients, Spanish female 45-59 conditional on working two years before entering the sample (competing risk hazard regression model)

Figure 13: Predicted hazard rate men 45-59 by temporary/permanent contract of employment

Figure 14: Predicted hazard rate men 45-59 by minimum/maximum unemployment rate of the sample period

Figure 15: Predicted hazard rate men 45-59 by minimum/maximum GDP growth of the sample period

1. 2009-25: “Business Cycle Effects on Labour Force Transitions for Older People in Spain”, Sergi Jiménez-Martín y Judit Vall Castello.
2. 2009-24: “Inequality of Opportunity and Growth”, Gustavo A. Marrero y Juan G. Rodríguez.
3. 2009-23: “A Characterization Of The Judicial System In Spain: Analysis With Formalism Indices”, Juan S. Mora.
4. 2009-22: “Anthropometry and Socioeconomics in the Couple: Evidence from the PSID”, Sonia Oreffice y Climent Quintana-Domeque.
5. 2009-21: “Stimulating Graduates' Research-Oriented Careers: Does Academic Research Matter?”, Mauro Sylos Labini y Natalia Zinovyeva.
6. 2009-20: “Papers or Patents: Channels of University Effect on Regional Innovation”, Robin Cowan y Natalia Zinovyeva.
7. 2009-19: “Innovation, Tangible and Intangible Investments and the Value of Spanish Firms”, Aitor Lacuesta, Omar Licandro, Teresa Molina y Luis A. Puch.
8. 2009-18: “Estimation of Elasticity Price of Electricity with Incomplete Information”, Xavier Labandeira, José M. Labeaga y Xiral López-Otero.
9. 2009-17: “MEDEA: A DSGE Model for the Spanish Economy”, Pablo Burriel, Jesús Fernández-Villaverde y Juan F. Rubio-Ramírez.
10. 2009-16: “Greenhouse gases emissions, growth and the energy mix in Europe: A dynamic panel data approach”, Gustavo A. Marrero.
11. 2009-15: “Impact of the Rise in immigrant unemployment on public finances”, Pablo Vazquez, Mario Alloza, Raquel Vegas y Stefano Bertozzi.
12. 2009-14: “Responding to Financial Pressures. The Effect of Managed Care on Hospitals´ Provision of Charity Care”, Núria Mas.
13. 2009-13: “Domestic Transport Cost Reductions and Firms’ Export Behaviour”, Pedro Albarran, Raquel Carrasco y Adelheid Holl.
14. 2009-12: “Compatibility with Firm Dominance”, María Fernanda Viecens.
15. 2009-11: “Pricing Strategies in Two-Sided Platforms: The Role of Sellers’ Competition”, María Fernanda Viecens.
16. 2009-10: “Scheduled Service Versus Personal Transportation: the Role of Distance”, Volodymyr Bilotkach, Xavier Fageda y Ricardo Flores-Fillol.
17. 2009-09: “Social Preferences and Strategic Uncertainty: An Experiment on Markets and Contracts”, Antonio Cabrales, Rafaele Miniaci, Marco Piovesan y Giovanni Ponti.
18. 2009-08: “Hidden Information, Bargaining Power and Efficiency: An Experiment”, Antonio Cabrales, Gary Charness y Marie Claire Villeval.
19. 2009-07: “Democracy and the curse of natural resources”, Antonio Cabrales y Esther Hauk.
20. 2009-06: “Social Interactions and Spillovers: Incentives,Segregation and Topology”, Antonio Cabrales, Antoni Calvó-Armengol e Yves Zenou.
21. 2009-05: “Chance Constrained Programming with one Discrete Random Variable in Each Constraint”,Emilio Cerdá Tena y Julio Moreno Lorente.
22. 2009-04: “Economic Value of Weather Forecasting Systems Information: A Risk Aversion Approach”, Emilio Cerdá Tena y Sonia Quiroga Gómez.
23. 2009-03: “Population Ageing, Inequality and the Political Economy of Public Education”, Francisco Martínez-Mora.
24. 2009-02: “Real Wages over the Business Cycle: OECD Evidence from the Time and Frequency Domains”, Julian Messina, Chiara Strozzi y Jarkko Turunen.
25. 2009-01: “The Determinants Of Misreporting Weight And Height: The Role Of Social Norms”, Joan Gil y Toni Mora.
26. 2008-42: “Social Security Incentives, Exit from the Workforce and Entry of the Young”, Michele Boldrin, Pilar García-Gómez y Sergi Jiménez-Martín.
27. 2008-41: “The Evolution and Main Determinants of Productivity in Brazilian Electricity Distribution 1998- 2005: an Empirical Analysis”, Francisco Javier Ramos-Real, Beatriz Tovar, Mariana Iootty, Edmar Fagundes de Almeida y Helder Queiroz Pinto Jr..
28. 2008-40: “Immigration and Housing Prices in Spain”, Simón Sosvilla.
29. 2008-39: “Modeling the Immigration Shock”, Ana Montes y Michele Boldrin.
30. 2008-38: “Immigration and the Demand for Health in Spain”, Sergi Jiménez, Natalia Jorgensen y José María Labeaga.
31. 2008-37: “Immigration and Students' Achievement in Spain”, Natalia Zinovyeva, Florentino Felgueroso y Pablo Vázquez.