Procedia - Social and Behavioral Sciences 195 ( 2015 ) World Conference on Technology, Innovation and Entrepreneurship

Transcription

1 Available online a ScienceDirec Procedia - Social and Behavioral Sciences 195 ( 015 ) World Conference on Technology, Innovaion and Enrepreneurship Tesing he Impac of Unemploymen on Self-Employmen: Evidence from OECD Counries Ferda Halicioglu a, Sema Yolac b a Deparmen of Economics, Yediepe Universiy, Isanbul, 34755, Turkey Deparmen of Business Adminisraion, Isanbul Universiy, Isanbul, 3445, Turkey b Absrac The impac of unemploymen on self-employmen is raher an ambiguous issue in economics. According o refugee effec approach, here are wo couner argumens: he heory of income choice argumen suggess ha increased unemploymen may lead o increased self-employmen aciviies whereas he couner argumen defends he view ha an increase in unemploymen raes may decrease he endowmens of human capial and enrepreneurial alen causing a rise in unemploymen raes furher. The empirical evidence on his issue seems o suppor boh hypoheses. This research presens fresh and more comprehensive evidence on his issue from 8 OECD counries using he ARDL approach o co-inegraion echnique over he period The empirical resuls indicae ha he firs hypohesis holds in he case of Belgium, Canada, Sweden and he UK whereas he second hypohesis is valid in he case of Greece, Luxembourg and Porugal. The empirical resuls for he remaining OECD counries did no reveal any long-run relaionship beween he variables in quesion. The empirical resuls are also evaluaed briefly for policy recommendaions. 015 The The Auhors. Published Published by by Elsevier Elsevier Ld. Ld. This is an open access aricle under he CC BY-NC-ND license (hp://creaivecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibiliy of Isanbul Universiy. Peer-review under responsibiliy of Isanbul Univerisy. Keywords: Self-employmen; unemploymen; coinegraion, OECD Corresponding auhor. Tel.: ; fax: address: fhalicioglu@yediepe.edu.r The Auhors. Published by Elsevier Ld. This is an open access aricle under he CC BY-NC-ND license (hp://creaivecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibiliy of Isanbul Univerisy. doi: /.sbspro

2 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) Inroducion The relaionship beween self-employmen and unemploymen presens a lively debae in economics. The origin of his debae is relaed o refugee effec which forms wo conflicing hypoheses. According o he heory of income choice, as he level of unemploymen raises i is expeced ha self-employmen sars o increase oo. As far as he couner argumen is concerned, he increased level unemploymen also leads o he depreciaion of human capial and skills which exacerbaes he exising unemploymen siuaion. The firs argumen of he refugee effec is also known as he unemploymen push hypohesis saes ha high unemploymen may reduce he opporuniy o gain salaried employmen and hus posiively affec self-employmen as discussed in Glocker and Seiner (007). According o Audresch e al. (005), he second hypohesis is coined as he unemploymen pull which suggess ha unemployed people end o possess lower endowmens of human capial and enrepreneurial alen o sar and susain a new firm. The empirical research on his issue seems o be raher ambiguous since he growing number of sudies presen evidence for he exisence of boh hypoheses. The ambiguiy in his issue migh be relaed o he fac ha he ime span, economeric mehodology and he variables in quesion vary considerably as far as he sudies are concerned. The main moivaion of his research is based on he fac ha self-employmen is regarded as one of he maor economic policy soluions o reduce he unemploymen in all counries. Thus, measuring he impac of unemploymen on self-employmen should reveal valuable policy informaion for policy makers. This research aims o conribue o he exising lieraure by providing furher ime series evidence on he refugee effec from 8 OECD counries using Auo Regressive Disribued Lag (ARDL) approach o coinegraion procedure. To our exising knowledge, here exiss no oher sudy uilizing his mehod previously in esimaion of he refugee effec. Moreover, he daa span and he exen of OECD counries in his sudy exceed he scope of oher sudies in he same caegory. Thus, he empirical resuls should be considered as more comprehensive. However, he primary aim of his sudy is o analyze empirically only he refugee effec; he economeric model is based on a univariae funcion disregarding oher possible facors ha may have impac on he refugee effec. This research is oulined as follows: he nex secion provides a brief review of heoreical and empirical sudies; secion 3 oulines he economeric mehodology; secion 4 presens and evaluaes he empirical resuls; and he final secion is devoed o conclusions.. A Brief Lieraure Review The discussions beween self-employmen and unemploymen lead he way o a growing body of empirical sudies in he lieraure in he las wo decades. Considering he size limiaions, his research focuses on seleced number of sudies o provide he main discussion poins of he lieraure. The back bone of his discussion revolves around he concep of he refugee effec. The refugee effec originaed from he simple heory of income choice which argues ha increased unemploymen will lead o an increase in sar-up aciviy on he grounds ha he opporuniy cos of saring a firm is less han being unemployed. In he same srand of his lieraure, a couner argumen indicaes ha he impac of unemploymen migh be derimenal on self-employmen due o he fac ha unemployed people no only lose heir obs bu hey may be deprived of he human capial and enrepreneurial skills which are required for new business aciviies. The second srand of he discussion is relaed o he Schumpeer effec which indicaes ha new-firm sar up reduces he level of unemploymen. Tha implies ha he direcion of relaionship runs from self-employmen o unemploymen. Differen aspecs of he refugee and Schumpeer effecs have been discussed and evaluaed heoreically and empirically in alarge number of sudies (Evans and Leighon, 1990; Alba-Ramirez, 1994; Audresch and Thurik, 000; Audresch e al., 00, 005; Carree e al., 00, 007; Risila and Tervo, 007; Bapisa and Preo, 007; Glocker and Seiner, 007; Faria e al., 010; Fairlie, 011; Yu e al., 014; and Aubry e al., 015). Self-

3 1 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) employmen is regarded as he maor proxy variable for he concep of enrepreneurship and i is also viewed as he engine of economic and social developmen in world economies. Thus, he problem of unemploymen can be alleviaed subsanially wih he incenives provided for he self-employmen aciviies. Wih a few excepions, a large number of empirical sudies provide suppor for his view. The empirical sudies on he refugee effec, by and large, appear o be focusing on developed counries especially in OECD counries. This endency is relaed o he fac ha he daa availabiliy and qualiy are beer and more easily accessible in developed counries. However, in recen empirical sudies, he numbers of developed counries were also invesigaed for his issue. Anoher aspec of he empirical sudies is ha analyses conain large number of local regions wihin counries in order o provide regional dispariies. Table 1 summarizes some of he empirical sudies on he refugee effec. I is crysal clear ha he resuls should be evaluaed on he basis of he ime span of he daa, he empirical mehodology and he counries or regions in quesion. Neverheless, Table 1 also demonsraes ha he economeric mehodology seems o be geing more sophisicaed as he ime ges close o he curren dae. The economeric mehodologies range from Ordinary Leas Squares (OLS) o sophisicaed Panel Economeric Mehods (PEM). In he recen empirical sudies, he researchers seem o be adoping more sophisicaed economeric procedures and longer daa ime span or he number of observaions. Increased daa qualiy and advanced economeric procedures encourage he researchers o conduc more comprehensive analyses on he refugee effecs. However, i should be emphasized again ha he previous empirical sudies on he refugee effecs have no uilized ye he economeric procedure of he ARDL approach o coinegraion as far as his research is concerned. Table 1. Summary Empirical Resuls on Refugee Effec Auhor(s) and Dae Daa Mehod Counries Refugee Effec (+/-) Evans and Leighon (1990) CS OLS 3 OECD + Audresch and Frisch (1994) CS OLS/WLS 75 regions in Germany - Alba-Ramirez (1994) CS OLS Spain and US + Audresch and Thurik (000) CS OLS 3 OECD + Audresch e al. (001) CS WLS 3 OECD + Caree e al. (00) CS WLS 3 OECD + Risila and Tervo (00) CS Probi Finland + Audresch e al. (005) TS VAR 3 OECD + Bapisa and Preo (007) TS VAR 30 regions in Porugal + Carree e al. (007) PD WLS 3 OECD + Glocker and Seiner (007) PD PEM Germany + Golpe and Seel (007) TS VAR 17 regions in Spain Mixed Faria e al. (009) TS VAR US, UK, Ireland, Spain Mixed Faria e al. (010) TS VAR 10 OECD + Fairlie (011) PD PEM US + Ghavidel e al. (011) PD SEM 7 Developing and 3 OECD Mixed Yu e al. (014) PD PEM US + Aubry e al. (015) PD PEM regions in France Mixed Noes: CS (Cross-Secion), TS (Time Series), PD (Panel Daa), OLS (Ordinary Leas Squares), PEM (Panel Economeric Mehods), SEM (Simulaneous Equaion Modelling), WLS (Weighed Leas Squares), VAR (Vecor Auo Regression).

4 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) Model and Economeric Mehodology 3.1 The Model Following he lieraure, we form he following long-run relaionship beween self-employmen and unemploymen raes, in double logarihmic linear form as: s a a1u 0 (1) where s is self-employmen raes and u unemploymen raes wih he subscrip indexes ime period wih =1986,, 013; and indexes for he counry in quesion, and is he classical error erm. 3. The Economeric Mehodology Advances in economeric lieraure dicae ha he long-run relaion in Eq. (1) should incorporae he shor-run dynamic adusmen process. I is possible o achieve his aim by expressing equaion (1) in an error-correcion model as suggesed by Engle-Granger (1987). Then, equaion (1) becomes as follows: m1 m b i si b i u 1,,, i, 1, i1 i0 s, b0 () where represens change, m i is he number of lags, is he speed of adusmen parameer and 1 is he one period lagged error correcion erm, which is esimaed from he residuals of equaion (1). The Engle-Granger mehod requires all variables in equaion (1) are inegraed of order one, I(1) and he error erm is inegraed order of zero, I(0) for esablishing a coinegraion relaionship. If some variables in equaion (1) are non-saionary we may use a new coinegraion mehod proposed by Pesaran e al. (001). This approach is also known as auoregressivedisribued lag (ARDL) ha combines Engle-Granger (1987) wo seps ino one by replacing 1 in equaion () wih is equivalen from equaion (1). 1 is subsiued by linear combinaion of he lagged variables as in equaion (3). n1 n c i si c i u i c s c u 1,,,, 3 1, 4 1, i1 i0 s, c0 v (3) To obain equaion (3), one has o solve equaion (1) for and lag he soluion equaion by one period. Then his soluion is subsiued for 1 in equaion () o arrive a equaion (3). Equaion (3) is a represenaion of he ARDL approach o coinegraion. Pesaran e al. (001) coinegraion approach, also known as bounds esing, has some mehodological advanages in comparison o oher single coinegraion procedures. Reasons for he ARDL are: i) endogeneiy problems and inabiliy o es hypoheses on he esimaed coefficiens in he long-run associaed wih he Engle-Granger (1987) mehod are avoided; ii) he long and shor-run coefficiens of he model in quesion are esimaed simulaneously; iii) he ARDL approach o esing for he exisence of a long-run relaionship beween he variables in levels is applicable irrespecive of wheher he underlying regressors are purely saionary I(0), purely non-saionary I(1), or muually coinegraed; iv) he small sample properies of he bounds esing approach are far superior o ha of mulivariae coinegraion, as argued in Narayan (005). The procedure is no longer valid in presence of I() series. The ARDL approach involves wo seps for esimaing he long run relaionship. The bounds esing procedure is based on a Wald ype (F-saisics) and is he firs sep of he ARDL coinegraion mehod. Accordingly, a oin

5 14 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) significance es ha implies no coinegraion under he null hypohesis, (H 0 : c c 0 ), agains he alernaive 3 4 hypohesis, (H 1 : a leas one c o c 0 ) should be performed for equaion (3). The F es used for his procedure 3 4 has a non-sandard disribuion. Thus, Pesaran e al. (001) compued wo ses of asympoic criical values for esing coinegraion for a given significance level wih and wihou a ime rend. One se assumes ha all variables are I(0) and he oher se assumes hey are all I(1). If he compued F-saisic exceeds he upper bound criical value, hen he null hypohesis of no coinegraion can be reeced. Conversely, if he F-saisic falls below he lower bound criical value, he null hypohesis canno be reeced. Lasly, if he F-saisic falls beween hese wo ses of criical values, he resul is inconclusive. The shor-run effecs beween he dependen and independen variables are inferred by he size of coefficiens of he differenced variables in equaion (3). The long-run effec is measured by he esimaes of lagged explanaory variables ha is normalized on esimae of c 3. Once a long-run relaionship has been esablished, equaion (3) is esimaed using an appropriae lag selecion crierion. A he second sep of he ARDL coinegraion procedure, i is also possible o obain he ARDL represenaion of he error correcion model. To esimae he speed wih which he dependen variable aduss o independen variables wihin he bounds esing approach, following Pesaran e al. (001) he lagged level variables in equaion (3) are replaced by EC -1 as in equaion (4): s. 0 k1 k i si i u i EC 1,,,, 1, i1 i0 (4) A negaive and saisically significan esimaion of no only represen he speed of adusmen bu also provides an alernaive means of supporing coinegraion beween he variables. 3.3 Alernaive Evidence of Coinegraion I has been proven by Bahmani-Oskooee and Goswami (003) ha F-esing sage is very sensiive o he seleced lag lenghs in equaion (3). Therefore, he resuls obained a his sage are no very conclusive. According o Bahmani-Oskooee and Ardalani (006), his siuaion can be avoided if he lagged linear combinaion of all variables in equaion is subsiued by EC -1 as expressed in equaion (4). Equaion (4) presens an alernaive evidence of co-inegraion by he coefficien esimae of EC -1. A negaive and significan coefficien of EC -1 could also reflec coinegraion among he variables. In paricular, his indicaes clear suppor for he shor-run adusmen oward long-run equilibrium as well as coinegraion. Moreover, Kremers e al. (199) and Baneree e al. (1998) also proved ha a negaive and significan EC -1 could be used as an alernaive evidence of coinegraion in he case of he Engle-Granger (1987) approach. Therefore, his sudy will uilize he resuls from error correcion model o esablish he exisence of coinegraion alernaively if he pre-esing sage of he Pesaran e al. (001) fails o do so. 3.4 Daa The daa period for each counry along wih variable definiions and daa sources are presened in his secion. All daa comes from OECD Main Economic Indicaors. The daa span is no he same for all he counries due o missing years. The counries and heir annual daa span used in his sudy are as follows: Ausralia ( ), Ausria ( ), Belgium ( ), Canada ( ), Finland ( ), France ( ), Germany ( ), Greece ( ), Hungary ( ), Iceland ( ), Ialy ( ), Japan ( ), Souh Korea ( ), Luxembourg ( ), Mexico ( ), The Neherlands ( ), New Zealand ( ), Norway ( ), Poland ( ), Porugal ( ), Spain ( ), Sweden ( ), Swizerland ( ), Turkey ( ), UK ( ), and USA ( ).

6 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) Variables s is he annual percenage of self-employmen in oal employmen in naural logarihm. Source: OECD u is he harmonized annual unemploymen rae in naural logarihm. Source: OECD. 4. Empirical Resuls 4.1 The F-es Equaion (3) is esimaed for 8 OECD counries using seleced annual daa over he period Time series properies of he variables are checked wih alernaive uni roo esing procedures. The uni roo resuls demonsrae ha all variables are in he order of inegraion of eiher I(0) or I(1). However, his sage of he economeric resuls is no repored here due o space consideraions. The pre esing sage of he ARDL approach o coinegraion is sensiive o he number of lags o be imposed on each differenced variable in equaion (3). To avoid his problem, an iniial lag of is imposed on each differenced variable o minimize he loss of degrees of freedom since we use limied annual daa. Then, R AIC (Akaike Informaion Crierion), SBC (Schwarz Bayesian Crierion) or HQC (Hannan Quinn Crierion) are being employed o selec he opimum number of lags. All resuls, herefore, belong o opimum models. The summary resuls from he ARDL approach o coinegraion are repored in Table. In regards o F saisics, here are only hree coinegraion relaionships namely in he case of Belgium, France and he UK. Considering a negaive and saisically significan EC -1 is considered o be an alernaive way of supporing coinegraion, Ausria, Canada, Belgium, France, Greece, Luxembourg, Porugal, Spain, Sweden and he UK fall ino his caegory. Finally, as far as he saisically significan long-run parameer and lagged error correcion erm in quesion is being considered o be plausible choice, here exis only seven counries, namely Belgium, Canada, Greece, Luxembourg, Porugal, Sweden and he UK. Table. ARDL Approach o Coinegraion Summary Resuls Counry Order of a ARDL F saisics Longrun slope EC1 R Shor Run Diagnosics SC FF Ausralia SBC (,) Ausria AIC (,0) Belgium R (, ) Canada HQC (1,0) Denmark R (, 1) Finland SBC (0,0) NA 0.10 NA NA NA NA NA NA France SBC (,1) Germany SBC (1,0) Greece SBC (1,) Hungary SBC (1,0) NA NA NA NA NA NA Iceland R (, 1) Ireland SBC (1,0) Ialy SBC (1,0) Japan R (, 0) Souh Korea R (, 1) Luxembourg AIC (1,1) Mexico AIC (1,0) Neherlands SBC (1,1) New Zealand HQC (1,0) Norway AIC (1,0) Poland SBC (1,0) Porugal R (1, 0) N H

7 16 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) Spain HQC (,0) Sweden AIC (1,) Swizerland HQC (1,0) Turkey HQC (1,0) UK SBC (1,0) USA AIC (1,0) a R, AIC, SBC, and HQC crieria are uilized appropriaely o selec he order of ARDL. The order of opimum lags is based on he specified ARDL model. For example, SBC (, ) for Ausralia suggess ha lags are imposed on Self-employmen rae and lags on Unemploymen rae in equaion (3). F sands for he compued F saisics for he bounds es. F criical value for esing he exisence of a long-run relaionship a 95 and 90 % level of significances respecively for I(0) and I(1)., and indicae, 1%, 5% and 10% significance levels, respecively. NA (No Available) sands for he fac ha here exis no dynamic economeric resuls from hese cases. SC, FF, N, and H are Lagrange muliplier saisics for ess of residual correlaion, funcional form mis-specificaion, non-normal errors and heeroskedasiciy, respecively. These saisics are disribued as chi-squared variaes wih degrees of freedom in parenheses. The criical values for (1) 3.84 and () a 5% significance level. 4. The Long-Run Resuls According o he resuls displayed in Table, he saisically significan slope parameers are available in wo differen accouns. If one deems ha he resuls from ARDL approach o coinegraion procedure should be considered only appropriae long-run oucomes, in his case here are 3 counries: Belgium, France and he UK. The empirical resuls from hese counries are also associaed wih saisfacory economeric diagnosics which make he inferences reliable and consisen. Wihin his caegory, he case of Belgium and he UK exend he suppor for he hypohesis of he refugee effec ha increased unemploymen will also raise he self-employmen level, whils he reverse hypohesis is valid in he case of France. As far as he degree of posiive impac of unemploymen is concerned, i is he highes in Belgium. The slope parameer of he Belgium economeric model, 0.77 indicaes ha a 1% rise in unemploymen on average increases self-employmen by 0.77 % on average during he esimaion period. If we assume in broad erms ha he long-run relaionships occur wih a saisically significan lagged error erm and long-run slope parameer, wihin his caegory here are seven counries: Belgium, Canada, Greece, Luxembourg, Porugal, Sweden and he UK. I appears ha he long-run slope parameer of Belgium, Canada and he UK economeric equaions provide suppor for he posiive relaionship beween unemploymen and selfemploymen whereas he resuls of Greece, Luxembourg and Sweden are relaed o negaive effec of he refugee effec. I is clear ha he negaive impac of he refugee effec on self-employmen is no a desirable choice. However, in his siuaion he counries should design proacive economic policies o comba hese impacs effecively. Designing appropriae economic policies will be relaed o he exen of he problem and he economic srucure of he counry quesion. 5. Conclusions The aim of his paper was o es he exisence of he refugee effec for 8 OECD counries. This obecive was aided by he echnique of Pesaran e al. (001) approach o coinegraion which presens non-spurious esimaes. Subsequenly, our work provides fresh evidence on he relaionship beween self-employmen and unemploymen. The resuls reveal ha here exis long-run relaionships in he case of only seven counries in broad erms. Wihin hese long-run relaionships, Belgium, Canada, Sweden and he UK offer suppor for he posiive impac of increased unemploymen on self-employmen suggesing ha increased unemploymen will simulae he new business sars ups whils i is observed empirically ha Luxembourg, Greece and Porugal will be suffering he

8 Ferda Halicioglu and Sema Yolac / Procedia - Social and Behavioral Sciences 195 ( 015 ) derimenal effec of increased unemploymen furher. The economeric resuls did no reveal any long-run relaionships in he case of ou of 8 counries of OECD. I is crysal clear ha he empirical resuls in his sudy are subec o some limiaions in regards o he daa ime span and omied variable bias. However, his is he firs ime series sudy ha has uilized he ARDL approach o coinegraion procedure. I is envisaged ha his paper will simulae furher empirical sudies in his naure o reveal more comprehensive insighs ino he undersanding of he refugee effec. As for policy recommendaions, counries should provide furher incenives for enrepreneurial spiris for sound and susainable economic growh. Selfemploymen as he mos plausible proxy of enrepreneurship is he backbone of economies. Therefore, i is essenial ha his ype of business aciviies requires special aenion and incenives all he ime so ha counries mainain sound and susainable economic growh raes. References Alba-Ramirez, A. (1994). Self-employmen in he mids of unemploymen: he case of Spain and he Unied Saes. Applied Economics, 6, Aubry, M., Bonne, J., & Renou-Maissan, P. (015). Enrepreneurship and business cycle: he Schumpeer effec versus he refugee effec a French appraisal based on regional daa. Annals of Regional Science, 54, Audresch, D.B., & Frisch, M. (1994). The geography of firms birhs in Germany. Regional Sudies, 8(4), Audresch, D.B., & Thurik, A.R. (000). Capialism and democracy in he 1s Cenury: from he managed o he enrepreneurial economy. Journal of Evoluionary Economics, 10, Audresch, D.B., Carree, M.A. & Thurik, A.R. (001). Does enrepreneurship reduce unemploymen? Tinbergen Insiue Discussion Paper, 074/3, Audresch, D.B., Carree, M.A., van Sel, A.J., & Thurik, A.R. (005). Does self-employmen reduce unemploymen? Scales Paper, Bahmani-Oskooee, M., & Goswami, G.G. (003). A disaggregaed approach o es he J-curve phenomenon: Japan versus her maor rading parners. Journal of Economics and Finance, 7, Bahmani-Oskooee, M., Ardalani, Z. (006). Exchange rae sensiiviy of U.S. rade flows: evidence from indusry daa. Souhern Economic Journal, 7, Bapisa, R. & Preo, M. T. (007). The dynamics of causaliy beween enrepreneurship and unemploymen. Inernaional Journal of Technology, Policy and Managemen, 7(3), Baneree, A., Dolado, J.J., & Mesre, R. (1998). Error-correcion mechanism ess for coinegraion in a single equaion framework. Journal of Time Series Analysis, 19, Carree, M., van Sel, A., Thurik, R., & Wennekers, S. (00). Economic developmen and business ownership: an analysis using daa of 3 OECD counries in he period Small Business Economics, 19, Carree, M., van Sel, A., Thurik, R., & Wennekers, S. (007). The relaionship beween economic developmen and business ownership revisied. Enrepreneurship & Regional Developmen, 19, Engle, R.F., & Granger C.W.J. (1987). Coinegraion and error correcion: represenaion, esimaion, and esing. Economerica, 55, Evans, D.S., & Leighon, L.S. (1990). Small business formaion by unemployed and employed workers. Small Business Economics,, Fairlie, R. (011). Enrepreneurship, economic condiions, and he grea recession. IZA Discussion Paper, 575, Faria, J.R., Cuesas, J.C., & Mourelle, E. (010). Enrepreneurship and unemploymen: a nonlinear bidirecional causaliy? Economic Modelling, 7, Faria, J.R., Cuesas, J.C., & Gil-Alana, L.A. (009). Unemploymen and enrepreneurship: a cyclical relaion? Economic Leers, 105, Ghavidel, S., Faradi, G., & Mohammadpour, A. (011). The relaionship beween enrepreneurship and unemploymen in developed and developing counries. Inernaional Conference on Applied Economics, Glocker, D. & Seiner, V. (007). Self-employmen: a way o end unemploymen? Empirical evidence from German pseudo-panel daa. IZA Discussion Paper, 561, 1-5. Golpe, A. & van Sel, A. (007). Self-employmen and unemploymen in Spanish Regions in he period Jena Economic Research Papers, 01, Kremers, J.M., Ericsson, N.R., & Dolado, J.J. (199). The power of coinegraion ess. Oxford Bullein of Economic Saisics, 54, Narayan, P.K. (005). The saving and invesmen nexus for China: evidence from coinegraion ess. Applied Economics, 37, Pesaran, H.M., Shin, Y., & Smih, J.R. (001). Bounds esing approaches o he analysis of relaionships. Journal of Applied Economerics, 16, Risilä, J. & Tervo, H. (00). Effecs of unemploymen on new firm formaion: micro-level panel daa evidence from Finland, Small Business Economics, 19, Yu, L., Orazem, P.F. & Jolly, R.W. (014). Enrepreneurship over he business cycle. Economic Leers, 1,