Wage curve in dual labor markets: cross-sectional evidence from Japan

Transcription

1 Economic and Buine Letter 4(2), 51-56, 2015 Wage curve in dual labor market: cro-ectional evidence from Japan Kazuyuki Inagaki * Graduate School of Economic, Nagoya City Univerity, Japan Received: 26 January 2015 Revied: 21 April 2015 Accepted: 22 April 2015 Abtract Uing a threhold regreion model, we find that the wage-unemployment relationhip in Japan depend on the degree of labor market duality. The unemployment elaticity of wage i negative for the group with a lower degree of duality and, therefore, the Japanee wage curve hold for that group. Howeve the wage-unemployment relationhip i revered if the hare of non-regular worker exceed approximately 30%. Therefore, the Japanee wage curve doe not hold for the group with a higher degree of duality, and alternative theorie uch a compenating differential may be ueful to explain thi group. Keyword: wage curve, dual labor market, threhold, non-regular worker JEL Claification Code: J30, R10 1. Introduction The purpoe of thi paper i to examine the wage curve for Japan. The wage curve how a negative relationhip between wage and unemployment rate; thi relationhip i conitent with the bargaining and efficiency wage theorie (Blanchflower and Owald, 1994). For example, the bargaining theory ugget that high unemployment reduce the bargaining power of trade union and prompt employee to accept lower wage. Thi relationhip ugget that a worker who i employed in a region of high unemployment earn le than doe an identical individual who work in a region of low unemployment. The lope coefficient of the wage curve can be interpreted a wage flexibility, which meaure the enitivity of wage to an exceive upply in a local labor market (Aixalá and Pelet, 2010). Blanchflower and Owald (1994) find that the unemployment elaticity of wage i 0.1 and i table for many countrie. Following Blanchflower and Owald (1994), many tudie have confirmed thi relationhip for more than 40 countrie. 1 Howeve little attention eem to have been paid to the impact of labor market duality on the wage curve. A the * inagaki@econ.nagoya-cu.ac.jp. Citation: Inagaki, K. (2015) Wage curve in dual labor market: cro-ectional evidence from Japan, Economic and Buine Letter, 4(2), For detail, ee Blanchflower and Owald (2005). Oviedo Univerity Pre 51 ISSN:

2 Wage curve in dual labor market in Japan inider-outider theory of the labor market ugget, unequal treatment of worker affect wage and unemployment. Therefore, the tructure of the wage curve i likely to depend on the degree of labor market duality. 2 To examine thi impact, the ue of Japanee data i helpful. Labor market duality in Japan ha increaed dramatically in recent year (OECD, 2013). The hare of non-regular worker ha increaed from 20% in 1994 to 35% in Howeve non-regular worker are confronted with everal problem, uch a lower wage, untable employment, poor working condition, and limited ocial inurance coverage. For example, the gap between the hourly wage of full-time and part-time worker i 31% for male and 48% for female. Similarly, about half of non-regular worker do not have ocial inurance coverage wherea almot all regular worker have the coverage. Furthermore, only 12.8% of non-regular worker pend more than ten year at the ame firm, compared to 49.4% of regular worker, and thi hort tenure reduce their opportunitie for job training. 4 The unequal treatment of regular and non-regular worker i one of the important characteritic of the Japanee labor market. Since bargaining power and employment protection in Japan are greater for regular worker than for non-regular worker (OECD, 2013), the wage-unemployment relationhip for regular worker i likely to be explained by the theorie behind the wage curve. Howeve the behavior of non-regular worker in Japan may not be well decribed by thee theorie, ince thee worker condition are much different from thoe of regular worker. In thi repect, there i room for further invetigation. Since labor market duality i alo ignificant in everal other developed countrie, our empirical analyi may provide ueful inight into the tructure of labor market in uch countrie. 2. Method In thi pape we conider the hare of non-regular worker a the degree of labor market duality in Japan for the reaon mentioned in Section 1. The unemployment elaticity of wage i the main focu of the literature on the wage curve. To examine the impact of labor market duality on the Japanee wage curve, we ue the threhold regreion technique developed by Hanen (2000) and aume that the unemployment elaticity of wage depend on the hare of non-regular worker. Thi aumption enable u to obtain information on the degree of labor market duality that ignificantly affect the tructure of the Japanee wage curve. We etimate the following model: W 1U I( N ) 1U I( N ) ( N ) I( N Z X, (1) W a r, 0 I 0 ) U, where, denote the real hourly wage; a, r, the unemployment rate; a, r, the hare of non-regular worker; Z a, r,, a vector of dummy variable; X a, r,, a vector of control variable; a,, an error term; and 1, 0, 1, 0,, and, the coefficient to be 2 Baltagi et al. (2012), Berg and Contrera (2004), and Ramo et al. (2010) are a few notable tudie focuing on the wage curve for informal worker. 3 In Japan, the number of non-regular worker i uually calculated by umming the number of part-time, temporary, dipatched, contract, entruted, and other non-regular employee. 4 For further detail, ee Jone and Uraawa (2011) and OECD (2013). 4(2), 51-56, N,

3 Wage curve in dual labor market in Japan etimated. The ubcript r, and repreent age group, region, and ex, repectively. All variable are natural logarithm. The indicator function I ( N a, r, ) take the value of 1 for N a, r, and 0 otherwie, and I N ) 1 I( N ). i the threhold parameter ued to plit the ample ( into two regime. Therefore, the ignificance of the threhold effect in Eq. 1 produce two etimate of the unemployment elaticity of wage ( 1, 1), depending on the degree of labor market duality. 5 The dummy variable in Eq. 1 include age group, region, and ex. 6 The control variable in Eq. 1 include educational attainment, length of ervice, the marriage rate, the hare of profeional and engineering worker, and the employment hare of the primary ector. 7 Following Hanen (1999), we aume that the coefficient on the control variable are the ame for each of the two regime. Let S n( ) be the um of quared error obtained from the ordinary leat quare etimation of Eq. 1 conditional on. Then, ˆ i defined a ˆ argmin Sn( ). (2) N a r, The obervation of the threhold variable, contitute the value of. Following Hanen (2000), we calculate the 95% confidence interval of from the likelihood ratio tatitic and it critical value. We then compute the p-value for the null hypothei of no threhold effect by uing the boottrap technique with 1,000 replication. 3. Data A Deller (2011) point out, the literature on the wage curve can be claified into two group: one group uing individual-level data (e.g., Baltagi et al., 2009, 2012, 2013; Berg and Contrer 2004; Ramo et al., 2010) and the other uing aggregate data uch a regional data (e.g., Delle 2011; Shilov and Mölle 2009). We ue aggregate data mainly becaue the data for individual regular and non-regular worker in Japan are not available for thi tudy. We obtain our cro-ectional data from the 2010 Japan Cenu and the 2010 Baic Survey on Wage Structure. 8 Japan ha 47 prefecture. In thi tudy, we divide each prefecture into nine age group (20 24, 25 30,, 60 64) and further divide them into male and female. Thu, the ample ize i 846. Annual earning are calculated by umming monthly contractual cah earning multiplied by 12 and annual pecial cah earning. Hour worked are calculated a the um of cheduled hour and overtime hour worked. We ue thee data and the regional difference index of conumer price to calculate real hourly wage. 9 We calculate the unemployment rate by dividing the number of unemployed people by the total number of people in the labor force. 5 For other iue uch a patial and dynamic effect, ee Baltagi et al. (2009, 2012) and Deller (2011). 6 If thee dummy variable are excluded from the threhold regreion, the main reult remain unchanged. 7 Further detail are provided in Section 3. 8 Panel data i very ueful in wage curve tudie. Howeve labor market duality in Japan ha increaed dramatically in recent year. Since the Japan Cenu i conducted every five year, the latet cro-ectional data eem mot ueful for thi tudy. 9 If we intead ue data on real annual earning, the following reult are unchanged. 4(2), 51-56,

4 Wage curve in dual labor market in Japan We obtain the number of non-regular worker by umming the number of part-time, temporary, dipatched, and other non-regular employee; thi i equal to the number of total employee minu the number of regular employee. Educational attainment i calculated a the hare of worker whoe educational level i higher than high chool. Profeional and engineering worker include individual engaged in technical work, uch a reearcher, engineer, and medical doctor (Major Group B of the Japan Standard Occupational Claification), and we ue the data for thoe worker whoe highet educational level i college, univerity, or graduate chool a a meaure of highly killed worker. 4. Reult A Table 1 how, the point etimate of i 0.305, and the 95% confidence interval i very narrow. Furthermore, the null hypothei of no threhold effect i rejected. Therefore, the tructure of the Japanee wage curve depend on the degree of labor market duality. Table 1. Tet for threhold effect (A) Threhold parameter Etimate 95% confidence interval [0.288, 0.307] (B) Tet for threhold effect Note: 0, ) and 0, ) ( 1 ( 1 Null hypothei 0 : Boottrap p-value Table 2 how the etimation reult of the Japanee wage curve with the threhold effect. 10 The unemployment elaticity of wage i for the regime in which N , and a, r, for the regime in which N The coefficient on the control variable have a, r, the expected ign. The number of obervation i 348 for the regime in which N Thi regime a, r, conit mainly of middle-aged male, and approximately 90% of them are regular worker. The Japanee wage curve hold for thi regime. Thi reult i conitent with the previou tudie that ue individual-level data for full-time worker (e.g., Baltagi et al., 2009, 2012). Howeve the Japanee wage curve doe not hold for the regime in which N , ince the unemployment elaticity of wage i poitive in thi regime. a, r, 10 The etimation reult of the coefficient on the dummy variable are not reported in thi paper for pace reaon but are available from the author upon requet. 11 To check for poible endogeneity in Eq. 1, we ue the intrumental variable (IV) technique. Following Aixalá and Pelet (2010), we aume that the unemployment rate i an endogenou variable and ue the lagged unemployment rate a an intrumental variable. Baed on the ame threhold effect a that reported in Table 1, the main reult remain eentially unchanged. The IV etimation reult are not reported in thi paper but are available from the author upon requet. 4(2), 51-56,

5 Wage curve in dual labor market in Japan Table 2. Etimation reult of Eq. 1 Variable Coefficient Standard error I ( N a, r, 0.305) 2.265** [0.145] I ( N a, r, 0.305) 2.668** [0.145] Unemployment rate I ( N a, r, ) 0.038* [0.016] Unemployment rate I ) 0.103** [0.017] ( N a, r, Educational attainment 0.107** [0.027] Length of ervice 0.368** [0.024] Marriage rate 0.044* [0.021] Share of profeional and engineering worker 0.027* [0.013] Employment hare of primary ector 0.025** [0.008] Note: The White heterokedaticity conitent tandard error are reported. The reult remain unchanged if we ue the uual leat quare tandard error. ** and * indicate ignificance at the 1% and 5% level, repectively. 5. Concluding remark We find that the wage-unemployment relationhip in Japan depend on the degree of labor market duality. The unemployment elaticity of wage i negative for the group with a lower degree of duality and, therefore, the Japanee wage curve hold for that group. Since thi group conit mainly of regular worker whoe bargaining power and employment protection are higher than that of non-regular worker, our reult are conitent with the economic theorie behind the wage curve. Howeve the wage-unemployment relationhip i revered if the hare of non-regular worker exceed approximately 30%. Therefore, the Japanee wage curve doe not hold for the group with a higher degree of duality. A poible explanation for thi reult i that labor mobility of non-regular worker i higher than that of regular worker. For example, 17.1% of non-regular worker have changed their job ince 2007 on annual average, compared to only 7.2% of regular worker. 12 Since the wage curve i derived from theorie that do not give weight to labor mobility (Shilov and Mölle 2009), thi regime may be explained by the theory of compenating differential (Harri and Todaro, 1970), which explicitly incorporate labor mobility and lead to a poitive wage-unemployment relationhip. 13 It i alo important to note that non-regular worker in Japan have little chance of achieving regular tatu, ince there i a high degree of employment protection for regular worker (Jone and Uraaw 2011). OECD (2013) pointed out that only about 10% of Japanee non-regular worker become regular worker. Thi fact i conitent with the inider-outider theory of the labor market. Almot all non-regular worker have fixed-term contract, and firm hire thee worker a a buffer to protect regular worker from dimial (Jone and Uraaw 2011). Therefore, non-regular worker are forced to change their job more frequently. For thi reaon, the theory of compenating differential eem to provide an explanation for the regime with a higher degree of duality. Finally, we dicu a further implication of the etimation reult of Eq. 1. The reult may be ued in aeing the effect of economic policie that aim to reduce labor market duality 12 Employment Statu Survey For further detail, ee Blanchflower and Owald (1994). 4(2), 51-56,

6 Wage curve in dual labor market in Japan in Japan. Given that problem uch a lower wage and employment intability for non-regular worker increae their labor mobility and, conequently, lead to the poitive wage-unemployment relationhip for the group with a higher degree of duality, the impact of labor market duality can be checked by etimating the unemployment elaticity of wage for the group. The ign and ignificance of the elaticity and probably the threhold effect reported in thi paper will be affected by improvement of labor market duality in Japan. Acknowledgement. I would like to thank Hideyuki Adachi, Yauyuki Oumi, Tamotu Nakamur Takahi Arai, Yautaka Tunehiro, Shin Imoto, and eminar participant at the Univerity of Hyogo and Onomichi City Univerity for their helpful comment. Furthermore, thi paper owe much to the comment and uggetion of the three anonymou referee. I am olely reponible for any error. Reference Aixalá, J. and Pelet, C. (2010) Wage curve veru Phillip curve: a microeconomic etimation for the Spanih cae, Análii Económico, 58, Baltagi, B.H., Başkay Y.S. and Hülagü, T. (2013) How different are the wage curve for formal and informal worker? Evidence from Turkey, Paper in Regional Science, 92, Baltagi, B.H., Blien U. and Wolf, K. (2009) New evidence on the dynamic wage curve for Wetern Germany: , Labour Economic, 16, Baltagi, B.H., Blien U. and Wolf, K. (2012) A dynamic patial panel data approach to the German wage curve, Economic Modelling, 29, Berg, J. and Contrer D. (2004) Political-economic regime and the wage curve: Evidence from Chile, , International Review of Applied Economic, 18, Blanchflowe D.G. and Owald, A.J. (1994) The Wage Curve, MIT Pre: Cambridge, MA. Blanchflowe D.G. and Owald, A.J. (2005) The Wage Curve Reloaded, NBER Working Paper 11338, National Bureau of Economic Reearch. Delle S. (2011) Spatial heterogeneity in the wage curve, Economic Letter, 113, Jone, R.S. and Uraaw S. (2011) Labour market reform in Japan to improve growth and equity. OECD Economic Department Working Paper 889, OECD Publihing: Pari. Hanen, B.E. (1999) Threhold effect in non-dynamic panel: etimation, teting, and inference, Journal of Econometric, 93, Hanen, B.E. (2000) Sample plitting and threhold etimation, Econometric 68, Harri, J.R. and Todaro, M.P. (1970) Migration, unemployment and development: a twoector analyi, American Economic Review, 60, OECD (2013) OECD Economic Survey of Japan, OECD Publihing: Pari. Ramo, R., Duque, J.C. and Surinach, J. (2010) I the wage curve formal or informal? Evidence for Colombi Economic Letter, 109, Shilov, A. and Mölle J. (2009) The wage curve in Ruia , Economic Letter, 102, (2), 51-56,