Partial Disability System and Labor Market Adjustment: The Case of Spain

Transcription

1 Upjohn Institute Working Papers Upjohn Research home page 2013 Partial Disability System an Labor Market Ajustment: The Case of Spain Jose I. Silva University of Kent Juit Vall-Castello Universitat e Girona Upjohn Institute working paper ; **Publishe Version** Labour Economics 48 (October 2017): Citation Silva, Jose I., an Juit Vall-Castello "Partial Disability System an Labor Market Ajustment: The Case of Spain." Upjohn Institute Working Paper Kalamazoo, MI: W.E. Upjohn Institute for Employment Research. wp This title is brought to you by the Upjohn Institute. For more information, please contact ir@upjohn.org.

2 Partial Disability System an Labor Market Ajustment: The Case of Spain Upjohn Institute Working Paper No Jose I. Silva University of Kent, UK an Universitat e Girona, Spain jose.silva@ug.eu Juit Vall-Castello Universitat e Girona an Centre for Research in Economy an Health, Universitat Pompeu Fabra juit.vall@ug.eu July 2013 ABSTRACT Although partially isable iniviuals in Spain are allowe to combine the receipt of isability benefits with a job, the empirical evience shows that employment rates for this group of iniviuals are very low. Therefore, in this paper we construct labor market moel with search intensity an matching frictions in orer to ientify the incentives an isincentives to work provie by the partial isability system in Spain from the point of view of both isable iniviuals an employers. Accoring to the moel, the high employment rate gap observe between nonisable an isable workers can be partially explaine by the presence of a lower level of prouctivity an higher searching costs among isable iniviuals that iscourage them from looking for jobs. Moreover, the esign of the Spanish Disability System also contributes in explaining this gap. We also analyze the role of business cycle conitions in shaping the labor market transitions of isable iniviuals. JEL Classification Coes: I18, J64, J68 Key Wors: isability system, job search intensity, flow analysis.

3 Partial Disability System an Labor Market Ajustment: The Case of Spain In recent years, isability policies have attracte particular attention in OECD countries because they represent an important source of public spening an also because societies are increasingly concerne about the nee to strengthen the integration of people with isabilities. For these reasons, the possibility of increasing the number of isable iniviuals who work is regare as a goo strategy to ecrease the pressures on the financial stability of the social security systems as well as to achieve the social integration of people with isabilities. As the OECD notes, Helping (isable) people to work is potentially a win-win policy: it helps people avoi exclusion an have higher incomes while raising the prospect of more effective labor supply an higher economic output in the long term (OECD 2007). The promotion of the employment of isable iniviuals is particularly relevant in Spain, where partially isable iniviuals are allowe to combine the receipt of (part) of the isability benefits with a job without any limit on the maximum wage or number of hours worke. Nonetheless, the country reporte an average employment rate of just 11 percent for the group of people with partial isability from 2001 to 2011, which contrasts with the observe rate for nonisable employees (79.9 percent). The employment rate falls to less than 6 percent when only consiering people with isabilities in the age bracket (which represents 45 percent of all the partially isable iniviuals in Spain). Therefore, in this paper we perform an analysis of the incentives an isincentives to work provie by the partial isability system in Spain from the point of view of both isable iniviuals an employers. At the same time, we also ientify the role of business cycle conitions in shaping the labor market transitions of isable iniviuals using ata for the 1

4 perio , which inclues the last expansion an recession of the Spanish economy. The central goal of the stuy is to isolate the role of the isability system in explaining the low employment rates of partially isable iniviuals in Spain vis-à-vis their nonisable counterparts. The final aim is to reach some conclusions about the types of policy initiatives that coul be more effective in increasing both iniviual incentives to work as well as employers incentives to hire isable workers. To o this, we consier a search an matching moel of iniviuals with isabilities an their interaction with nonisable iniviuals in the search for jobs. Aitionally, we also inclue in the moel the hiring ecisions mae by companies an the incentives available in the legislation to hire isable workers. We assume that, ue to their isabling conition, isable workers are, on average, less prouctive an incur higher job searching costs than nonisable iniviuals. The presence of a prouctivity gap in our moel is consistent with the recent empirical evience presente by Malo an Pagan (2012). 1 They show that between 68 an 74 percent of the Spanish wage ifferential between nonisable an isable workers is ue to the ifferences in workers characteristics, which generate a prouctivity gap of nonisable workers with respect to their isable counterparts. 2 In contrast, the authors show that less than 30 percent of the wage gap between these two groups of workers can be attribute to the presence of iscrimination. 3 1 This paper uses the Oaxaca-Bliner wages ecomposition metho for Spain an other European countries. 2 Because the authors use ata from the ECHP survey, they efine isable workers as iniviuals that are hampere in their aily activities. 3 In their paper, Malo an Pagan (2012) show that Spanish workers with isabilities hampere in their aily activities earn between 1.71 an 1.44 euros per hour less than workers who are not isable. They also fin that most of this wage gap is relate to workers an jobs characteristics that lea to a prouctivity gap. Moreover, the wage ifferences between nonisable an isable workers who are not hampere for their aily activities are not statistically significant, so that the presence of iscrimination is not empirically prove. Similar results are foun for most of the 11 European countries consiere in their ata. 2

5 We calibrate an simulate the structural moel to match a number of stylize facts observe in the aministrative ata provie by the Spanish Social Security Aministration (the Continuous Sample of Working Lives). Our simulate moel helps to unerstan the ifferences between isable an nonisable workers. More in etail, it simulates an employment rate of 24.0 percent among isable workers, which is much lower than the simulate one for nonisable iniviuals (79.9 percent). The moel also shows that the job fining rate for isable workers is much lower an the job separation rate is higher than for nonisable workers. These results are in line with the ata presente in this paper. The sensitivity analysis shows that both workers an firms are sensitive to the main policy parameters, especially for those remaining operative uring the ifferent labor market conitions. Thus, accoring to the moel, part of the employment rate gap observe between nonisable an isable workers can be attribute to the esign of the Spanish Disability System, which contributes to generate a 64 percent job search intensity gap between isable an nonisable iniviuals. Part of this lower search intensity can be explaine by the presence of cost gaps in both prouctivity an job searching that iscourage isable workers from looking for jobs. However, our moel also inicates the level of isability benefits for isable workers increases their employment opportunity costs an, therefore, reuces their job search intensity. This result is in line with the one reporte by Silva an Vall-Castello (2012a), who focus on the employment effects of an increase in isability benefits that is grante to isable iniviuals who turn 55. They esign, calibrate, an simulate a structural labor supply moel of job search with labor force participation ecisions of iniviuals with isabilities. They show that the employment rate for isable iniviuals is lower because the increase in the level of isability benefit generates the incentive to stay out of the labor force. However, this stuy only 3

6 reprouces the behavior of isable iniviuals without incluing the interactions between isable an nonisable workers, an it also neglects the role of employers. Our simulate moel also shows that 87.0 percent of the times the firm an the isable worker meet, a new employment relationship is create. This result occurs because of the policy incentives in place for employers to hire isable workers. In other wors, the moel suggests that although firms are willing to hire isable iniviuals, these workers look for jobs with much less intensity than nonisable iniviuals. Moreover, since isable workers receive isability benefits, they have a higher ajuste employment opportunity cost an, therefore, are separate from their jobs with higher frequency than nonisable ones. In terms of policy interventions, the sensitivity analysis shows that the employment rate gap between isable an nonisable workers can be consierably reuce by ecreasing the percentage of the regulatory base receive by partially isable iniviuals, by increasing the euction to Social Security Contributions pai by the employer, or by increasing the tax euction for isable workers. In contrast, the moel shows that transitory lump-sum subsiies have a much lower impact in the employment rate of isable iniviuals. In contrast to the strong incentive to hire isable iniviuals from the unemployment status, employers o not ten to maintain their workers in the firm when they receive the negative health shock an become partially isable. Accoring to our moel, less than 10 percent of the employees who become isable are kept in the same firm. This result is in line with the low employment survival rate for new isable employees observe in the ata (17.3 percent). It is important to emphasize that, in this case, firms o not receive a lump-sum tax if they keep the new isable employee in the firm. 4

7 Finally, the moel shows that the employment rate, the job fining rate, an the job survival rate in the same firm are positively correlate with the gross omestic prouct (GDP). This result goes in line with the procyclical behavior of these rates shown by the ata. Furthermore, the job estruction rate is negatively correlate with GDP, which is in line with its empirical countercyclical response. Besies the aforementione work by Silva an Vall-Castello (2012), we have foun no literature that stuies the job search process of isable iniviuals. There is also no literature that analyzes the matching problems in the labor market for iniviuals with isabilities through the use of structural moels. However, two stuies analyze the labor supply behavior of iniviuals with isabilities using life-cycle moels in the Unite States: Benitez-Silva, Buchinsky, an Rust (2010) an Yin an Benitez-Silva (2009). Both stuies focus on the U.S. economy, where the isability system oes not allow isable iniviuals to combine the benefits with a job, which is very ifferent from the Spanish system. Furthermore, these stuies o not consier interactions with nonisable workers or the role of employers. In our moel, we inclue search intensity an matching frictions because we think that they play a central role in etermining employment outcomes of isable iniviuals. With respect to the empirical evience on the labor market behavior of isable iniviuals, there is an extensive literature analysing the U.S. system (see Autor an Duggan [2006, 2007, 2008]; Autor, Duggan, an Lyle [2011]; Burkhauser an Daly [2011], among others), but the literature is still very limite for the Spanish case (Cervini-Plá, Silva, an Vall-Castello 2012; Malo an Pagan 2012; Marie an Vall- Castello 2012; Vall-Castello 2012). The paper procees as follows. The next two sections escribe the specificities of the Spanish system of isability benefits an compare the Spanish labor market behavior for 5

8 partially isable an nonisable iniviuals from 2001 until The following sections introuce the job search an matching moel an present its calibration. The paper then presents the steay state benchmark simulation an compares it with the ata, an after that presents the sensitivity analysis to the policy parameters. It then looks at the business cycle behavior of the isable workers. Finally, some conclusions are erive in the last section. MAIN FUTURES OF THE SPANISH SYSTEM OF DISABILITY BENEFITS After the perio of the Franco regime, the first comprehensive piece of legislation that was passe in Spain with respect to iniviuals with isabilities was the Ley e Integración Social e los Minusvalios (LISMI), which was approve in The law inclue various aspects ranging from isability benefits as a way to economically protect isable iniviuals to employment promotion measures to promote the labor market integration of isable iniviuals. After this first important step, a number of small changes have been aopte uring the 1990s an 2000s in terms of employment promotion measures to increase the incentive to work an hire isable workers. Economic Incentives for Employers to Hire Disable Workers As shown in Table 1, currently there are three main economic incentives for employers that hire isable workers. First, there is a lump-sum subsiy of 3, euros for each isable worker hire (this amount is ajuste proportionally for part-time contracts). Secon, employers can benefit from euctions to the Social Security contributions. These euctions are linke to the gener an the intensity of the isability of the worker. In general the euctions are 4,500 euros per year, but they increase to 5,350 for women, to 5,700 for women above age 45, to 5,100 for iniviuals with a severe isability, to 5,950 for women with a severe isability, an to 6,300 6

9 for women with a severe isability an age above 45. Finally, the thir element that is inclue in the package of financial incentives for employers to hire isable workers is another subsiy that aims to aapt the working space to the special nees that the isable worker may have. The maximum amount of this subsiy is set at 902 euros an is only pai one time for each contract. Furthermore, a new piece of legislation was passe in 2001 to stress the fact that all these measures woul no longer be applie to certain contracts, such as those to family members, to workers who ha worke in the firm in the past 24 months with a permanent contract, or to workers who ha ene a permanent contract uring the last three months. Table 1 Summary of the Economic Incentives for Employers Lump-sum subsiy Deuctions to the Social Security (one for each contract) contributions 3, (open to part-time contracts; 4,500 euros/year proportional) 5,350 women 5,700 women > 45 5,100 severe isability 5,950 women severe isability 6,300 women severe isability>45 a Other subsiies Ask to INEM for subsiies to aapt the working space b up to a maximum of 902 euros NOTE: a If it is a part-time contract the euctions will be 100 percent if 75 percent of full-time work, 75 percent if she works between 50 an 75 percent, 50 percent if she works between 25 an 50 percent, an 25 percent if she works less than 25 percent. This was change in RD2/2009: the euctions will equal the percentage worke plus 30 percent (with a 100 percent limit). b In 1994 the maximum amount of this subsiy was set at euros. Economic Incentives for Disable Workers The isability system in Spain istinguishes between two types of permanent isability benefits: 1) contributory, which are given to iniviuals who have generally contribute to the Social Security system before the onset of the isabling conition; 2) an noncontributory, which are given to iniviuals who are assesse to be isable but have never contribute to the Social Security system (or o not reach the minimum contributory requirement to access the contributory system). Noncontributory isability benefits are means-teste an manage at the regional level. 4 4 Income is evaluate yearly. The income threshol in 2010 was set at 4, euros/year for an iniviual living alone. This amount is ajuste if the iniviual lives with other members. 7

10 The size of the noncontributory system is relatively small compare to the contributory system (197,126 iniviuals receive noncontributory isability benefits in 2009, while 920,860 receive contributory benefits uring the same year). The amount of benefits receive is also smaller in the noncontributory case (the average noncontributory pension is euros/month compare to an average contributory isability pension of euros/month). Because we want to assess the effect of isability on wages, in the remainer of the paper we focus only on the permanent contributory isability system in Spain. The Spanish Social Security Aministration efines the permanent contributive isability insurance as the economic benefits to compensate the iniviual for losing a certain amount of wages or professional earnings when affecte by a permanent reuction or complete loss of his/her working ability ue to the effects of a pathologic or a traumatic process erive from an illness or an accient. In orer to capture the ifferent situations in which a person can be after suffering from a isabling conition, the Spanish Social Security A In 1994 the maximum amount of this subsiy was set at euros. In 1994 the maximum amount of this subsiy was set at euros. In 1994 the maximum amount of this subsiy was set at euros. In 1994 the maximum amount of this subsiy was set at euros. In 1994 the maximum amount of this subsiy was set at euros. In 1994 the maximum amount of this subsiy was set at euros. Aministration uses a classification of three main egrees of isability that epen on the working capacity lost: 5 5 There is a fourth egree of isability benefits (permanent limite isability), but this type of benefit is alreay extinguishe an it only consists of a one-time lump-sum payment. 8

11 1) Partial isability (57 percent of claimants): the iniviual is impaire to evelop all or the funamental tasks of his/her usual job or professional activity, but he/she is still capable of eveloping a ifferent job or professional activity. 2) Total isability (40 percent of claimants): the iniviual is impaire for the evelopment of any kin of job or professional activity. 3) Severe isability (3 percent of claimants): iniviuals who, as a result of anatomic or functional losses, nee the assistance of a thir person to evelop essential activities of aily living, such as eating or moving. The eligibility requirements an the pension amount epen on the source of the isability (orinary illness, work-relate or unrelate accient, or occupational illness); the level of the isability; an the age of the onset of the isabling conition. Table 2 summarizes the main parameters of both the eligibility criteria an the pension formula. With respect to eligibility, the number of years of contributions require epens on the age of the onset of the isable conition for common illness, while there are no contributory requirements if the health impairment is ue to either an accient or an occupational illness. The total amount of the pension is obtaine by multiplying a percentage, which varies epening on the type of pension an the egree of isability (as shown in the last rows of Table 2) to the regulatory base, which epens on the source of the isability an on previous salaries. 6 The percentage is 55 percent or 75 percent for partial isability beneficiaries, 100 percent for total isability an 150 percent for severe isability. Partial isability beneficiaries receive 55 percent of the regulatory base but this percentage can be increase to 75 percent for iniviuals who are oler than 55 an have ifficulties fining a job ue to lack of eucation or characteristics of the social an labor market of the region where they live. 6 Benefit = Regulatory base Percentage. 9

12 The number of years inclue in the regulatory base epens on the source of the isability; for common illness the regulatory base is calculate by iviing by 112 the wage in the last 96 months (8 years) before becoming isable. When the source of the isability is an accient unrelate to work, the regulatory base is calculate by iviing by 28 the wage in the last 24 months before becoming isable. The iniviual can choose these 24 months from the last 7 years of work. For a work-relate accient or professional illness, the regulatory base is calculate by iviing by 12 the wage in the last 365 ays before becoming isable. 7 With respect to the taxes that isable iniviuals have to pay, we can see in Table 3 that iniviuals with a total isability benefit are exempt from paying income taxes. On the contrary, partially isable workers are require to pay income taxes (they are only exempt if the iniviual resies in the Basc country an the partially isable worker oes not have a job). However, there is a reuction in the employment income use to calculate the income tax for partially isable workers. This reuction is 2,800 euros per year if the isability level is between 33 an 65 percent an 6,200 euros per year if the isability level is more than 65 percent (or if the iniviual has reuce mobility). Because the aim of the paper is to analyze the incentives an isincentives to work provie by the isability system in Spain, we focus only on the group of partially isable iniviuals, as they have been assesse to keep certain capacity to work by the Social Security meical team (in a professional activity that is ifferent from the one evelope before the onset of the isability). We o not inclue in our sample iniviuals with a total or a severe isability 7 There was a reform in the calculation of the level of isability benefits for orinary illness introuce in After the reform, there was a percentage that epene on the number of years contribute to the system that was multiplie by the regulatory base. We o not moel this legislative change in this paper, an we inclue the prereform system in our moel. However, to see the effect of this change on the inflow into the isability system, see Silva an Vall-Castello (2012b). 10

13 because they are suppose to be impaire for the evelopment of any kin of job or professional activity. Table 2 Summary of the Parameters to Calculate Permanent Disability Pensions Accient unrelate to Orinary illness work Eligibility Regulatory base Percentage applie to the regulatory base Age >= 31: Contribute 1/4 time between 20 years ol an isabling conition. Minimum of 5 years. Age < 30: Contribute 1/3 time between 16 years ol an isabling conition. No minimum number of years require. Average wage last 8 years of work. No minimum contributory perio require. Average annual wage of 24 months within the last 7 years of work. Work-relate accient or professional illness No minimum contributory perio require. Average wage last year of work. Partial isability: 55 percent. Iniviuals oler than 55 with ifficulties to fin a job ue to lack of eucation or characteristics of the social an labor market of the region where they live: 75 percent. Total isability: 100 percent. Severe isability: 100 percent + 50 percent. Table 3 Summary of the Fiscal Measures for Iniviuals Receiving Disability Benefits Partially isable iniviuals Totally isable iniviuals Pay normal income taxes. Exempt from income taxes. Income taxes Exempt if Basc a country an no job. Reuction in employment income use to calculate the income tax 2,800 euros/year (if isability level between 33 percent an 65 percent) if working. 6,200 euros (if isability more than 65 percent or below that but isable with reuce mobility) if working. NOTE: a Disable iniviuals in the provinces of Vizcaya, Alava, an Guipuzcoa, which constitute the Basc country, are exempt from paying income taxes on total isability pensions if they on t work. DATABASE The stuy will use the Continuous Sample of Working Lives (Muestra Continua e Vias Laborales, [MCVL]), which is a microeconomic ata set base on aministrative recors provie by the Spanish Social Security Aministration. It contains a ranom sample of 4 percent of all the iniviuals who, at some point uring 2010, ha contribute to the Social 11

14 Security system (either by working or being on an unemployment scheme) or ha receive a contributory benefit. The ranom sample selecte contains over 1 million people. There is information available on the entire employment an pension history of the workers, incluing the exact uration of employment, unemployment, an isability pension spells, an for each spell, several variables that escribe the characteristics of the job or the unemployment/isability benefits. There is also some information on personal characteristics such as age, gener, nationality, an level of eucation. The macroeconomic variables use to capture the economic business cycle are erive from the Spanish Instituto Nacional e Estaistica. For the sample of isable workers we select an inflow sample of all iniviuals that starte receiving partial isability benefits between 2001 an 2010, an we follow their labor market transitions until 2010 or until they reach age 65, an they are automatically transferre to the ol-age pension system. For the sample of nonisable iniviuals we select a 30 percent ranom sample of all iniviuals that have never receive (an never will receive) a isability pension, an we follow their transitions in the labor market from 2001 to For both samples, we consier an iniviual as employe if they are observe as working on December 15th. 8 EMPLOYMENT AND TRANSITION RATES IN THE SPANISH LABOR MARKET FOR DISABLED AND NONDISABLED WORKERS In this section we present a set of inicators that compare the Spanish labor market behavior for partially isable () an nonisable (n) iniviuals from 2001 until Figure 8 We have also worke with a ifferent efinition of employment an the results o not change substantially. We have ecie to use the December 15th efinition because it prove to be the efinition with a lower irregular component. 12

15 1 shows the annual employment rates of these two types of iniviuals, as a proportion of the working-age population. The first noteworthy result is the presence of an important employment gap between nonisable an partially isable workers uring the whole perio. More in etail, while the employment rate of nonisable employees fluctuates between 78.2 percent an 82.3 percent, the corresponing rate of partially isable employees moves between 8.1 percent an 13.3 percent. Both employment rates show a somewhat similar tren, increasing between 2001 an 2007 an ecreasing uring the current ownturn. Figure 1 Spanish Employment Rates for Disable an Nonisable Workers ( ) Nonisable Disable Disable Nonisable 7.0 SOURCE: Authors elaboration from the MCVL. We also analyze the ins an outs of employment by consiering measures of separation an job fining rates erive from the MCVL. Let eu t i an ue t i enote the gross flows from employment to nonemployment an from nonemployment to employment with i=,n, i i respectively, an let e t 1 an u t 1 inicate the measure stocks of employe an nonemploye workers in year t 1, respectively. Then, the annual separation an job fining rates are 13

16 etermine by ψ i t = eu t i i an f i t = ue t i. Figures 2 an 3 show the evolution of the job fining an e t 1 u i t 1 job separation rates for both isable an nonisable workers. As can be observe in Figure 2, the job fining rate for the nonisable is much higher than that for isable people. On average, the job fining rate for the former is 10 times higher (0.329 an 0.033, respectively). In aition, both rates isplay similar behavior uring the perio, showing a correlation coefficient of In turn, Figure 3 shows that the job separation rate of isable workers is 2.5 times higher than that for nonisable workers (0.215 vs , on average). Moreover, both rates are positively correlate, isplaying a correlation coefficient of Figure 2 Job Fining Rates for Disable an Nonisable Workers ( ) Nonisable Disable Di sa bl e Noni s abl e 2. SOURCE: Authors elaboration from the MCVL. Summarizing, we fin that the job fining rate for isable workers is much lower than for nonisable workers, while the job separation rate is higher. These results imply that nonemployment spells for the isable are much larger while, at the same time, these workers lose or separate from their jobs with higher frequency. 14

17 Figure 3 Job Separation Rates for Disable an Nonisable Workers ( ) Nonisable Disable Dis able N onis ab le 18.0 SOURCE: Authors elaboration from the MCVL. As we have seen above, the Spanish system allows partially isable iniviuals to work while receiving isability benefits. Accoring to the law, however, partially isable iniviuals cannot work in the same job position when becoming isable, although they are still capable of oing a ifferent activity. Therefore, once the iniviual becomes isable, either the firm can reallocate these workers in a ifferent job position insie the firm or the new isable iniviual has to change his current job an search for a job in another firm. Accoring to Figure 4, less than 20 percent of the nonisable employees who become partially isable continue working in the same firm just after the isabling conition (17.3 percent, in average). This survival rate, ρ e,t, is etermine by the ratio between the new isable iniviuals who remain employe in the same firm, e n t 1 e t, an the total flow of new isable iniviuals, e t 1 e t + e t 1 u t. Therefore, using these transition rates that we have empirically obtaine from the Spanish Social Security atabase, in the next sections we present an simulate a moel with matching n n 15

18 Figure 4 Survival Employment Rate for New Disable Employees ( ) SOURCE: Authors elaboration from the MCVL. frictions an search intensity that inclues the main aspects of the Spanish Disability System in orer to explain the labor market ifferences between isable an nonisable workers. THE THEORETICAL MODEL The economy consists of a continuum of risk-neutral, infinitely live firms, an iniviuals who iscount future payoffs at a common rate, β. We normalize the population to 1. Moreover, capital markets are perfect an time is iscrete. Iniviuals may be either nonisable (n) or partially isable (). A partially isable iniviual receives a pension equivalent to a proportion, α, of their average wage for the years previous to the accient or illness, w 0. A nonisable iniviual can be converte into partially isable with exogenous probability, π(the health shock). In turn, a partially isable iniviual exits from the labor market with exogenous probability, ρ. 16

19 Accoring to the Spanish isability system, an iniviual with partial isability cannot evelop all of the funamental tasks of his usual job, but he is still capable of eveloping a ifferent job or professional activity. We assume that the firm can ecie to offer him a reallocation in a ifferent professional activity. If there is not reallocation insie the firm, the new isable iniviual has to search for a new one. Both partially isable an nonisable iniviuals can either be employe or unemploye. All workers compete in the labor market for the same jobs. Unemploye iniviuals enjoy an instantaneous utility b each perio. This employment opportunity cost has to be given up when the worker fins a job. Following Pissaries (2000), unemploye workers have search intensity. Let s j t with j = n, be a variable measuring the intensity of search by each type of unemploye workers. Unemploye workers incur in convex job search costs, b s j t π, expresse in terms of the employment opportunity costs, b. It is more ifficult for a person with a partial isability to look for jobs than it is for a person without one. Thus, we assume that the search costs for isable iniviuals are proportional to the level of isability, is. Each perio, any job position may be enogenously terminate. Exogenous separations may also occur with probability φ for any type of worker. When an employment relationship is broken, the worker becomes unemploye. Each firm consists of one job that is either fille or vacant an uses only labor as input. Before a position is fille, the firm has to open a job vacancy with cost c per perio. A firm s output epens on aggregate prouctivity A t, a match-specific term z t, an the worker s type. In particular, a job fille with nonisable workers prouces A t z t, whereas with a partial isability in continuing jobs prouces (1 is)a t z t, where (1 is) is the permanent prouctivity 17

20 gap between isable an nonisable iniviuals. 9 Following Cervini-Plá, Silva, an Vall- Castello (2012), we also assume that the prouctivity gap in new job positions inclues a temporary component, is t, which is relate to the assimilation costs of working in a ifferent job or professional activity. This temporary gap component isappears after a certain perio spent working in the new job. The match-specific prouctivity term z t is assume to be inepenent an ientically istribute across firms an time, with a cumulative istribution function G(z) an support [0, z ]. We also assume that log(a t ) follows a Markovian stochastic process, while the iiosyncratic prouctivity z is assume to be log-normally istribute with normalize mean (μ) an stanar eviation (σ). Employers receive a one-time lump-sum subsiy ζ e when they maintain a new isable iniviual in the firm or ζ u if they hire an unemploye isable worker. Firms also receive an annual euction of ξ to Social Security contributions for each isable worker. In turn, firms incur extra hiring costs when they hire a new isable worker. For example, they nee to provie aapting working space costs so that workers with isabilities can o their jobs, an they have to take into account the extra time that supervisors or coworkers will nee to spen to assist workers with isabilities. We assume that these net entry costs, τ, are incurre uring the first year of the employment relationship. Finally, employe iniviuals earn a wage net of taxes, w j t, an isable workers receive a net income tax euction, p, when they are working. There is a time-consuming an costly process of matching unemploye workers an job vacancies. As in en Haan, Ramey, an Watson (2000), we assume that the meeting function takes the following form: 9 As we mentione in the beginning of the paper, the presence of a prouctivity gap in our moel is consistent with the recent empirical evience presente by Malo an Pagan (2012). 18

21 (1) su t tν t m( su, ν ) =, t t t ϑ ϑ ( su ) 1/ t t + ν t ϑ where u t enotes the unemployment rate, v t are vacancies, an s t efines the average job searching intensity. Each perio an unemploye iniviual meets a firm with probability Since λ = s j t j t (, ν ) m su t t t su with probability t t. there is not search intensity for firms, they meet an unemploye worker (, ν ) m su qt = ν t t t t. We assume that there is free entry for firms. Hence firms open vacancies until the expecte value of oing so becomes zero. Therefore, in equilibrium the value of a vacancy, V t, is equal to zero: (2) V t = 0. To escribe the firms behavior, let us efine the Bellman equations characterizing the value of the fille position for a nonisable worker, J t n (z t ), the new job position for isable worker remaining in the same firm, J e,t (z t ), the new isable position for a worker coming from the unemployment status, J u,t (z t ), an the position for a isable employee in a continuing job, J t (z t ). { z, + 1, 1 } t+ 1 z + et, + 1 (3) J z z n( z n n ) = A z w ( z ) + ( ϕ) β E ( π) J ( z ) G ( z ) + π J ( z ) G ( z ) 1 1 n t t t t t t t t et z (4) J ( z ) ( 1 is is ) A z w ( z ) ζ ξ τ ( 1 ϕ)( 1 ρ) βe J ( z) G ( z) t = ut, t t t t ut, t u t z t { z + } + 1 z (5) J ( z ) = ( 1 is is ) A z w ( z ) + ζ + ξ τ + ( 1 ϕ)( 1 ρ) βe J ( z) G ( z) t et, t t t t et, t e t t 1 19

22 z (6) J ( z ) ( 1 is) A z w ( z ) ξ ( 1 ϕ)( 1 ρ) βe J ( z) G ( z) t = + + t t t t t t t z t where E t is the expectation operators an z n, z u, z e an z are prouctivity threshols efine such that nonprofitable matches (i.e., with negative surplus) are severe. Accoring to Equation (3), each perio, those nonisable employees who receive a health shock, π, have the possibility of being reallocate to a ifferent job position insie the firm. If the firm reallocates the worker it receives the lump-sum subsiy, ζ e (see Equation [5]). In case of isagreement, the firm opens a new vacancy that may be fille by either a isable or a nonisable worker. If an unemploye isable worker fills the new vacancy, the firms will receive a lump-sum subsiy, ζ u (Equation [4]). In both cases, the firm only gets the subsiy uring the first perio of the new employment relationship an incurs ajustment costs, τ. As Equation (6) shows, continuing isable job positions o not receive the lump-sum subsiy but still receive the euction ξ to Social Security contributions. New hires are etermine accoring to the expecte value of a contact with an unemploye worker. This value is the average of the expecte hiring values of isable an nonisable unemploye workers. Thus, it epens on the effective searching share of isable an nonisable iniviuals looking for jobs, δ j t = s j n t ut. The average expecte value of a new s t u t fille position is equal to (7) c z n = βe q t δ n t n J t+1 (z) G(z) + δ t J u,t+1 (z)g(z) t z t+1 z z u,t+1 20

23 Accoring to Equation (7), in equilibrium the average expecte value of a new fille position is equal to the hiring costs, c q t. The conitions efining the threshols for job creation an estruction are (8) J t n (z tn ) = 0, (9) J e,t z e,t = 0, (10) J u,t z u,t = 0, (11) J t z t = 0. The first expression, (8), captures both the job creation an job estruction conition for a nonisable iniviual. Expression (9) is the job reallocation conition for the isable worker who remains in the same firm. In turn, expression (10) is the job creation conition for a isable worker coming from the unemployment status, u. Finally, expression (11) is the job separation conition for each isable employee in a continuing job position. From the worker s perspective, the values of being unemploye U j t, employe an W j t (z t ) are z n n n n λ t W n t 1 z G z G z n n z t 1 U t 1 t 1 Ut b b st π βe = + t n n ( 1 λt ) Ut+ 1 πβ EU + + t t+ 1 ω (12) ( ) ( 1 ) ( ) ( ) + ( ) z λ t W u ut, 1 z G z G z z ut, 1 U t 1 t 1 Ut b b is st uw0 ρ βe = t ( 1 λt ) U + t+ 1 ϖ (13) ( 1 )( ) α ( 1 ) ( ) ( ) + ( ) 21

24 (14) 1 π β { 1 ϕ ( n z ) ϕ + 1 t } + 1 z (, 1 ( ) ( ) ) (, 1) z et, + 1 z ( ) ( ) ( ) ( ) ( ) ( ) ( ) W z = w z + E W z G z + G z U + U n n n n n n t t t t t t t t t + πβ E W z G z + G z U t et et t (15) ( ) α ( ) ( 1 ) ( Wt 1 z G z G zt 1 U + + t+ 1) W zt 1 ut, zt = w0 + wut, zt + p+ ρ βe + t + ϕu t+ 1 (16) ( ) α ( ) ( 1 ) z ( 1 ϕ ) ( ) ( ) + ( ) ( Wt 1 z G z G zt 1 U + + t+ 1) W zt 1 et, zt = w0 + wet, zt + p+ ρ βe + t + ϕu t+ 1 (17) ( ) α ( ) z ( 1 ϕ ) ( ) ( ) + ( ) ( Wt 1 z G z G zt 1 U + + t+ 1) W z = ( w0 + w z ) + p+ ( ) E z ( 1 ϕ ) ( ) ( ) + ( ) 1 z 1 t t t t ρ β t. t + + ϕu t+ 1 When a health shock hits a nonisable worker, he receives isability benefits, α j w 0, regarless of the labor market status. However, accoring to Equations (13) an (15) (17), the isability scheme for an unemploye worker, α u, can be ifferent with respect to the one for the employment status, α. More in etail, accoring to the Spanish Disability System, the amount of pension receive varies accoring to the isability egree in which the iniviual is classifie. Iniviuals in the partial isability scheme receive, in general, 55 percent of the regulatory base (which is an average of the last salaries). This 55 percent can be increase to 75 percent for iniviuals age 55 conitionally of being unemploye. Moreover, Equations (15) (17) show that if the partially isable iniviual works, he receives a net reuction of p in the employment income use to calculate the income tax. Unemploye nonisable an isable workers fin jobs with the following probabilities: 22

25 (18) f n (1 ) n 1 ( n t = πλ t G z t+ 1) (19) f ( 1 ) 1 ( t = ρ λ t G z ut, + 1) Moreover, it follows that nonisable an isable workers separate from employment to unemployment with probabilities n n (20) ψt = ( 1 π) ϕ+ ( 1 ϕ ) G( z t+ 1) (20) ψt = ( 1 ρ) ϕ+ ( 1 ϕ ) G( z t+ 1) In turn, the reallocation an nonreallocation rates for new isable employees are (22) χet, = ( 1 ρ) π 1 G( z et, + 1) (23) χut, = ( 1 ρ) π G( z et, + 1) Notice that the survival rate insie the firm for new isable workers is just ρ e,t = 1 G z e,t+1. Neither workers nor employers can instantaneously fin an alternative match partner in the labor market, an because hiring an firing ecisions are costly, a match surplus exists. To ivie this surplus we assume wages to be the result of bilateral Nash bargaining between workers an firms. They are revise every perio upon the occurrence of new shocks, an the Nash solution is the wage that maximizes the weighte prouct of the workers an the firms net 23

26 return from the job matches. The first-orer conitions for the isable an nonisable employees yiel the following four equations: n n n (24) ( 1 η) W ( z ) U = ηj ( z ) t t t t t (25) ( 1 η) W ( z ) U = ηj ( z ) ut, t t ut, t (26) ( 1 η) W ( z ) U = ηj ( z ) et, t t et, t (27) ( 1 η) W ( z ) U = ηj ( z ) t t t t t where η (0,1) enotes workers bargaining power relative to firms. Finally, each type of unemploye worker chooses search intensity s j t to maximize the present-iscounte value of their expecte income U j t uring the search process, taking the other market variables as given. Each optimal s j t satisfies (28) ( ) (, ) ( ) ( ) m su v 1 π β ( ) ( ) 1 ( ) = ϖ su t t t z n n n n Et W n t 1 z G z G z z t 1 Ut 1 b s t t+ 1 t t ϖ 1 (29) ( ) (, ) ( ) ( )( ) m su v 1 ρ β ( ) ( ) 1 ( ) = ϖ 1+ su t t t z Et W u ut, 1 z G z G z z ut, 1 Ut 1 b is s t t+ 1 t t ϖ 1 To fully characterize the ynamics of this economy, we nee to efine the law of motion for unemployment an employe workers (u t j an e t j ). These evolve accoring to the following ifference equations: (30) e n n t = e t 1 + f n n t u t 1 ψ n n t e t 1 χ n e,t e t 1 χ n u,t e t 1 24

27 (31) e t = e t 1 + f t u t 1 ψ t e t 1 + χ n e,t e t 1 ρe t 1 (32) u n n t = u t 1 f n n t u t 1 + ψ n n t e t 1 n πu t 1 (33) u t = u t 1 f t u t 1 + ψ t e t 1 + πu n t 1 +χ n u,t e t 1 ρu t 1 (34) e t n + e t + u t n + u t = 1 (35) u t n + u t = u t (36) e t n + e t = e t (36) CALIBRATION We calibrate the moel in the steay state at annual frequency to be consistent with certain empirical Spanish labor market facts. In particular, the parameterization must match the main labor market characteristics of nonisable iniviuals between 2001 an 2011: 1) A job separation rate of 9.2 percent, so ψ n = ) A job fining rate of 28.4 percent, f n = ) the 93.2 percent average proportion of nonisable workers in the labor force, e n +u n u+e = ) A relative number of nonisable workers looking for jobs of un u 5) Following Silva an Vázquez-Grenno (2011), the hiring costs parameter is calibrate to match a quarterly hiring costs equivalent to 3 percent of nonisable wages, c 0.12 =. wn 4 6) Cervini-Plá, Silva, an Vall-Castello (2012) estimate an average wage gap of 21 percent between a isable an nonisable iniviual, wn = w 25

28 7) Base on Castillo, Jimeno, an Licanro (1998), the calibration also must match our target elasticity, ε u, of 0.80 in the matching function with respect to unemployment. We normalize the aggregate labor prouctivity to one, A = 1. We fix the iscount factor at β = 0.96, which implies a reasonable interest rate of nearly 4 percent. We also set the bargaining parameter η to 0.5. Using the efinition of partial isability, we assume that partial isability represents a reuction of between 33 percent an 65 percent of the working ability an set the permanent isability gap at is = From the Spanish atabase, the transition rate from nonisability to partial isability is set at 0.14 percent, π = In turn, Cervini-Plá, Silva, an Vall-Castello (2012) calculate that the transitory component of the wage gap represents 6 percent of the wage of a isable worker. Thus, we set the transitory isability gap, is t, to match this target, w w u = Iniviuals above 55 years of age in the partial isability scheme receive 75 percent of the regulatory base if they o not work an 55 percent if they work (which is an average of the last eight years of salaries). As in Silva an Vall-Castello (2012a), we assume that iniviuals are below 55 years ol. Thus, we set α u = α = The average wage net of tax for a partially isable worker is 14,168 euros/year. 10 If the iniviual works, he receives an income euction net of taxes of 420 euros/year in the employment income if the isability level is between 33 percent an 65 percent. 11 This amount represents aroun 3 percent of the average wage for isable workers. In turn, the average regulatory base is 10,716 euros/year, which represents 0.76 times the average net wage of a partially isable worker. Thus, we calibrate w 0 an p to match these two targets. 10 Accoring to Cervini-Plá, Silva, an Vall-Castello (2012), the gross average wage is 16,668. Using the OECD Tax Database, we calculate an average income tax of 15 percent. 11 Multiplying the income tax euction of 2,800 by the tax rate of 0.15 we obtain 420 euros/year. 26

29 In turn, firms receive a lump-sum subsiy of 3,900 euros when hiring an unemploye isable iniviual, which represents aroun 28 percent of the average annual net wage of a isable worker. Thus, we set ζ u to match this target. Since 2001, this subsiy cannot be applie to workers that ha worke in the firm in the past 24 months or to workers that ha ene a contract uring the last 3 months. We assume that ζ e = 0. Firms also receive 4,500 euros/year, which accounts for reuction to the Social Security contributions. This magnitue represents 32 percent of the net wage for a isable iniviual. As before, we set ξ to match this target. The logarithm of the iiosyncratic prouctivity z is assume to be N (µ=0, σ=0.2). Finally, the parameter of the searching costs, π, the matching function parameter, θ, the exogenous separation probability, φ, the transitory isability gap, is t, the employment opportunity cost, b, the vacancy costs parameter, c, income euction, p, the exit rate from the labor market for isable iniviuals, ρ, the lump-sum subsiy for hiring a isable worker ζ u, the firm s euction to the Social Security contributions, ξ, the firm s ajustment costs, τ,an the regulatory base parameter, w 0, are calibrate to match our 12 targets. Table 4 resumes the targets an the calibrate parameters in the economy. Table 4 Annual Calibrate Parameters for the Average Spanish Labor Market, Parameters Value Definition A 1 Normalization of aggregate labor prouctivity β 0.96 Discount factor σ w 0.20 Stanar eviation for the istribution of log(a) µ 0.00 Mean of the istribution of log(a) b The employment opportunity cost θ Matching function elasticity φ Exogenous job exit probability. ρ Exit rate from the labor market for isable iniviuals c Vacancy costs p The income euction for isable workers ζ u 0.22 The firm s lump-sum subsiy for hiring an unemploye worker with isability 27

30 Table 4 (Continue) Parameters Value Definition ζ e 0.0 The firm s lump-sum subsiy for hiring an employe worker with isability α u 0.55 Partial isability scheme for unemployment α 0.55 Partial isability scheme for employment w Regulatory base for isability is 0.50 Permanant egree of isability is t Transitory egree of isability π 3.7 Searching costs η Workers bargaining power π Transition from nonisability to partial isability ξ 0.25 Firm s euction to Social Security contribution τ Creation costs for new isable positions Targets f n Nonisable job fining rate ψ t Nonisable job separation rate u n The relative number of nonisable unemploye workers u e n + u n Average proportion of nonisable workers in the labor force u + e c w n 0.03 Vacancy costs ratio w n 1.21 Average wage gap between isable an nonisable workers w p Worker s subsiy ratio for isable iniviuals w w Regulatory base ratio for isable iniviuals w ζ u 0.28 The firms lump-sum subsiy ratio w ξ 0.32 Firm s ratio of euction to Social Security contribution w ε u 0.81 Elasticity of the matching function w w u 1.06 Transitory isability gap STEADY STATE SIMULATED RESULTS Table 5 shows the benchmark simulation of the moel in the steay state an compares it with average values observe in the Spanish labor market between 2001 an The simulation shows an employment rate of 24.0 percent among isable workers, which is much lower than the observe one for nonisable iniviuals (79.9 percent). It also shows that the job fining rate for isable workers (0.101) is significantly lower than the observe one for 28

31 nonisable iniviuals (0.329). The lower average job fining rate of isable workers is ue to their search intensity rate of 0.167, which is much lower than the job search intensity rate of nonisable workers (0.465). The simulation also shows that 87.0 percent of the times the firm an the isable unemploye worker meet, a new employe relationship is create [G z u,t+1 = 0.130]. In other wors, the moel suggests that although firms are willing to hire isable unemploye iniviuals, these workers look for jobs with much less intensity than nonisable iniviuals. While part of this behavior is relate to the presence of both a prouctivity gap an higher job searching costs that iscourage these types of workers from looking for jobs, our moel also inicates that the receipt of the isability benefits increase the employment opportunity cost an therefore reuce their job search intensity. With respect to the firms sie, although the lump-sum subsiy generates the incentive to hire new isable workers this incentive is reuce once the subsiy isappears in the next perio. More in etail, Table 5 shows that the simulate job estruction rate for continuing job isability contracts, ψ, amounts to 0.311, which is higher than the job estruction rate of nonisable employees (0.083). Similarly, the moel shows that less than 10 percent of the employees who become isable are kept in the same firm after the isabling conition. That is, the employment survival rate for new isable employees is ρ e,t = As explaine above, this low survival rate can be explaine by the fact that firms o not receive a lump-sum subsiy if they ecie to keep the new isable employee working in the firm. Finally, although the moel can help to explain the ifferences between isable an nonisable workers, notice that it overestimates most of the average transition rates for the 29