Overeducation in Cyprus

Transcription

1 Overeducaton n Cyprus Andr Kyrz The Unversty of Lecester Department of Economcs Abstract In ths paper we nvestgate the effects of excessve and lmted schoolng on the ndvduals earnngs n Cyprus. Ths mportant ssue has not yet been tested analytcally for Cyprus due to a dearth of ndvdual level data. Usng the conceptual framework of over- under- and adequate educaton we fnd that overeducated ndvduals stll receve a postve return for ther surplus schoolng whereas the undereducated receve a negatve return on ther earnngs for the defct years of schoolng. Addtonally, usng a logt model we tested whether beng overeducated ncreases the probablty of beng unemployed under perods of economc crss. We fnd statstcally sgnfcant evdence that ndvduals wth surplus schoolng are more lkely to lose ther jobs whle undereducated ndvduals are less lkely to become unemployed. Ths mght be due to the lower reservaton wage of the undereducated relatve to ther overeducated counterparts. Consequently employers do not fnd t proftable to employ overeducated ndvduals durng hard tme such as the current global downturn. Introducton It has been, so far, well establshed emprcally and theoretcally that the more years of educaton one has the hgher the earnngs. Followng the Human Captal Theory, each extra year of educaton should be accompaned wth a respectve ncrease n earnngs. Educaton has been apprecated by many as a form of nvestment 1 and more and more ndvduals have been observed to enter hgher educaton nsttutons because of expectatons for hgher returns. Almost every advanced country n the world has been experencng an ncrease n the supply of hgher educaton graduates, rasng a lot of concerns as to what the results would be of such an effect. Ths paper nvestgates the prevalence of overeducton n Cyprus labour market, somethng that has not yet been tested. What s nterestng about Cyprus s that, despte ts small sze, over the past decade there has been a very hgh amount of hgh school graduates enterng tertary educaton. Menon (1998) has argued that 60% of young people have been plannng to enter hgher educaton and ths has caused major concerns whether the lmted 1 We acknowledge that the Human Captal s effects are debatable by the Sgnallng Theory whch vews educaton as a form of screenng devse for employers to dstngush among hgh and low ablty ndvduals.

2 capacty of the Cyprot economy can support ths ncrease n supply. After the entrance of Cyprus n the European Unon, n 2004, the percentage of people enterng hgher educaton ncreased from 66% to 75% n just one school year. Addtonally, the ncrease n the supply of students n the tertary educaton was further enhanced by the reform of the educatonal system n Cyprus after several colleges ganed recogntons and became prvate unverstes that reached the 82% n Addtonally, the latest Statstcs of Educaton report (2008/2009) 2 showed that 79% of the total secondary educaton graduates contnued ther studes beyond the secondary level. Although ths may rase some concerns about the qualty of the graduates n Cyprus, t does not change the fact that more and more ndvduals hold a graduate certfcate. Furthermore, ths paper tres to move the research and lterature one step further by lookng whether overeducated ndvduals are more lkely to beng unemployed. The dea behnd ths hypothess s that overeducated ndvduals receve a hgher wage than ther adequate educated co-workers. As a result, employers under certan crcumstances may not fnd t proftable to employ overeducated ndvduals and ths wll result n an ncrease n unemployment. Cyprus s one of the most approprate countres to test ths snce the rate of ndvduals enterng the tertary educaton s extremely hgh n the last years. As Decreuse (2000) n one of the two man arguments n hs paper states that ndvduals may not protect themselves from unemployment by acqurng more years of educaton f all ndvduals are hghly educated. Ths paper contnues as follows. Secton 2 descrbes Prevous research followed by the theory behnd the methodology adopted n Secton 3. Secton 4 descrbes the data used and Secton 5 s dvded nto two sub-sectons dscussng the results for the effect overeducaton has on earnngs and whether overeducated ndvduals are more lkely to lose ther jobs and become unemployed n a perod of crss. Fnally, Secton 6 concludes. Prevous research on the effect of overeducaton on wages Much of the lterature has dealt wth the real cost of overeducaton on ndvduals, that s the wage penalty. Followng the ncreasng supply of graduates almost n every developed economy n the last 30 years, concerns were rased as to the effect ths would have. Duncan and Hoffman (1981) were the frst to test for the effect of overeducaton and concluded that overeducated ndvduals receved a hgher wage than ther adequately educated co-workers but lower than ndvduals holdng the same amount of educaton n well matched jobs. Addtonally, undereducated receved less returns than ther adequately co-workers but more compared to ndvduals holdng the same educaton n well matched jobs. Although dfferent methodologes were engaged n some of the emprcal works, results have been consstently the same (Rumberger 1987; Scherman, 1991). These results were also supported for dfferent countres. Hartog and Oosterbeek (1988) have confrmed the above results for Netherlands and Cohn and Ng (2000) for Hong Kong. On Portugal, Kker et al (1996) argue that research on the mpact of schoolng on earnng should not be done usng the Mnceran regresson as t s msleadng. Addtonally, n another paper (Mendes et al, 2000) they show the mportance of ncludng nteracton terms between over- under- and adequate educaton wth experence and tenure. 2 Statstcal Servce of Cyprus, Educatonal Statstcs, Seres I, Report No.41

3 Research on the UK has confrmed that overeducated ndvduals receve less than ndvduals wth smlar levels of schoolng but workng n matched jobs (Dolton and Vgnoles, 2000). Addtonally, t was found that ndvduals wth surplus schoolng experence a hgher wage penalty n ther frst job than n ther last (Dolton and Slles, 2008). Frenette (2004) found that overeducated Canadan college and bachelor s graduates experence an earnngs penalty whereas, although master graduates are more lkely n beng overqualfed, they experence lttle earnngs effect. Theory and Methodology There s no sold theory behnd the overeducaton model presented here. However, ths model derves from two mportant theores n labour economcs; the human captal model (1) and the job competton model (2). The supply sde of the labour market, Human Captal model, assumes that wages are determned solely from the personal characterstcs of ndvduals lke the years of educaton, experence, tranng etc. It s assumed that workers productvty s fully utlzed and the wage receved reflects the margnal productvty of the ndvdual. Human Captal Model (Mncer, 1974): ln Y S E E (1) where S are the ndvdual s years of schoolng 3 2 The demand sde, Job competton theory, suggests that t s not the personal characterstcs, lke years of schoolng of the ndvdual or experence, that determne wages but the jobs characterstcs, lke the requred level of schoolng. Job Competton Model(Thurow,1975) R 0 1S ln Y (2) where R S s the amount of schoolng Requred by the job. Fnally, Duncan and Hoffman (1981) by ncludng the requred, the surplus and defct level of educaton trggered the lterature on overeducaton. The attaned level of educaton s separated nto A=R+S-D. If an ndvdual has surplus (S) educaton the defct (D) varable takes a value of zero and when S s added to R then the acqured level of educaton s obtaned. Overeducaton Model /Assgnment Theory (Duncan and Hoffman, 1981) ln Y E (3) R O U 0 1S 2S 3S 4 The above equaton has become a trademark n the lterature of over- under and adequate educaton. They have ntellgently ncorporated both the supply and demand sdes of the labour market by ncludng job characterstcs and personal characterstcs. lny s the natural logarthm of hourly earnngs, ndvdual s surplus schoolng, S s the amount of schoolng Requred by the job, R U S O s the S the defct schoolng and E are the years of experence.

4 The human captal theory assumes that the coeffcents of requred, over and under-educaton ( 1, 2 and 3 ) should equal each other. The job competton model assumes that only requred schoolng determnes wages. Both theores are rejected usng the assgnment model. From the results of prevous work t s expected that and that although 1 2 the coeffcent on surplus schoolng s stll postve (Dolton and Vgnoles, 2000; Hartog and Oosterbeek, 1988). Data The data used n ths paper s the European Unon Statstcs on Income and Lvng Condtons (EU SILC) survey performed by Eurostat. It s a cross-sectonal and longtudnal mcro data collectng nformaton on ncome, poverty, lvng condtons, housng condtons, socal excluson, educatonal levels, health status, and employment status on a household and on an ndvdual level. Focusng on Cyprus, the survey was conducted by CYSTAT, the Statstcal Servce of the Republc of Cyprus, and by the tme ths chapter was wrtten only three waves, years 2005, 2006 and 2007, were avalable. We were not able to derve a panel snce the personal dentfcaton number was not the same through the years, and as a result only a cross-secton form and pooled data could be used. The varables used are the logarthmc of earnngs, the years of schoolng, the requred years of schoolng, actual experence, experence squared, gender, martal status, health status and whether the job s permanent or not. Occupaton and ndustry dummes were also used as control varables. Results The results are shown n the table below wth the Mnceran regresson n the frst column, the ORU regresson n column (2) and column (3) wth the ncluson of nteractve dummes for experence. Table 7: Full Sample Results for Schoolng and ORU model Varables lnearnngs (1) Schoolng lnearnngs (2) lnearnngs ReqSch (3) (50.13)** (29.33)** OverE (6.11)** (7.99)** UnderE (12.50)** (9.59)** Experence (28.15)** (30.53)** (22.77)**

5 Exp (19.18)** (21.45)** (22.46)** Gender (32.71)** (34.92)** (35.15)** Marred (9.95)** (9.15)** (8.71)** Healthstat (5.98)** (4.13)** (4.20)** Permanent (37.47)** (31.82)** (32.05)** OverExp (5.86)** UnderExp (3.60)** ReqExp (2.69)** _cons (238.99)** (227.34)** (141.83)** R N 11,742 11,579 11,579 * p<0.05; ** p<0.01 For the varable of nterest, the results show a return of 5.1% for every extra year of schoolng n the Mnceran regresson. As expected the rate of return of the actual years of schoolng s lower, by 3 percentage ponts, than the return to the requred years of schoolng. Indvduals holdng the exact amount of schoolng as requred by the job receve 8.1% for every year of schoolng they acqured. Indvduals holdng surplus schoolng receve a return of 1.3% for every year of extra schoolng they obtan above the requred level. On the other hand, undereducated ndvduals receve a wage penalty of 4.1% for every year they are undereducated. Lookng at the coeffcents of the nteractve terms, the experence and overeducaton explan a substtuton effect because of ther negatve value. Smlarly, for the adequately educated ndvduals the years of experence are a substtute. Overeducated and adequately educated ndvduals receve a somewhat lower return for every year of experence. Addtonally, the relatonshp between undereducated and experence s complementary suggestng that undereducated ndvduals receve a hgher wage for every year of hgher past experence. Overeducaton and Unemployment

6 Usng the Cyprus Labour Force Survey we nvestgate whether beng overeducated ncreases the lkelhood of beng unemployed especally n a perod of economc crss. The CY LFS contans nformaton on unemployed ndvduals regardng ther prevous occupaton and ndustry. Ths survey collected nformaton of ndvduals quarterly for years 2008, 2009 and Usng the years of schoolng and nformaton on last employment of the ndvdual we can dstngush whether the unemployed ndvdual was over- under or adequately educated. For the dependent varable, 1 was used for ndvduals who were at the tme of ntervew unemployed and 0 for those who were employed. The overeducated measure ncluded both unemployed and employed ndvduals wth the number of surplus years of schoolng. Every other varable n the logstc estmaton concerned everyone n the sample. To check f beng overeducated ncreases the probablty of beng unemployed a logt model was used. Table 9: Logt model Unemployed (All) Unemployed (Males) Unemployed (Females) Over Educated (2.10)* (3.43)** (0.46) Under Educated (2.55)* (3.52)** (0.50) Requred (1.99) (1.04) (2.64)** Experence (14.68)** (7.39)** (10.76)** Exp Sex (21.30)** (13.92)** (14.84)** (6.65)** Marred (3.77)** (9.71)** (3.15)** General (4.14)** (4.06)** (1.79) Chldren Educaton (1.72) (0.44) (1.43) Art and Human Studes (1.05) (0.69) (0.88) Greek and Foregn Language (1.67) (0.41) (1.66)

7 Socal Scence (4.44)** (3.14)** (3.14)** Maths and Computer Scence (0.89) (0.46) (0.88) Envronmental Studes (2.40)* (0.98) (2.12)* Scence (1.85) (0.25) (2.36)* Computer Scence (2.43)* (2.61)** (0.32) Archtecture and Engneerng (3.42)** (3.78)** (0.42) Geologst (0.20) (0.05) (0.32) Health and Socal Studes (0.88) (1.45) (0.08) Servce Studes (3.08)** (3.83)** (0.44) Larnaca (4.38)** (2.45)* (3.89)** Lmassol (2.63)** (0.61) (4.16)** Famagusta (6.92)** (3.40)** (6.58)** Paphos (6.11)** (3.04)** (5.17)** _cons (9.96)** (7.50)** (9.57)** N 16,071 8,238 7,833 * p<0.05; ** p<0.01 The table above shows that there s a postve and sgnfcant relatonshp between beng overeducated and between the probablty of beng unemployed. Therefore t can be concluded that n a perod of economc crss the frst group to lose ther jobs are overeducated ndvduals. Addtonally t s shown that undereducated ndvduals, who are

8 subsequently pad less than ther co-workers, are less lkely to be unemployed. As expected, more experenced ndvduals are less lkely to lose ther jobs and males are found to be less lkely n beng unemployed compared to females. Movng to the second and thrd columns overeducated males are more lkely to be unemployed and undereducated males are less lkely as n the full sample. Smlarly, marred males are agan less lkely to lose ther jobs. On the other hand the results for over and under educated females are nsgnfcant but strangely adequately educated females are more lkely to lose ther job. Addtonally, marred women are more lkely to become unemployed. In the full sample and the males sample beng adequately educated s statstcally nsgnfcant and we cannot draw any results n the model. Ths could be because only a small amount of ndvduals who hold the requred amount of schoolng are unemployed. Ths mght be due to the lower reservaton wage of the undereducated relatve to ther overeducated counterparts. Subsequently employers do not fnd t proftable to employ overeducated ndvduals durng hard tme such as the current global downturn. A possble polcy mplcaton of the above would be to encourage mgraton for overeducated ndvduals n other countres of the European Communty. After the Bologna Process agreement n 1999, European countres amed for an easer occupatonal moblty of ndvduals resultng from the compatblty of qualfcatons. Ths agreement enforced all EU countres that by 2010 a European Hgher Educaton Area would be created. 3 Concluson Overeducaton exsts greatly n Cyprus. Almost 50% of the tertary sector graduates are found as beng overeducated ndcatng that ths s creatng a problem. Late reports have shown that one out of four unversty graduates s unemployed durng the economc crss n Cyprus. The effect of over- under- and adequate educaton on earnngs was tested usng Cyprot data confrmng that overeducated ndvduals receve less although stll postve returns for each surplus of schoolng they acqure. Addtonally, undereducated receve less than ther co-workers but more than ther counterparts n less-qualfed jobs. A new hypothess was tested whether overeducated ndvduals are more lkely n losng ther jobs especally n a perod of crss. The results show that overeducated ndvduals have a greater probablty of becomng unemployed n a perod of recesson. Ths suggests that overeducaton does not secure ndvduals from unemployment but can even cause t. Overeducated although not pad the full amount of money they should receve, they stll receve a postve amount over and above ndvduals that hold the requred amount of schoolng. As expected at some pont, s not proftable for employers to employ overeducated workers as t s more costly. Furthermore, undereducated ndvduals are less lkely n losng ther jobs, as expected, snce they are pad a lower return for ther defct schoolng than ther adequately educated co-workers. References 3 For more nformaton on the Bologna Process

9 1. Cohn, E., and Ng, Y. C. (2000). Incdence and wage effects of Overschoolng an underschoolng n Hong Kong, Economcs of Educaton Revew, 19, Decreuse, B. (2000). Over-educaton and long-term unemployment, unpublshed paper, Unverstѐ de la Mѐdterranѐe 3. Dolton, P., and Vgnoles, A. (2000). The ncdence and effects of overeducaton n the U.K. graduate labour market, Economcs of Educaton Revew, 19:2, Dolton, P., and Slles, M. A. (2008). The effects of over-educaton on earnngs n the graduate labour market, Economcs of Educaton Revew, 27, Duncan, G.J., and Hoffman, S.D. (1981). The Incdence and wage effects of Overeducaton, Economcs of Educaton Revew, 1:1, Frenette, M. (2004). The overqualfed Canadan graduate: the role of the academc program n the ncdence, persstence, and economc returns to overqualfcaton, Economcs of Educaton Revew, 23, Groot, W., and Maassen van den Brnk, H. (2000). Overeducaton n the labour market: a meta-analyss, Economcs of Educaton Revew, 19:2, Hartog, J., and Oosterbeek, H. (1988). Educaton, Allocaton and Earnngs n the Netherlands: Overschoolng?, Economcs of Educaton Revew, 7:2, Kker, B. F., Santos, M. C., and Mendes De Olvera, M. (1997). Overeducaton and Undereducaton: Evdence for Portugal, Economcs of Educaton Revew, 16:2, Mendes de Olvera, M., Santos, M. C. and Kker, B. F. (2000). The role of human captal and technologcal change n overeducaton, Economcs of Educaton Revew, 19, Menon, M. (2008), Perceved rates of return to hgher educaton: Further evdence from Cyprus, Economcs of Educaton Revew, 27, Rumberger R. (1987). The mpact of surplus schoolng on productvty and earnngs, Journal of Human Resources, 22:1, Scherman, N. (1991). Overeducaton n the labour market, Journal of Labour Economcs, 9:2,