Far Eastern Studes Vol.7 May 2008 Center for Far Eastern Studes, Unversty of Toyama Analyss of Unemployment Durng Transton to a Market Economy: The Case of Lad-off Workers n the Bejng Area Jun MA 1 Hroko IMAMURA 2 and Guo Qng ZHAO 3 Abstract Usng self-gathered data and the hazard model, we analyzed the reasons lad-off workers n the Bejng area remaned unemployed. For the lad-off workers who were eventually reemployed, we used the Tobt model to explan the change n wages between ther pre- and post-layoff employment. The results of these analyses revealed that, although Chna s polcy of layng off workers from state-owned enterprses wthout completely severng the relatonshp accelerated the establshment of a labor market to some extent, wth prolonged tme the lad-off workers are out of work, the polcy was an obstacle to the labor-marketng effort. Moreover, t was an obstacle to mplementng and mprovng an unemployment benefts system. They also showed that lad-off workers wth worked experence under the tradtonal system need to be reeducated to enable them to adapt to the new economc envronment as well as to be taught new sklls, especally those workers who were managers. Addtonally, the "lay-off" s not only a polcy n specal phase durng the process of Chnese economc system reform, but also a specal phenomenon, whch s the structural unemployment revealed from Chnese economc system. Ths unemployment due to changes n ndustry structure means that the government must adopt a more actve employment polcy. Keywords: Transton Economy, unemployment duraton, state-owned enterprses (SOE), reemployment, Chnese Economy 1. Introducton A recent phenomenon n Chna s the reemployment of workers who had prevously been lad off. Under the prevous economc regme, workers were generally employed for lfe, so the concept of beng lad off was a foregn one. Now, wth the ntroducton of captalsm n Chna, ths concept s no longer alen. The government now faces the problem of helpng lad-off workers fnd new jobs, and new polces are needed to promote reemployment. We have nvestgated ths problem usng data we collected on lad-off workers n the Bejng area. From the mcroeconomcs vewpont, n ths paper we address the behavor dsplayed by lad-off workers durng ther unemployment. The results should help polcy makers n Chna better understand and address the problem of reemployng unemployed workers. 1 Faculty of Economcs, Unversty of Toyama, 3190 Gofuku Toyama-cty Toyama 930-8555, Japan, E-mal: majun@eco.toyama-u.ac.jp 2 Center for Far Eastern Studes, Unversty of Toyama, Japan 3 The School of Economcs, Renmn Unversty of Chna, Chna
Far Eastern Studes Vol.7 Ths paper s organzed as follows. Frst, by surveyng the latest related lterature, we clarfy the sgnfcance of and motvaton for our research. Second, by comparng our surveys wth macroeconomc data and mcro surveys, we descrbe the characterstcs and relablty of our surveys. Thrd, usng estmates of the hazard functon for the data of ndvdual lad-off workers, we dentfy the factors contrbutng to more rapd reemployment of lad-off workers. Fourth, we use the Tobt model to analyze these factors and to analyze the change n ncome after reemployment. Fnally, we address the polcy mplcatons of these results. 2. Lterature Revew A country transtonng to a market economy often must face a hgh unemployment rate and lengthy perods of unemployment. Studes on these problems have generally concentrated on two thngs: the rates and duratons are remarkably dfferent n the country n the same transtng-economy. The analyss of these dfferences s not only an mportant functon of transton economcs but also an mportant source of nformaton for polcy makers. Durng the transton, the government has to reduce ts nterventon n the marketplace and promote the development of a market mechansm. It must also mplement an effcent socal securty system. However, these tasks are somewhat at odds. Therefore, searchng for a polcy that addresses both has been a challenge for both researchers and polcy makers. A prmary task s dscussng the effects on the duraton of unemployment of a passve employment polcy (e.g., provdng unemployment nsurance) and of an actve employment polcy (e.g., provdng occupatonal tranng for unemployed workers). Although a passve employment polcy provdes socal securty for a more successful transton, t ncreases government nterventon and spendng, whch degrades economc effcency and seduces the moral hazard of unemployed workers. Although an actve employment polcy promotes the labor-marketng effort, whether the unemployed can be reemployed depends somewhat on the effectveness of the tranng. A dscusson of the factors affectng the unemployment duraton should provde basc nformaton useful to the development of an approprate polcy. Usng the hazard model and data on unemployed workers regstered at labor offces, Ham et al. (1998) analyss comes from ther Oaxaca-type decomposton of the dfference n the (non-lner) expected unemployment duraton between the Czech and Slovak republcs. Wth the average unemployment spell beng four tmes longer n Slovaka than the Czech Republc, the authors fnd that nearly one-thrd of ths dfference s explaned by dfferences n observable demand condtons and the ndustral employment structure n the
Analyss of Unemployment Durng Transton to a Market Economy two republcs. The remanng two-thrds are accounted for by the dfferent behavor of frms, ndvduals and nsttutons n the labor market, as reflected by dfferences n the coeffcents of the hazard functons. Very lttle of the dfference n expected unemployment duraton comes from dfferences n the demographc varables between the two republcs. As a country transtonng to a market economy, Chna confronts the questons of how to realze labor-marketng and how to desgn a socal securty system. In the md-1990s, Chna started to gradually layoff workers who were made redundant by the reform of state-owned enterprses (SOEs). Although these workers left ther workplace, they stll mantaned a relatonshp wth ther company, and the government encouraged them to fnd new jobs. Durng the reform, the government opened employment servce centers, pad lad-off workers an allowance, provded professonal tranng, and recommended jobs. Snce the lad-off workers mantaned a relatonshp wth ther company, the statstcal data on unemployment dd not nclude them. Hu (1998) addressed the relatonshp between the number of lad-off workers and the offcal unemployment rate n Chna from the macroeconomc vewpont. He calculated the number of unemployed workers and the true unemployment rate. As shown n fgure 1, a large proporton of the unemployed workers were lad-off workers, especally from 1998 to 2000, when the proporton was greater than 50%. Lad-off workers were thus a sgnfcant porton of the unemployed durng Chna s transton. Ths specal layoff polcy durng the reform of Chna s economc system and the development of Chna s economy had a dualstc nature. On the one hand, workers who had once enjoyed lfetme employment were dsmssed from ther jobs and pushed nto the labor market, whch promoted labor-marketng. On the other hand, the contract between the workers and the enterprses was not broken and s often ntervened by government. Therefore, durng the mperfecton of the unemployment benefts system, the government also provded unemployment securty. The queston then s what s best behavor that of the lad-off workers under ths specal layoff polcy? That s, what were the results of the layoff polcy? Dd t promote labor-marketng? Dd t mprove unemployment securty? Or dd t do both? Analyss of these questons should provde not only a factual bass for evaluatng the measures taken by Chna durng the reform of ts economc system but also provde a theoretcal bass for desgnng a polcy that promotes labor-marketng.
Far Eastern Studes Vol.7 1400 1400 1200 1200 1000 1000 800 800 600 600 400 400 200 200 0 1993 1994 1995 1996 1997 1998 1999 2000 1993 1994 1995 1996 1997 1998 1999 2000 Number of unemployed workers, (10,000) Number of unemployed workers, (10,000) Unemployment rate(ncludng lad-off workers) Unemployment rate(ncludng lad-off workers) Unemployment rate(no ncludng lad-off workers) Unemployment rate(no ncludng lad-off workers) 7% 7% 6% 6% 5% 5% 4% 4% 3% 3% 2% 2% 1% 1% 0% 0% Fgure 1 Number of unemployed workers and unemployment rate ncludng lad-off workers Source: 1993 1995: Chnese Academy of Scences, 1998, p. 106. 1996 2000: Chna Labor Statstcal Yearbook (1997 to 2001) 3. Characterstcs and Relablty of Surveys Used Many Chna scholars, from varous ponts of vew, have used feld surveys and questonnare surveys to analyze the specal layoff polcy,.e., the specal unemployment phenomenon n Chna and the polcy measures taken to promote reemployment of lad-off workers. Many scholars have also performed smple statstcal analyss. For example, the Task Team of the Insttute of Labor Scence (2000), whch s afflated wth the Mnstry of Labor and Socal Securty, used the results of feld surveys and questonnare surveys of enterprses and lad-off workers n 17 Chnese ctes to analyze the relatonshps between lad-off workers and SOEs and to analyze the rate of reemployment. They found that SOEs lad-off about 1.7 mllon workers, whch s more than the offcal number. As the number of lad-off workers ncreased, reemployment became more and more dffcult, and the hgher the age of a worker, the lower the rate of reemployment. Mo and Yue (2001) used sample data from the questonnare surveys to analyze the reemployment of lad-off workers. They found that the lad-off workers were more lkely to be women, older workers, workers wth a lower level of educaton, to have worked for an SOE or a manufacturng company, to have been the ordnary workers, and the lower ncome famles. They also found that nearly half the lad-off workers found new jobs, that men were more compettve and enterprsng than women, that younger lad-off workers had an advantage n terms of reemployment whle older lad-off workers had an advantage n terms of self-employment, and that lad-off workers wth no more than a mddle school educaton
Analyss of Unemployment Durng Transton to a Market Economy were more lkely to become self-employed after beng lad off. Those wth a hgh school educaton accounted for most of the lad-off workers who set up ther own enterprse whle those wth at least a junor college educaton were the most lkely to be reemployed by an enterprse. In general, the lad-off workers who were reemployed were most lkely to be employed by prvate enterprses and to be engaged n the tertary ndustry. Fnally, the communty servce was the most mpotent feld for those lad-off workers who became self-employed or set up a new busness. In addton, a survey conducted n Bejng by the Team of Study for Chna s Ctes and Towns Labor Moblty (2002) showed that the lad-off workers had several characterstcs n partcular. The layoff rate for women was approxmately twce that for men (8.9% vs. 4.5%). About 70% were between 26 and 45 years old, and about 60% were between 36 and 45 years old. More than 60% had worked n manufacturng. More than half had no more than a mddle school educaton, and, on average, ther educaton level was lower than that of the non-lad-off workers. About 60% had ether never swtched jobs or had swtched jobs only once before beng lad off. The layoff perod for more than half of them was between 18 and 24 months. A search of the relevant lterature revealed no reports of an econometrc model beng used to analyze the unemployment duraton or the effects of ther reemployment. It also revealed no reports of a mcroeconometrc model beng used to analyze the data for ndvdual lad-off workers. We have used self-gathered data for lad-off workers n Bejng who were eventually reemployed to conduct econometrc analyss of the unemployment duraton and the factors that affected the change n ther ncome after reemployment. We used the results of the analyss to nvestgate the effects of the specal layoff polcy. In October 2001, we conducted a survey of lad-off workers n Bejng by random samplng and collected 1214 usable questonnares 4. Most of the statstcal propertes of the sample were bascally consstent wth those of prevous surveys. The rato of men to women was 42.9: 57.1. About 90% of the workers were between 26 and 45 years old, 77% were between 36 and 45 years old. Most of the enterprses that had lad-off the workers were large or medum-sze SOEs founded between 1949 and 1978. More than 70% of the reemployed workers found work n manufacturng. About 66% had a hgh school educaton. About 66% had worked for at least 15 years. At the tme of the survey, more than half of the workers had been out of work for one to three years. 4 The questonnare survey was also admnstered to lad-off workers n Shangha, Shenyang, and Lanzhou. The data for Bejng was used because t has the strctest government regulaton as Bejng s the captal of Chna. Moreover, the functon of the market s the weakest n the whole country.
Accordng Far Eastern Studes Vol.7 to the Chna Statstcal Yearbook, of the 74,473 lad-off workers of Bejng n 2000, 63% were women, 67% were between 35 and 46 years old, and more than half had no more than a mddle school educaton. Comparson of these basc statstcal propertes wth ours confrms the credblty of our data. 4. Factors Contrbutng to More Rapd Reemployment of Lad-off Workers 4.1 Econometrc Model We used the hazard model to analyze the layoff duraton. Ths model s often used n the mcroeconometrc analyss of labor economcs to analyze the factors that affect the unemployment duraton. In ths secton we ntroduce the theoretcal background for applyng the hazard model to the unemployment duraton. Mortensen (1986) dscussed the relatonshp between the hazard model when t s appled to the unemployment duraton and the theory of job search and conducted a lterature revew on the foundaton and applcaton of ths model. Here we use the basc dea of the hazard model to dscuss the whole course of transtonng from beng lad off to becomng reemployed. We assume that lad-off workers fall nto only two categores: those who are unemployed and seekng work (ndcated by u) and those who have found new employment (ndcated by e). Retreat from the labor market s not consdered. For the convenence of analyss, we make several more assumptons. (1) Enterprse F provdes several employment opportuntes wth a wage level of w (wth dstrbuton functon p(w)) for lad-off workers, and a certan proporton of the lad-off workers () can obtan one of these opportuntes. The probablty that a lad-off worker can obtan one of these job opportuntes durng a short tme nterval s. The number of opportuntes and the dstrbuton of the wage level (p(w)) are ndependent. Each worker knows the dstrbuton of p(w) but not the actual wage level offered by an enterprse. (2) Durng the unemployment duraton, the utlty level of lad-off workers s v u, and t s ndependent of the wage level provded by the enterprse and of the dstrbuton of p(w). It does not change wth the unemployment duraton. (3) The utlty of lad-off workers after reemployment s a functon of the wage level: v e = v e (w). It s ndependent of the probablty dstrbuton of job opportuntes. (4) There s dmnshng margnal utlty,.e., dv e /dw > 0. (5) The tme when lad-off workers are gven a job opportunty s random, and they can accept or reject t.
Analyss of Unemployment Durng Transton to a Market Economy Under these assumptons, the optmum behavor of lad-off workers can be descrbed by the strategy wth a reservaton wage. That s, a lad-off worker wth reservaton wage w * wll accept a job opportunty f the offered wage level s hgher than w *. Otherwse, the worker wll declne the opportunty and contnue searchng for a job. The expected utlty level of reemployment s an ncreasng functon of the wage level and s ndependent of the wage levels that are not accepted. If, at wage level w, a lad-off worker prefers employment to unemployment, the worker wll also prefer employment to unemployment at wage levels exceedng w. The w * s the wage level at whch the worker s ndfferent between employment and unemployment. Therefore, the probablty of a job opportunty beng accepted by a lad-off worker s the probablty that the wage level s hgher than w *. The probablty can be defned by w p ( w ) dw *. (1) The probablty of transtonng from unemployment to reemployment depends on the probablty of a reemployment opportunty beng accepted by a lad-off worker. That s, the hazard functon of the unemployment duraton dstrbuton s. (2) However, the functon does not contan the observed value and a varable that represents the unemployment duraton. We thus defne the dstrbuton of the unemployment duraton for lad-off workers by usng the log-logstc hazard functon: The survvor functon s x t 1 x t 1 x, t S x t 1 1 x t. (3),. (4) We use the maxmum lkelhood estmaton for the survvor functon to estmate the effects of varous factors on unemployment duraton. Ths regresson model contans explanatory varables for the factors that affect the job opportunty offered to lad-off workers and those that affect the acceptance of a job opportunty.
4.2 Model Applcaton Far Eastern Studes Vol.7 From the data collected n our questonnare surveys, we extracted the date when each worker was lad off and the date of reemployment. We then calculated the unemployment duraton. Ths duraton reflects the speed of reemployment: the longer the duraton, the lower the speed of reemployment. We then appled Kaplan-Meer survval curves to the data. We found that, for heterogeneous lad-off workers, the unemployment duraton for about 50% of the workers was greater than four years (see Fgure 2) and that the proporton of men decreased over tme (see Fgure 3). Ths shows that the men were more quckly reemployed than the women. We also found that the duraton of the unemployment dffered between workers who were between 36 and 45 years old and those who were above 46 years old (see Fgure 4). Lad-off workers who were less than 35 years old had the shortest duraton. Furthermore, the duraton for lad-off workers whose educaton level was hgh school and above was, as expected, shorter than that of workers whose level was mddle school and below (see Fgure 5). That s, the educaton level affected the speed of reemployment. Next, we analyzed the factors that affect the lad-off duraton (or the speed of reemployment). In the questonnare surveys, we classfed these factors nto sx categores: (1) basc personal state of the lad-off workers; (2) enterprse state of the workers before beng lad off; (3) work state of the workers before beng lad off; (4) work state of the workers after beng reemployed; (5) enterprse state of workers after beng reemployed; (6) present work state of the workers. Fgure 2. Change n unemployment duraton for samplng of lad-off workers
Analyss of Unemployment Durng Transton to a Market Economy Fgure 3. Change n unemployment duraton by sex for samplng of lad-off workers Note: x11, x12 ndcate men and women, respectvely. Fgure 4. Change n unemployment duraton by age for samplng of lad-off workers Note: age 35, age 45, and age 55 ndcate below 35 years old, between 36 and 45 years old, and above 46 years old, respectvely.
Far Eastern Studes Vol.7 Fgure 5. Change n unemployment duraton by educaton level for samplng of lad-off workers Note: school3 and school4 ndcate mddle, and hgh school and above, respectvely. Frst of all, we consder how the factors n the enterprse state of the workers before beng lad-off category affect the unemployment duraton. These factors are the enterprse sze, the ownershp form, the ndustry and the management stuaton of the enterprse, the foundng date of the enterprse, and so on. Regulatons requre that the lad-off workers regster wth the reemployment servce center of ther orgnal enterprse, draw unemployment compensaton from the center, and take part n reemployment tranng offered by the center. The orgnal enterprse must help ts lad-off workers fnd new jobs. Ths means that the factors n ths category partly determne the ablty of the enterprse to provde reemployment opportuntes to lad-off workers. Of course, the fnancal capacty and wllngness of the orgnal enterprse to pay unemployment compensaton to ts lad-off workers affects ther utlty levels, reservaton wages, and so on. The enterprse s sze, ownershp form, and management stuaton determne ts capacty and wllngness to pay unemployment compensaton to ts lad-off workers. Snce these factors have dual effects, we can use regresson analyss to determne whch effects are sgnfcant. We desgnated several of the orgnal enterprse factors (sze, ownershp form, ndustry, management stuaton, and foundng date) as dummy varables for the analyss. We classfed enterprse sze nto fve ntervals: 100 employees or less, 101 to 300, 301 to 1000, 1001 to 2000, and above 2000. We suspect that the sze partly affects unemployment duraton, but whether the effect s postve or negatve s a queston to be answered. For example, a large enterprse may be more concerned about the welfare of ts lad-off workers, so t would be more wllng to open a reemployment servce center and to provde more generous unemployment compensaton n a tmely manner. On the one hand, the reemployment center would support ther efforts to fnd new employment, whch would
Analyss of Unemployment Durng Transton to a Market Economy tend to shorten the unemployment duraton. On the other hand, the compensaton would reduce ther motvaton to fnd new employment, whch would tend to lengthen the unemployment duraton. Usng econometrc analyss, we nvestgated how enterprse szes affect the unemployment duraton. To analyze the effect of sze, we set dummy varable Fsze to 1 f the sze was 300 or less and to 0 f t was above 301. We classfed enterprse form as SOE, collectve enterprse, prvate enterprse, jont venture enterprse, jont-stock enterprse, or other. Most of the lad-off workers had worked for an SOE, and the workers lad off from SOEs were lkely to have receved better treatment than those lad off from the other types of enterprses n terms of housng, medcal treatment, compensaton, etc. Ths better treatment may have affected ther motvaton to fnd new employment. To analyze the effect of form, we set dummy varable Fform to 1 f the enterprse was an SOE and to 0 otherwse. We classfed the ndustres nto 12 types n accordance wth the dvson standard of State Statstcal Bureau of the People s Republc of Chna. Snce prevous studes have shown that lad-off workers are mostly from manufacturng, we analyzed the effect of the enterprse beng a manufacturng one on the unemployment duraton n relaton to that of the other types of ndustres. To analyze the effect, we set dummy varable Fndu to 1 f the ndustry was manufacturng and to 0 otherwse. We classfed the enterprse condton as very good, good, far, slght defct, severe defct, beng acqured, shut down, or bankrupt. The condton of the orgnal enterprse should t affect ts treatment of the lad-off workers and thus affect ther motvaton to fnd new employment. To analyze the effect, we set dummy varable Fcond to 0 f the condton was very good, good, or far and to 1 otherwse. We assume that, gven the dstnct dfferences between the varous stages n the development of the economc system n Chna, enterprses founded n dfferent stages have dfferent ams. For example, those founded n the early years would tend to be more paternalstc whle those founded more recently would tend to be more focused on the bottom lne. These dfferences may lead to dfferences n the unemployment duraton. We grouped the enterprse foundng dates nto fve ntervals: before 1949, 1950 to 1965, 1966 to 1978, 1979 to 1989, and after 1990. We defned 1978 as the dvdng lne and compared the effect of the enterprse beng founded before 1978 on the unemployment duraton wth that of t beng founded after 1979. We set dummy varable Ffoun to 1 f the enterprse was founded before 1978 and to 0 otherwse. In addton, personal nformaton on the lad-off workers and ther stuaton pror to
Far Eastern Studes Vol.7 beng lad-off lkely affected ther unemployment duraton. We thus defned varables for several types of personal nformaton, ncludng gender, age, and martal status, and defned seven varables for ther pre-layoff stuaton before beng lad-off: educaton level, length of work experence, type of work done, poston n the orgnal enterprse, work experence n the orgnal enterprse, employment status (full-tme, part-tme, etc.), and ncome. Table 1 summarzes the defntons and settngs of all the varables used. As mentoned above, surveys have shown that men can generally fnd reemployment more easly than women. A negatve coeffcent for our estmaton would support ths. We thus set Gender to 1 f the surveyed ndvdual was female and to 0 otherwse. Several studes have shown that lad-off workers wth a hgh school or hgher educaton can generally fnd reemployment more easly than those wth a mddle school or lower educaton. We thus used two dummy varables to analyze the effects of educaton level on the unemployment duraton. School1 reflects the low educaton levels and was set to 1 for mddle school or below and to 0 otherwse. School2 reflects the hgh educaton levels and was set to 1 for junor college or hgher and to 0 otherwse. To analyze of effects of a change n the type of work done on the unemployment duraton, we classfed the type of work done nto (1) sklled works and (2) other. If the reemployed workers who dd sklled works at the present enterprse had had a shorter unemployment duraton than those who dd other work, we can conclude that the change n the type of work done dd not affect the duraton. If the reemployed workers who dd sklled works had had a longer or the same unemployment duraton as those who dd other work, we can conclude that the change dd affect the duraton. We set dummy varable Ljob to 1 f the worker dd sklled workers and to 0 otherwse.
Analyss of Unemployment Durng Transton to a Market Economy Table 1. Varable defntons and settngs. Varable Defnton Settng Gender Male or female Dummy varable: woman = 1; man = 0 Age Age Age at tme of survey Age2 Square of age = age age Fage Interacton effect of sex = gender age and age School1 Low educaton level Dummy varable: mddle school and below = 1; hgh school and above = 0 School2 Hgh educaton level Dummy varable: junor college and above = 1; otherwse = 0 Mstatus Martal status Dummy varable: marred = 1; sngle = 0 Genderm Interacton effect of sex = Gender Mstatus and marrage Fsze Sze of orgnal enterprse Dummy varable: less than 300 employees = 1; 300 or more employees = 0 Fform Form of orgnal enterprse Dummy varable: SOE = 1; otherwse = 0 Fndu Industry of orgnal Dummy varable: manufacturng = 1; otherwse =0. enterprse Fcond Condton of orgnal Dummy varable: very good, good, or far = 0; otherwse = 1 enterprse Ffoun Orgnal enterprse Dummy varable: before 1978 = 1; otherwse = 0 foundng date Ljob Type of work done at Dummy varable: sklled workers = 1; otherwse = 0 orgnal enterprse Lexp Total length of work before Perod of work before beng lad off unemployment Llenth Length of work n orgnal Perod of work n orgnal enterprse before beng lad off enterprse Lpos Poston n orgnal Dummy varable: management staff = 1; otherwse = 0 enterprse Lstat Employment status n orgnal enterprse Dummy varable: permanent poston or long-term contract wth orgnal enterprse = 1; otherwse = 0 Learn Income before beng Monthly ncome before beng lad off lad-off Ldur Duraton of unemployment From start of lay-off to tme of survey Nfsze Sze of present enterprse Dummy varable: less than 300 employees = 1; 300 or more employees = 0 Nfform Form of present enterprse Dummy varable: SOE = 1; otherwse = 0 Industry of present Dummy varable: manufacturng = 1; otherwse = 0. Nfndu enterprse Condton of present Dummy varable: very good, good, or far = 0; otherwse = 1 Nfcond enterprse Present enterprse foundng Dummy varable: before 1978 = 1; otherwse = 0 Nffoun date Length of work n present Perod of work after beng reemployed Nllenth enterprse Type of work done at Dummy varable: sklled workers = 1; otherwse = 0 Nljob present enterprse Poston n present Dummy varable: management = 1; otherwse = 0 Nlpos enterprses Nlstat Employment status n present enterprse Dummy varable: permanent poston or long-term contract = 1; otherwse = 0 Income n present Monthly ncome after beng reemployed Nlearn enterprse Remp Reemployment Dummy varable: reemployed workers = 1; otherwse = 0 Note: orgnal enterprse s enterprse before beng lad off; present enterprse s enterprse after beng reemployed.
Far Eastern Studes Vol.7 Determnng Determnng whether whether know-how the know-how obtaned obtaned by managements by managements under under the planned the planned economy economy can be can appled be appled to management to management under under the market the market economy economy s an mportant s mportant task task durng durng the Chna s the Chna s transton transton to a market to a market economy. economy. We set We dummy set dummy varable varable Lpos to Lpos 1 f to the 1 f the lad-off lad-off worker worker had been had a been manager a manager and to and 0 otherwse. to 0 otherwse. A negatve A negatve coeffcent coeffcent for ths for ths varable varable would would show that show the that know-how the know-how can be can appled. be appled. If t s If postve s postve or zero, or t zero, cannot t cannot be be appled. appled. The lad-off The lad-off workers workers had dfferent had dfferent employment employment status status and accordngly and accordngly receved receved dfferent dfferent treatments treatments before before beng lad beng off. lad These off. These dfferences dfferences lkely affected lkely affected ther motvaton ther motvaton to fnd to new fnd employment. new employment. We set We dummy set dummy varable varable Lstat to Lstat 1 for to lad-off 1 for lad-off workers workers who had who had been permanent been permanent employees employees or who or had who had had long-term had long-term contracts contracts and to and 0 otherwse. to 0 otherwse. In short, In the short, regresson the regresson equaton equaton had 18 had varables: 18 varables: Gender, Gender, Age, Age2, Age, Age2, Fage, School1, Fage, School1, School2, School2, Mstatus, Mstatus, Genderm, Genderm, Fsze, Fsze, Fform, Fform, Fndu, Fndu, Fcond, Fcond, Ffoun, Ffoun, Ljob, Lexp, Ljob, Lexp, Lstat, Lpos, Lstat, Lpos, Learn. Learn. Table 2 Table shows 2 shows the basc the sample basc sample statstcs statstcs for each for varable. each varable.
Analyss of Unemployment Durng Transton to a Market Economy Table 2. Basc sample statstcs for each varable Varable Number of Average Standard Mnmum Maxmum Gender 1214 0.571 0.495 0 1 Age 1214 39.309 4.478 22 56 Age2 1214 1565.223 344.971 484 3136 Fage 1214 21.790 19.140 0 50 School1 1214 0.187 0.390 0 1 School2 1214 0.047 0.212 0 1 Mstatus 1214 0.942 0.235 0 1 Genderm 1214 0.547 0.498 0 1 Fsze 1214 0.179 0.383 0 1 Fform 1214 0.647 0.478 0 1 Fndu 1214 0.721 0.449 0 1 Fcond 1213 0.728 0.445 0 1 Ffoun 1214 0.833 0.373 0 1 Ljob 1214 0.757 0.429 0 1 Lexp 1214 19.043 5.044 2 36 Llenth 1214 16.375 4.937 2 36 Lpos 1214 0.152 0.359 0 1 Lstat 1214 0.592 0.492 0 1 Learn 1214 736.279 213.913 100 3000 Ldur 1214 3.177 1.542 0.17 10.67 Nfsze 574 0.777 0.417 0 1 Nfform 574 0.082 0.274 0 1 Nfndu 574 0.287 0.453 0 1 Nfcond 574 0.033 0.179 0 1 Nffoun 574 0.138 0.345 0 1 Nllenth 574 2.081 1.168 0.08 9.75 Nljob 574 0.333 0.472 0 1 Nlpos 574 0.226 0.419 0 1 Nlstat 575 0.019 0.137 0 1 Nlearn 574 793.949 505.029 240 6000 Remp 1214 0.472 0.499 0 1
4.3. Far Eastern Studes Vol.7 Results We used the log-logstc hazard functon descrbed n secton 4.1 for regresson analyss of the factors that affected the unemployment duraton of the lad-off workers. The estmated parameters are summarzed n Table 3. Table 3. Estmated parameters for unemployment duraton (1) (2) Explanatory varable Estmated coeffcent Z value Estmated coeffcent Z value Gender 0.3417 1.0500 0.2376 0.8010 Age 0.0909 * 1.8460 0.1104 ** 2.3940 Age2 0.0000 0.0480 0.0002 0.3710 Fage 0.0119 1.3840 0.0071 0.9130 School1 0.1095 ** 2.1680 0.0753 * 1.7040 School2 0.1087 1.3400 0.0410 0.5650 Mstatus 0.0012 0.0130 0.0381 0.4670 Genderm 0.3510 ** 2.4640 0.2946 ** 2.3070 Fsze 0.0010 0.0190 0.0011 0.0230 Fform 0.0448 1.1180 0.0168 0.4730 Fndu 0.0791 ** 2.0150 0.0531 1.5140 Fcond 0.0670 * 1.6790 0.0400 1.1310 Ffoun 0.0571 1.0800 0.0341 0.7180 Ljob 0.0238 0.5220 0.0098 0.2430 Lexp 0.0803 *** 9.0930 Llenth 0.0934 *** 12.2800 Lpos 0.0335 0.6220 0.0052 0.1090 Lstat 0.1567 *** 4.1080 0.1658 *** 4.8570 Learn 0.0002 ** 2.3970 0.0002 ** 2.0220 Const 0.7741 0.8040 1.3722 1.5110 Number of observatons = 1213 Number of observatons = 1213 Number of falures = 572 Number of falures = 572 Log lkelhood = 629.05 Log lkelhood = 592.93 Notes: *, **, and *** ndcate sgnfcance level of 10, 5, and 1%, respectvely. (1) The regresson coeffcents for the age varable (Age) and the educaton level varable (School1) are greater than zero. Ths means that, assumng the other varables are fxed, the older the lad-off worker and the lower the educaton level, the longer the unemployment duraton.
Analyss of Unemployment Durng Transton to a Market Economy (2) The regresson coeffcents for the gender varable (Gender) and the marrage status varable (Mstatus) were not sgnfcant, but the regresson coeffcent of the nteracton effect of sex and marrage varable (Genderm) was sgnfcantly greater than zero. Therefore, the martal status affected the unemployment duraton only for marred women. (3) The varables that reflected the ndustry of the orgnal enterprse (Fndu) were all greater than zero, but the sze, form, and foundng date of the enterprse had not sgnfcant effects. (4) The regresson coeffcent for the length of tme worked before beng lad off (Lexp) was sgnfcantly less than zero, and that for the length of tme worked n the orgnal enterprse (Llenth) was also sgnfcantly less than zero. (5) Whether the former job of a reemployed worker was sklled workers (Ljob) or management (Lpos) dd not sgnfcantly affect the unemployment duraton. (6) The regresson coeffcent for prevous employment status (Lstat) was sgnfcantly greater than zero. Ths means that the lad-off workers who had had permanent postons or long-term contracts had longer unemployment duratons than the other lad-off workers. (7) The regresson coeffcent for ncome before beng lad off (Learn) was sgnfcantly less than zero. Ths shows that the hgher the ncome before beng lad-off, the shorter the unemployment duraton. 5 Use of Tobt Model to Analyze Factors and Change n Income after Reemployment 5.1. Econometrc Model The above analyss of the factors affectng the unemployment duraton revealed that the lad-off workers who were younger, had a hgher educaton levels and a longer experence, and had a hgher ablty fndng new jobs more quckly. However, even f workers found new employment, ther ncomes were lower than before the layoff, the reemployment was lower level. That s, the standard of lvng was lower, whch resulted n a loss to the human captal. We thus conducted a regresson analyss of the factors that affected the ncome change to clarfy an effect of the reemployment. The results should provde basc nformaton for desgnng a sutable polcy for provdng unemployment nsurance and approprate occupatonal tranng of unemployed workers. In econometrcs, model analyss on the ncome of employed workers usually uses a censored sample, but ordnary least-squares estmatons of a censored regresson model generate based and nconsstent parameter estmates. Tobn developed the Tobt model to solve ths problem. We used the Tobt model to analyze the factors that affect the ncome change.
The basc form of the Tobt model s Far Eastern Studes Vol.7 y * x (5) y y 0 y * f f y y * * 0 0. (6) To analyze the ncome change, we defned * y as dfference between the logarthms of the earnngs before and after beng lad-off. * y log( Nl earn Learn ) (7) The basc assumpton of formula (6) s that, f lad-off workers are reemployed, ther earnng have changed. That s, y y 0 y * f f Nlearn Nlearn 0 0. (8) We used the probt model to estmate the probablty that the lad-off workers were reemployed. Snce we dd not collect data about the workers durng ther unemployment, we could only use the data we collected about them for the perod before they were lad off to analyze the factors that affected ther chances of beng reemployed. Smlar to our analyss of the factors that affected the unemployment duraton, we classfed the factors nto two groups. In one group were the factors that affected the opportunty that the lad-off workers accepted. It mostly contans varables reflectng the basc nformaton about the workers and ther experence before beng lad off. In the other were exteror factors that affected ther new employment opportunty. Frst we consder the characterstc varable of the orgnal enterprse. Whether lad-off workers fnd reemployment was formulated as Pr ob( Nlearn 0) z, (9) where z = (Gender, Age, Fage, School1, School2, Mstatus, Genderm, Fsze, Fform, Fndu, Fcond, Ffoun, Ljob, Lexp, Lpos, Lstat, Learn). Wth respect to the factors that affect the ncome after reemployment, except for the two factors above, we also consdered the propertes of the enterprse after reemployment and the effects of these propertes on the ncome change. The resultng model s formulated as log( Nlearn / Learn ) x, (10)
Analyss of Unemployment Durng Transton to a Market Economy where x = (Gender, Age, Age2, Fage, School1, School2, Mstatus, Genderm, Ljob, Lexp,, Nfsze, N, Nfndu, Nfcond, Nffoun, Nljob, Nlexp, Nlpos, Nlstat) The basc meanngs of the varables above are presented n Table 1, and ther basc sample statstcs are summarzed n Table 2. 5.2. Parameter Estmaton The model descrbed above and a censored sample were used to estmate the maxmum lkelhood for the varables affectng the ncome change. The results are shown n Table 4. The personal nformaton on the lad-off workers (Gender, Age, Age2, Fage) sgnfcantly affected the ncome change. For example, the older workers tended to have a lower ncome after reemployment, but the dfference gradually decreased wth an ncrease age. Addtonally, the women tended to have a lower ncome after reemployment, partcularly the older ones. The regresson coeffcents for the educaton levels (School1, School2) were not sgnfcant; the one for Lexp was sgnfcantly less than zero. The workers who obtaned a sklled job tended to have a hgher ncome than before whle the ones who obtaned a management or other type job had about the same ncome. The regresson coeffcents for the sze of the new enterprse (Nfsze) and the foundng date (Nffoun) were sgnfcantly less than zero. Ths ndcates that the lad-off workers who found employment n a smaller enterprse tended to have a lower ncome. Ths s attrbuted to smaller enterprses havng lower wage levels. The workers who found employment at enterprses founded before 1978 tended to have lower ncome. Ths s attrbuted to enterprses founded before 1978 havng lower wage levels than those founded after 1978.
Table 4. Far Eastern Studes Vol.7 Estmated results for ncome change Heckman selecton model (regresson model wth sample selecton) Number of observatons = 1212 Censored observatons = 641, Uncensored observatons = 571 Log lkelhood = 970.2942 Log(Nlearn /Learn ) Select (Prob(Nlearn > 0)) Explanatory varables Estmated coeffcents Z value Explanatory varables Estmated coeffcents Z value Gender 0.541 * 1.885 Gender 0.353 1.086 Age 0.101 ** 2.548 Age 0.121 5.181 Age2 0.002 ** 2.941 School1 0.071 0.631 Fage 0.015 * 1.900 School2 0.342 1.762 School1 0.055-1.152 Mstatus 0.191 0.818 School2 0.080 1.061 Genderm 0.561 1.678 Mstatus 0.055 0.628 Fsze 0.021 0.190 Genderm 0.014 0.108 Fpor 0.012 0.152 Ljob 0.141 ** 3.032 Fndu 0.039 0.447 Lexp 0.024 ** 2.547 Fcond 0.215 2.523 Lpos 0.041 0.058 Ffoun 0.205 1.823 Nfsze 0.155 *** 3.334 Ljob 0.288 2.819 Nfform 0.108 1.555 Lexp 0.106 5.188 Nfndu 0.012 0.323 Lpos 0.068 0.544 Nfcond 0.004 0.049 Lstat 0.056 0.684 Nffoun 0.212 *** 3.445 Learn 0.001 6.693 Nllenth 0.114 *** 8.276 Const. 4.937 7.407 Nljob 0.066 * 1.807 Nlpos 0.348 *** 8.006 Nlstat 0.010 0.081 Const. 2.061 ** 2.711 LR test of ndep. eqns. (rho = 0); Wald ch2(22) = 240.48; Prob > ch2 = 0.0000 Note: *, **, and *** ndcate sgnfcance level of 10, 5, and 1%, respectvely. The regresson coeffcent for work tme n the new enterprse (Nlenth) was sgnfcantly more than zero. And, the lad-off workers who found employment as managers tended to have ncreased ncome. In contrast, those took sklled jobs tended to have reduced ncome.
Analyss of Unemployment Durng Transton to a Market Economy 6. Concluson and Recommendatons Usng the hazard functon and a lmted dependent varable model, we analyzed the factors that may affect the unemployment duraton of lad-off workers n the Bejng area and the dfference n ncome before beng lad off and after reemployment. Age was found to have sgnfcantly affected not only the unemployment duraton, but also the ncome change. Whle gender dd not affect the duraton, t dd affect the ncome change. Analyss of the factors that affected the duraton revealed that the duraton for workers lad off by enterprse wth bad benefts tended to be longer. It also revealed that workers who were permanent employees or had a long-term contract wth ther prevous enterprse had longer unemployment duraton than those who had a short-term contract. From the results above, we can draw a prmary concluson: workers who are older women, especally those who are marred, wll consder the cost of fndng a new job and the opportunty cost of reemployment (such as the allowance for unemployment, the cost of famly labor, the mental pressure of fndng a new job, and the mental pressure of the works. If these costs exceed the benefts of reemployment, they wll tend to reject the reemployment opportunty; however, they do not gve up ther lad-off status and do not wthdraw from the labor market. They hold on to ther lad-off status as long as possble due to the benefts, such as the allowance, specal treatment for housng, and medcal nsurance. These benefts reduce the motvaton of lad-off workers who are older or who are marred women to fnd new employment. Gven these conclusons, we can say that, although the specal layoff polcy has promoted the establshment of a labor market to some extent, t has been an obstacle to labor-marketng because t tends to prolong the tme lad-off workers are out of work. Moreover, f the condton of the orgnal enterprse s bad and t does not have suffcent capacty to help ts lad-off workers fnd reemployment, and snce t has to shoulder the burden of provdng an allowance to the lad-off workers, ts condton wll become worse and worse. Ths wll make t more dffcult to set up and mprove a system of unemployment benefts under the market economc system. Governments of countres transtonng from a planned economc system to a market-orented one are often skeptcal of the role played by the market and tend to rely more on admnstratve decrees than market forces when problems occur durng the transton. As shown by the current analyss, a typcal example of ths s Chna s specal lay-off polcy, whch we consder a falure. Although admnstratve decrees regardng lad-off workers ntally played a postve role n the downszng of the state owned
Far Eastern Studes Vol.7 enterprses, as unemployment ncreased, the labor markets could not keep up. Moreover, the polcy dd not cultvate a compettve mndset n the SOEs. A more effectve polcy would be for the government to provde subsdes to enterprses employng lad-off workers and for job tranng of lad-off workers. Adopton of such polces would promote the role of the labor markets and make better use of the market mechansm. Analyss of the factors that affected the unemployment duraton also revealed that the lower the educaton level, the longer the duraton, the longer the work experence, the shorter the duraton (even f the experence was all at one enterprse), and the hgher the ncome before beng lad-off, the shorter the duraton. Analyss of the factors that affected the ncome change revealed that the longer the tme worked at the orgnal enterprse, the lower the ncome level after reemployment and the longer the tme worked at the new enterprse, the hgher the ncome level. From these results, we can draw a general concluson related to labor economcs. Because nformaton about the labor market s asymmetrcal, when a lad-off worker fnds a new job, the educaton level, work experence, and ablty of the worker becomes mportant nformaton to the new enterprse. On the other hand, a part of the human captal that was accumulated through prevous experence s a frm specfc skll, and the longer the tme worked at the orgnal enterprse, the larger the proporton of the frm specfc skll to the human captal. Consequently, the new ncome s less than the prevous one. Nevertheless, as the tme worked at the new enterprse ncreases, the worker s capacty gradually mproves, and the ncome naturally ncreases. From the analyss of the factors that affected the unemployment duraton, we found that whether a worker had a sklled job or a management job before beng lad off dd not sgnfcantly affect the unemployment duraton. In addton, from the analyss of the factors that affected the ncome change, we found that the workers who accepted a sklled job, tended to have a hgher ncome than before. In contrast, the workers who accepted a management or other job tended to have about the same ncome as before. From these results we can say that f the ndustry structure changes (for example heavy ndustry or labor ntensve ndustry decrease), demand of labor also changes. Durng the job search, the past experences nor sklls are not be advantage. The knowledge and sklls whch lad-off workers obtaned former enterprses do not have mportant condtons. For the economc system transton, the management experence ganed under the planned economc system can not be regarded as an effcent nformaton that the market economy. Therefore, n comparson wth the other types of workers, they do not have the advantage of the know-how durng the job search. Then the management knowledge and
Analyss of Unemployment Durng Transton to a Market Economy experence they ganed have not entrely accommodated to the market economy. In comparson to the workers who dd not fnd a management job, they do not have the advantage of the management skll under planed economc system, and ther ncome change s nsgnfcant. Ths tells us that the management knowledge and experence obtaned under the planed economc system are not entrely accommodated to the work demand under the market economc system. To help more lad-off workers fnd new jobs, the government must reeducate them so that they can adapt to the new economc envronment and t must teach them the sklls workers normally acqure under a tradtonal economc system, especally those who worked as managers. In a word, t s necessary to reeducate not only ther sklls but also the way of thnkng and the knowledge structures to make t adjust to work n the enterprse n the market economy. The duraton of the unemployment n the manufacturng was longer. Ths shows that the lay-off of many redundant workers was not only due to the economc transton, but also to changes n Chna s ndustral structure. That s, the lay-off was also a specal phenomenon they reflected the structural unemployment nherent n the planned economc system. Ths means that the government must mplement a more actve employment polcy, ncludng provdng more support for professonal tranng and promotng the creaton of prvate enterprses that can effectvely support such tranng. Although we dd not collect nformaton on the lad-off workers durng ther perod of unemployment, we estmated that some of them had been under employed or nvsbly employed durng the tme they were searchng for a new job. The unemployment duraton (or the job-search actvty) was probably affected by the earnngs from ths under or nvsble employment. Unfortunately, we were unable to analyze ths factor. In addton, we were unable to analyze the human captal for the reemployment qualty of the lad-off workers. To do ths, we need to nvestgate the relatonshp between the characterstcs of the new job and the past experence. These tasks are left for future study.
References Far Eastern Studes Vol.7 1. Dong, X. and Putterman, L., Investgatng the Rse of Labor Redundancy n Chna s State Industry, Chna Economc Quarterly (n Chnese), 2002, 1, 397 417. 2. Greene, W. H., Econometrc Analyss, 4th ed. Prentce-Hall, 2000. 3. Academy of Macroeconomc Research, Natonal Development and Reform Commsson, P. R. Chna, Chna s employment stuaton durng 1999 to 2001 and a study of related polces, Management World (n Chnese), 1999, 4, 71 81. 4. State Statstcal Bureau, P. R. Chna, Chna Labor Statstcal Yearbook, 1997 to 2001, Chna Statstcs Press. 5. Ham, J., Svejnar, J., and Terrell, K., Unemployment and the Socal Safety Net Durng Transtons to a Market Economy: Evdence from the Czech and Slovak Republcs, Amercan Economc Revew, 1998, 88(5), 1117 1142. 6. Hu, Angang. An Analyss of the Status of Unemployment n Chna s Towns and Ctes, Management World (n Chnese), 1998, 4, 47 63. 7. Kefer, N., Economc Duraton Data and Hazard Functons, Journal of Economc Lterature, 1998, 26, 646 679. 8. Lancaster, T., The Econometrc Analyss of Transton Data, Cambrdge, 1990. 9. Qang, L., Junsheng, H., and Dayong, H., A Comparatve Analyss of Unemployment and Lad off, 2001, Tsnghua Unversty Press. 10. 10. Mortensen, D. T, Job Search and Labor Market Analyss, Handbook of Labor Economcs, 1986, 2, 849 919 11.Rong, M. and We, Y., A Study to Increasng Job Opportuntes of lad-off workers, Management World (n Chnese), 2001, 2, 88 94. 12. Svejnar, J., Labor Markets n the Transtonal Central and East European Economes, Handbook of Labor Economcs, 1999, 3B, 1809 1857 13. Insttute of Labor Scence, A Study on Labor Relatons of Lad-off workers and State-owned Enterprses, Management World (n Chnese), 2000, 1, 104 110. 14. Chnese Academy of Scences, Report of State of the Naton, No. 6, Employment and Development, 1998, Laonng People s Publshng House. 15. The Team of Study for Chna s Ctes and Towns Labor Moblty, Labor Market Structure and Labor Moblty n Chna, Management World (n Chnese), 2002, 3, 74 79. Receved 18 January 2008; accepted 21 Aprl 2008