Deregulation and Labor Earnings: Three Motor Carrier Industries in Japan

Transcription

1 Deregulaton and Labor Earnngs: Three Motor Carrer Industres n Japan Akra Kawaguch Faculty of Polcy Studes, Doshsha Unversty Kezo Mzuno School of Busness Admnstraton, Kwanse Gakun Unversty May, 2007 Abstract Ths paper examnes the effect of product market deregulaton on labor earnngs. The dfference-n-dfference (DID) approach s used for an emprcal analyss wth the wage data n three motor carrer ndustres n Japan. Based on a theoretcal model that characterzes four regulatory regmes, we obtan the followng emprcal fndngs. Frst, the sharp wage declne of the drvers n regulated sectors n the 1990s and 2000s s partly explaned by deregulatons n transport ndustry and partly explaned by the deteroraton of the labor market. Second, we ensure that the rate deregulaton had a sgnfcant mpact on wage declne. Thrd, the entry deregulaton reduces wages only f the rate regulaton s not rgd. Keywords: product market deregulaton, labor earnngs, motor carrer ndustres. JEL classfcaton: J30, L51, L91. Correspondng author: Kezo Mzuno School of Busness Admnstraton, Kwanse Gakun Unversty Uegahara, Nshnomya, Hyogo , Japan. Tel: , Fax: , E-mal: kmzuno@kwanse.ac.jp 1

2 1 Introducton Deregulaton n a product market has effects not only on the product market tself but also on labor markets. Indeed, the recent enlargement of ncome nequalty n Japan, reported repeatedly n TV and newspapers, seems to have some relatonshp wth the wave of product market deregulaton that has rushed concurrently. 1 However, the exact mechansm concernng how product market deregulaton nduces ncome nequalty, especally a reducton n labor earnngs, s not so clear. Furthermore, there have not been many emprcal lteratures that gve the evdence for the effect of deregulaton on labor earnngs. The purpose of ths paper, then, s to clarfy the effect of product market deregulaton on labor earnngs by provdng some emprcal evdences wth a smple theoretcal model. In order to provde the emprcal evdences, we choose three motor carrer ndustres n Japan; taxes, buses, and trucks. There are two reasons for the choce of the ndustres. Frst, all the three ndustres have experenced product market deregulaton snce the 1990s. However, the strctness of the regulaton and the tmng of deregulaton are dfferent. It s of great nterest to compare the effect of deregulaton on wages of these drvers. Second, there are deal control groups: prvate car drvers and prvate truck drvers, whose wages were not drectly affected by deregulatons n transport ndustry. Prvate car drvers, such as courtesy bus drvers, transport passengers wthout any charge. On the other hand, prvate truck drvers are employed by owners of goods that are transported. Those drvers belong to varous ndustres such as mnng, manufacturng and wholesale ndustres, whch are not drectly affected by the deregulaton n transport ndustry. We apply the dfference-ndfference (DID) method to test the effect of deregulaton on wages of tax, bus and truck drvers usng prvate car drvers and prvate truck drvers as control groups. Several studes have examned the effect of product market deregulaton on labor earnngs. Peoples (1998) examned the deregulated four ndustres (truckng, ralroad, arlnes, and telecommuncatons), and concluded that the four ndustres took the dfferng approaches to reducng labor costs when facng deregulaton. In partcular, ndustry labor earnngs premums fell sharply n truckng, somewhat n arlnes, slghtly n telecommuncatons, and barely n ralroads. Focusng on US truckng ndustry, Hrsh and Macpherson (1998) found sgnfcant wage decreases followng deregulaton for unon for-hre drvers, whereas smaller wage 1 In the academc lterature, there are some debates about whether the ncome nequalty has been enlarged or not n Japan snce 1990s. See Ohtake (2005) for the detals. 2

3 declnes were found for non-unon for-hre drvers and for prvate carrage drvers. 2 The analyss we perform n ths paper has three dstnctve features. Frst, we treat the aspects of product market deregulaton n a more specfc way. In partcular, concernng entry regulaton, we deal wth not only drect entry regulaton but also ndrect entry regulaton, as wll be explaned n secton 3. Ths treatment has an analytcal mert, snce the deregulaton from the two regulatory regmes slghtly dffer each other. Furthermore, ntroducng rate regulaton n addton to entry regulaton, we explctly characterze the four regulatory regmes and apply the regmes to the real regulatory envronment n the three motor carrer ndustres. Ths analytcal approach makes the effect of each regulaton on labor earnngs clearer than the prevous studes. Second, n addton to entry regulaton and rate regulaton, we nclude another government nterventon, a government subsdy, whch cannot be gnored especally when consderng local passenger bus ndustres. 3 Thrd, we adopt the DID approach as an emprcal estmaton method. As s well-known, the DID approach s useful to examne the effect of polcy changes. 4 The data used for the estmaton comes from the Basc Survey on Wage Structure n Japan, where workers are categorzed by occupaton, age and experence. There are about 40 age-experence groups n each occupatonal group. Hence, we can obtan enough samples to examne the effect of deregulaton. In fact, as far as we know, our research s the frst attempt to study the ssue by usng the data n Japan. The man results obtaned n ths paper are as follows. Frst, the sharp wage declne of the drvers n regulated sectors n the 1990s and 2000s s partly explaned by deregulatons n transport ndustry and partly explaned by the deteroraton of the labor market. Second, we ensure that the rate deregulaton had a sgnfcant mpact on wage declne. Thrd, the entry deregulaton reduces wages only f the rate regulaton s not rgd. The outlne of ths paper s as follows. In the next secton, we brefly summarze some characterstcs of the three motor carrer ndustres n Japan and revew the process of product market deregulaton that occurred so far n the ndustres. Secton 3 provdes a theoretcal model that gves some predcton on the effects of entry and rate deregulaton. Secton 4 provdes the framework of the emprcal analyss, whose results are explaned n secton 5. Secton 6 concludes the paper. 2 Crémux (1996) and MacDonald and Cavalluzzo (1996) also reported the trend of wage declnes after deregulaton, whle they dealt wth the data n arlne ndustres and ralroad ndustres, respectvely. 3 Romly (2001) emphaszed the effect of government subsdy on local passenger bus fares n Brtan. He dd not, however, menton ts effect on labor earnngs. 4 Hrsh and Macpherson (1998) also used the DID approach. 3

4 2 Characterstcs of Three Motor Carrer Industres n Japan We brefly summarze the characterstcs of the three motor carrer ndustres (.e., taxs, buses, and trucks) n Japan. Fgure 1 llustrates ndexes of monthly wages of tax, bus and truck drvers, where monthly wages of prvate car drvers and prvate truck drvers are set to 100. The survey s mplemented n June of each year. The fgure shows that the wages of tax drvers are lower than the drvers n the other ndustres, and had sharply decreased n the 1990s. The wages of bus drvers had also decreased n the 1990, but the magntude s smaller than those of tax drvers. The wages of truck drvers are relatvely stable, though there was a small declne n the late 1990s. In these ndustres the bottom was n 1999, and wages have been recovered slghtly snce then. [Insert Fgure 1] Fgure 2 depcts the transport ndexes of the three ndustres. 5 Accordng to the fgure, the transport ndex of tax ndustry decreases rapdly snce 1993 (.e., the year of the so-called bubble economy burst ). Indeed, the magntude of ndex after 2000 s about three-fourth of that n On the other hand, the transport ndex of bus ndustry has been nearly constant, whle that of truck ndustry has been ncreasng. The transport ndex of truck ndustry n 2003 s twce as much as that n [Insert Fgure 2] Fgure 3 shows the annual ncreasng rate of corporatons n tax, chartered bus, local passenger bus and truck ndustres. In tax ndustry, prvately-owned companes are not ncluded. In spte of the sharp declne n the transport ndex of taxes, the number of corporaton dd not declne n the 1990s, and substantally ncreased n FY2002, FY2003 and FY2004. [Insert Fgure 3] Corporatons n bus ndustres have ncreased n the 1990s and 2000s n spte of the slght declne n the transport ndex of buses n the same perod. Corporatons n chartered bus 5 The ndexes are based on person-klometers and ton-klometers. The fgure n 1985 s set to

5 ndustry had steadly ncreased n the late 1980s and early 1990s at an average annual rate of fve percent, whch had been accelerated n the late 1990s. The annual ncrease rate exceeded 10 percent from FY1997 to FY2001. Local passenger bus ndustry experenced a steady ncrease n the number corporatons n the 1980s and 1990s at an average annual rate of two per cent. The ncrease rate exceeded fve percent n FY2002 and FY2003. Truck ndustry has experenced a qute stable ncrease n the number of corporatons snce The annual growth rate was about two percent n the 1980s and 2000s, whle t was slghtly hgher n the 1990s. The changes n the ncrease rates of the three ndustres are closely related to the tmng of the entry deregulatons as we wll dscuss later. 2.1 Tax ndustry As s well-known, tax ndustry s a labor-ntensve ndustry. The proporton of labor cost n total cost s about 80%, whch has not been changed n the last decade. The number of companes n 2005 s 55,126, whch conssts of 8,766 corporatons and 46,360 ndvdually-owned companes. The number of drvers s 428,253. The average age of drvers employed by corporatons s 52 years old, and that of owner-drvers s 59 years old. The total amount of revenues n the ndustry s about 2 trllon yen, whle a thrd of the companes experence defcts. 6 The pay system n tax ndustry s characterzed by a commsson system. Hence, wage dfferental among workers s large. In addton, the level of wage earnngs n the ndustry s lower than those n bus and truckng ndustres. Wage earnngs n the ndustry are also senstve to the busness cycle. The deregulaton n tax ndustry began n October 1993 when the dentcal fare n the same dstrct system was abolshed. Untl then, only the sngle tax fare system n each dstrct had been approved by the Mnstry of Transport, and no companes had been allowed to desgn the fare system by themselves. Interestngly, t s often clamed that the regulated tax fare had been hgher than the equlbrum fare determned by the market mechansm n order to protect the workng condtons and standards of lvng of drvers. 7 In May 1997, the regulaton on startng fare was relaxed, so that tax companes can choose ther own startng fare wthn a certan range. At the same tme, the mnmum vehcle holdng regulaton was relaxed, so that the entry to the tax ndustry becomes easy. In February 2002, the Road Transport Law was sgnfcantly amended. By ths amendment, the lcense to run a tax company has been removed, and was replaced by the entry 6 See the Mnstry of Land, Infrastructure and Transport (2005). 7 See Matsuzawa and Ishda (1993). Gärlng et al (1995) also nssted the same role of fare regulaton n the Swedsh tax-cab Industry. 5

6 approval system. Under the entry approval system, obtanng the permsson for entry was easer than before. At the same tme, companes became able to desgn ther own fare systems, as long as they notced them n advance to the Mnstry of Land, Infrastructure and Transport (MLIT), and ther fare systems satsfed the prce-celng stpulated by the Mnstry. Then, several new types of fare systems such as dscount fares and flat fares have been ntroduced. As Fgure 3 shows, the effect of the 1997 entry deregulaton on the number of corporatons s small, whle that of the 2002 deregulaton s remarkable. The number of tax corporatons ncreased by 4.7% n FY2002, 9.1% n FY2003 and 8.9% n FY Bus ndustry A bus ndustry conssts of chartered bus ndustry and local passenger bus ndustry, whose characterstcs dffer from each other. In chartered bus ndustry, there are about 3,600 companes. Most of them are small, and 47% of them experence defcts. The deregulaton began wth the relaxaton of the mnmum vehcle holdng regulaton n May Then, the Road Transport Law was amended n February In the amendment, the clause of supply and demand adjustment was removed, as well as the rate regulaton clause. As a result, the busness lcense system was replaced by the entry approval system. As Fgure 3 shows, the effect of ths deregulaton s substantal. Corporatons n ths ndustry ncreased by 14.6% n FY1997, 22.6% n FY2000 and 14.6% n FY2003. In local passenger bus ndustry, there are 500 companes, 72% of whch experence defcts. Frms n the ndustry has been recevng subsdy. The subsdy scheme was, however, revsed n 2001, and the amount of subsdy reduced. Furthermore, by the amendment of the Road Transport Law n February 2002, the busness lcense system was replaced by the entry approval regulaton. Also, the prce approval regulaton changed to the prce-celng approval regulaton. After the entry deregulaton, the number of corporatons ncreased by 7.8% n FY2002 and 5.4% n FY2003, as Fgure 3 shows. 2.3 Truck ndustry In ths ndustry, there are two gant companes (.e., Nhon Express Co. ltd. and Yamato Transport Co. ltd.), whle 99% of companes are small busnesses. The deregulaton began wth the establshment of two carrer laws (.e. the Motor Truck Transport Busness Law and the Freght Transport Management Law) n By the laws, the dstncton between route-busness and area-busness was abolshed, and the busness lcense system was replaced by the entry approval system. Moreover, the rate regulaton was eased such 6

7 that the approval system was replaced by the advance notfcaton system. Fgure 3 shows, however, that the effect of the entry deregulaton n truck ndustry s not large. The annual ncreasng rate of corporatons s about 3% n the 1990s, and t s stable. Moreover, as Sato (2004) ponted out, t was dffcult for the authortes to montor prces, snce there were so many small frms n truck ndustry. As a result, actual prces reflected the demand and supply of the market even before the deregulaton of prces. The rate regulaton and entry regulaton were almost totally abolshed n Aprl 2003 when three carrer laws (.e., the Motor Truck Transport Busness Law, the Freght Transport Management Law and the Ralway Busness Law) were amended. The busness dstrct regulaton and the advance notfcaton system were abolshed. However, ths deregulaton dd not accelerate the ncrease n the number of corporatons, as Fgure 3 shows. 3 A Theoretcal Model In ths secton, we provde a smple theoretcal model that llustrates the effect of deregulaton n the product market on the earnngs n the labor market. 8 As for product market regulaton, we examne three specfc types of government nterventon adopted n the three motor carrer ndustres n Japan; entry regulaton, rate regulaton, and government subsdes. The formulaton of these three types of government nterventon wll be explaned shortly. 3.1 The framework of the model There are an nfnte number of potental frms wth the same producton technology. Each frm could enter an ndustry and produce homogenous goods f t were proftable to do so. We consder a two-stage game. In the game, a frm frst ncurs some entry (sunk) cost when enterng the ndustry and then, once a frm enters, a manager and a labor unon of the frm bargans on wage and employment, takng the other frms barganng as gven. As s well-known, because of the exstence of the entry cost, the ndustry can be olgopolstc. 9 Let us descrbe the model specfcally. We assume that the nverse demand functon s gven by P ( Q) where Q s the ndustry output. That s, '( Q) < 0 P and Q = q 1 N 8 Haskel and Szymansk (1993) gave a smple model to examne the effect of prvatzaton and deregulaton on the performances n labor markets. In ther model, frms were monopolstcally compettve n product market, whereas ours are olgopolstc. They manly focused on the effect of the change n a frm s objectve and ts market power. 9 Dowrck (1989) frstly analysed the nteracton of olgopoly n a product market and unon-frm barganng. See also Nckel (1999) for the study on the nteracton of market power n a product market and the performance of labor unon. 7

8 where q s frm s output and N s the number of actve frms that have entered nto the ndustry n the frst stage. For analytcal smplcty, we assume that the relevant nput for producton s only labor, and each frm s producton functon s smply represented by employment. 10 Frm s proft s gven by ( Q) q ( w s) l E = ( P( L) ( w s) ) l E q = l where l s frm s π = P, (1) where L s the total employment n the ndustry (.e., L = l 1 ), w s frm s wage, s s a government subsdy per nput (or output), 11 and E s an entry cost. Frm s unon utlty s represented by U ( w u) l =, (2) where u s the expected alternatve ncome, whch depends postvely on the unemployment beneft, outsde wages and the dscount rate, and negatvely on the unemployment rate and the turn over rate. 12 For analytcal smplcty, we assume that the barganng power of the manager and that of the unon are the same. Wage and employment for frm are set to solve Max w, L ( π + E) U. (3) Note that the entry cost s already sunk n the barganng stage, so that the relevant proft of the manager at the barganng stage s π E. + To restrct our attenton to an nteror soluton for a frm s barganng problem, we assume the followng condton on the nverse demand functon. P Q P Q q. Assumpton 1 '( ) "( ) < 0 + Note that ths s a suffcent condton for the decreasng margnal revenue property that guarantees the concavty of a frm s proft n the standard Cournot olgopoly. 13 As stated above, we consder two-stage game. The tmng of the game s specfed as follows. Gven some types of government nterventon, the manager of each frm smultaneously determnes whether or not to enter the ndustry n the frst stage. In the second stage, once enterng the ndustry, a manager and a labor unon of each frm smultaneously N 10 As stated n the ntroducton, the three motor carrer ndustres are characterzed by hgh labor ntensty, so that ths smplfcaton can be justfed. 11 Here, we restrct our attenton to the subsdy per nput. Indeed, some local governments (e.g., Me prefecture n Japan, etc.) practced ths type of subsdy. 12 See Chapter 2 of Layard et al. (1991) for a detaled dscusson. 13 See Chapter 4 of Vves (1999) for a thorough treatment of the Cournot olgopoly model. 8

9 bargans on wage and employment, takng the other frms barganng as gven, and the market clears n that stage. We can derve a subgame perfect equlbrum of the game by a backward nducton argument. In the followng analyss, we examne the equlbra n four regulatory regmes: (1) drect entry regulaton wth no rate regulaton (Regme DER&NRR), (2) drect entry regulaton wth rate regulaton (Regme DER&RR), (3) ndrect entry regulaton wth no rate regulaton (Regme IER&NRR), and (4) ndrect entry regulaton wth rate regulaton (Regme IER&RR). 14 Here, we defne the drect entry regulaton regme as the regme where a regulator drectly controls the number of actve frms n the ndustry. Otherwse, we call the regme ndrect entry deregulaton, rrespectve of the degree of dffculty of frms entry caused by some types of regulaton. Two remarks on the entry (de)regulaton deserve to be mentoned. Frst, n ths paper, two types of entry deregulaton can be defned. The frst type of entry deregulaton s the transton from drect entry regulaton to ndrect entry regulaton, whle ts second type s defned by a decrease n the (admnstratve) entry sunk cost E n the ndrect entry regulaton. 15 Second, the equlbrum under the drect entry regulaton exactly corresponds to the equlbrum n the second stage of our model. On the other hand, the equlbrum under the ndrect entry regulaton corresponds to the equlbrum n the whole game. Smlarly, we defne a rate regulaton such that a regulator drectly controls the market prce n the ndustry. When there exsts a rate regulaton, we represent the regulated market prce by P The equlbra n the four regulatory regmes Let us derve the equlbrum n each of the four regulatory regmes. Regme DER&NRR At frst, we examne the equlbrum n the drect entry regulaton wth no rate regulaton regme. Ths s smply the equlbrum n the second stage of the model. In ths equlbrum, the number of actve frms n the ndustry s taken as gven, whereas the equlbrum market prce s endogenously determned. Suppose N frms have already been allowed to enter the ndustry by ncurrng the entry 14 We admt the government subsdy n all the four regulatory regmes. 15 In our sense, Blanchard and Gavazz (2003) examned only the second type of entry deregulaton. 16 The case where a regulator sets only the prce celng can be consdered ether as the rate regulaton regme or as the no-rate regulaton regme, dependng on whether the prce celng s bndng n the real polcy arena. 9

10 cost E. Then, each frm smultaneously sets wage and employment by takng the other frms barganng as gven. Takng the log of frm s objectve functon of (3) and substtutng the nverse demand functon nto t, we can rewrte the problem as follows. Maxln w, l ( P( L) ( w s) ) + 2lnl + ln( w u) Rearrangng the two frst-order condtons, we have the followng equatons. P '( L) l + 2( P( L) ( w s) ) = 0 ( L) u + s = w, (5) P + 2, =1,, N. (6) Solvng (6) for w and substtutng t nto (5), we have ( L) + P' ( L) l + ( s u) = 0 P. =1,, N. (7) We restrct our attenton to the symmetrc equlbrum; l ( N; s)( l =... = l ) (7) s rewrtten as 1 N (4). Then, P( Nl ) + P' ( Nl ) l + ( s u) = 0 Hence, from (6), the symmetrc equlbrum wage w ( N; s)( w =... = w ) expressed as follows. ( P( Nl ) + u + s). (8) s 1 w = 2 It s then easy to practce the comparatve statc exercses by usng (8) and (9). 17 Denotng the equlbrum output by Q Nq = Nl comparatve statcs on the equlbrum wage as a proposton., we report the results of 1 N (9) Proposton 1: In the equlbrum n the drect entry regulaton wth no rate regulaton regme (Regme DER&NRR), we have: () As the number of frms n the ndustry ncreases, the wage decreases. 18 () As the government subsdy decreases, the wage decreases (ncreases, respectvely) f and only f the elastcty of the slope of nverse demand functon s smaller (larger, respectvely) than one,.e., ( Q ) < ( > )1 δ where ( Q ) QP" ( Q) P' ( Q) δ. Proof See Appendx. 17 See Okuno-Fujwara and Suzumura (1993), whch carefully deals wth the comparatve statcs n a symmetrc Cournot olgopoly, as a reference of the analyss here. 18 For analytcal smplcty, we gnore the nteger problem when examnng the effect of changng the actve number of frms N. 10

11 An ncrease n the number of actve frms n the ndustry mples an ncrease n the ferceness of competton, so that the manager of each frm asks the labor unon to decrease the unt producton cost,.e. the wage (Proposton 1()). Proposton 1() states that the effect of subsdy on wage depends on the magntude of the elastcty of the slope of nverse demand functon, whch often plays a crucal role n the comparatve statcs n Cournot olgopoly. 19 Ths result s explaned as follows. A decrease n subsdy means an ncrease n unt producton cost. Then, the frm has an ncentve to decrease ts producton. Note that a small elastcty of slope of nverse demand functon (especally when t s less than one) mples a small ncrease n prce caused by a decrease n producton. 20 Ths mples a decrease n profts, so that the manager would lke to share ts loss wth the labor unon by decreasng ts wage payment. The opposte scenaro works n the case where the elastcty s large. Regme DER&RR Let us ntroduce the rate regulaton n addton to the entry regulaton. As stated above, the rate regulaton s represented by some predetermned level of the market prce, P, whch s stpulated by a regulator. In our Cournot competton framework, ths means each frm s output decson or employment decson s restrcted by the regulatory constrant, whereas ts wage decson s agan characterzed by (6). Then, n the symmetrc equlbrum wth the bndng rate regulaton, we have the followng condtons. (See Appendx for the dervaton.) * DR ( Nl ) P = P, (10) P * DR + u + s = 2w, (11) where DR l * and DR w * are each frm s employment and wage n the symmetrc equlbrum wth the regulated prce, P, respectvely. Eq.(10) unlaterally determnes l * DR, whle (11) determnes DR w * becomes as follows:. Hence, the equlbrum wage and employment under the rate regulaton l 1 * DR P 1 * P + s + u = ( P ) and w DR =, N 2 1 where P (). s the demand functon. The results of comparatve statcs are easly derved. 19 See Secton 4.3 of Vves (1999) on ths pont. 20 For example, when the nverse demand functon s concave,.e., P < 0, the slope elastcty s negatve. In ths case, a manager and a labor unon of a frm agree on a wage decrease. On the other hand, when the prce elastcty of demand s 1, whch mples no change n revenue when decreasng one unt of output, the slope elastcty s 2, so that they agree on a wage ncrease. 11

12 Proposton 2: In the equlbrum n the drect entry regulaton wth rate regulaton regme (Regme DER&RR), we have; () A change n the number of actve frms n the ndustry does not affect the wage. () As the government subsdy decreases, the wage decreases. () As the regulated prce decreases, the wage decreases. The reason that an ncrease n the number of actve frms n the ndustry does not affect the wage (Proposton 2()) s as follows. Notce that one of the relevant varables that affect the FOC wth respect to wage (.e. (11)) s the market prce P. Hence, even though the number of actve frms n the ndustry ncreases, the wage does not change as long as the predetermned market prce does not change. Regme IER&NRR Let us turn to the regme n whch ndrect entry regulaton s practced. At frst, we examne the case n whch rate regulaton s not practced. In ths regme, the regulator allows frms to enter the ndustry, whereas she may stll mpose some (regulatory) entry costs (e.g., admnstratve cost to enter the ndustry) on frms. Hence, the degree of entry deregulaton can be measured by the magntude of a decrease n entry cost E. In our model, ths regme corresponds to the subgame perfect equlbrum n the whole game. That s, gven an entry cost E and a government subsdy s, we can obtan the number of actve frms, whch s endogenously determned. We can use a backward nducton argument to derve the subgame perfect equlbrum n the whole game. The equlbrum n the second stage s the same as that n Regme DER&NRR, gven the number of actve frms that has been determned n the frst stage. Let us denote the * equlbrum wage and employment by w IN ( N s) ; and l ( N s) ( 1,..., N ) ; =, respectvely. Let us then examne the frst stage. Entry decson of a manager of frm depends on whether the proft, π, s nonnegatve or not. For analytcal smplcty, we gnore the nteger problem. Then, the addtonal equlbrum condton requred s the followng zero proft condton. N ( P( l j ( N; s) ) ( w ( N; s) s ) l ( N s) = E = ; j 1, =1,, N. (12) The endogenously determned varables n ths regme are not only the levels of employment and wage of each frm, but also the number of actve frms n the ndustry. They are characterzed by (8), (9), and (12). 12

13 Agan, we focus on the symmetrc equlbrum, whch we denote by { l w, N } Then, (12) becomes as follows. ( P( N l ( N ; s ) ( w ( N ; s) s ) l ( N ; s) = E,. Usng (13) and the results n Proposton 1, we have the comparatve statc results n ths regulatory regme. Denotng the equlbrum output by them as a proposton. Q Nq = Nl (13), we summarze Proposton 3: In the equlbrum n the ndrect entry regulaton wth no rate regulaton regme (Regme IER&NRR), we have; () A decrease n the entry cost decreases the wage. () As the government subsdy decreases, the wage decreases (ncreases. respectvely) f and only f the elastcty of the slope of nverse demand functon s negatve (postve, respectvely),.e., ( Q ) < ( > )0 δ. Proof See Appendx. It s apparent that, snce a decrease n the entry sunk cost nduces entry nto the ndustry, a labor unon has to accept a decrease n wage n order for ts frm to survve n a ferce envronment (Proposton 3()). We can use the same nterpretaton to explan () of Proposton 3 as the counterpart of Proposton 1. That s, when the elastcty of slope of nverse demand functon s small, prce ncrease s small, so that a frm s proft decreases. Hence, the manager would lke to share ts loss wth the labor unon by decreasng ts wage payment. Regme IER&RR Last, we examne the regme where entry s ndrectly controlled, whereas the rate s regulated. 21 * IR * IR * IR Let us denote the symmetrc equlbrum n ths regme by { l w, N },. Then, the symmetrc equlbrum wage and employment are dentcal wth those n Regme DER&RR; l * IR 1 1 * = P N P + s + u =. (14) 2 IR * IR * IR ( P ) l ( N; P ) and w w ( P, s) 21 One may thnk that, when the entry s allowed by a decrease n the entry cost, the rate regulaton s usually abolshed concurrently. Ths s just because the objectve of allowng entry s to enhance consumer welfare through a decrease n the market prce wth compettve pressure. Although ths may be the case n the real polcy arena, the fourth regme can occur n the theoretcal framework. Hence, we report the results n ths regme. 13

14 The addtonal equlbrum condton requred s the followng zero proft condton. * IR * IR * IR * IR * IR * IR ( P( N l ( N ; P ) ( w ( P, s) s ) l ( N ; P ) = E * IR * IR * IR Substtutng (14) nto (15) and solvng for { l w, N } equlbrum values as follows. ( P + u s) * 1 w IR = + 2, l * 2E IR = P u +, and s (15),, we can explctly obtan ther 1 P N IR = 2E. Hence, we have the followng results of the comparatve statcs. ( P )( P + s u) *. Proposton 4: In the equlbrum n the ndrect entry regulaton wth rate regulaton regme (Regme IER&RR), we have; () A decrease n the entry cost does not affect wages. () As the government subsdy decreases, wages decrease. () As the regulated prce decreases, wages decrease. The ntutons of the results n Proposton 4 are almost the same as those n Proposton 2. All the results stated n the above four propostons are summarzed n Table 1. [Insert Table 1] 4 Emprcal Analyss In ths secton, we frstly descrbe the characterstcs of the data used n the emprcal analyss. Then, we explan the emprcal model that examnes the effect of product market deregulaton on labor earnngs. 4.1 The data The data used for the followng estmaton s from the Basc Survey on Wage Structure. The mean wages of worker groups that are categorzed by occupaton, age and experence are avalable form the publshed tables. Workers are dvded nto 129 occupatonal groups, each occupatonal group s dvded nto 12 age groups: aged 15-17, 18-19, 20-24, 25-29, 30-34, 35-39, 40-44, 45-49, 50-54, 55-59, 60-64, and 65 years old and over. Each occupaton-age group s further dvded nto 5 experence groups: less than 1 year, 1-4, 5-9, 10-14, and 15 years and over. Therefore, potentally 60 (=12*5) observatons are avalable for each occupatonal group. Some of the age-experence groups, however, do not exst. For example, workers aged 14

15 18-20 years old who have years of job experence do not exst. Hence, about 40 observatons are avalable for each occupatonal group per year. We pcked up sx occupatonal groups for the emprcal study tax drvers, busness bus drvers, small-szed busness truck drvers, large-szed busness truck drvers, prvate car drvers and prvate truck dvers. The last two groups are control groups. A large part of prvate truck drvers belong to constructon ndustry, whle prvate car drvers transport passengers wthout any charge, such as courtesy bus drvers. They are not drectly affected by deregulatons n transport ndustry. 4.2 The emprcal specfcaton We use the DID approach that s useful to examne the effect of polcy changes (.e., product market deregulaton n ths paper). In the DID approach, we measure the effect of deregulaton by the dfference n wages between the treatment group and the control group. In the case where ndvdual drvers panel data were avalable, the followng model would be approprate for the estmaton. ln w = α + β + δx + γd + u, t t t t t where w t s the log of hourly wages of ndvdual at year t, α s a varable that represents the ndvdual s characterstcs that do not change over tme, β t s a year specfc effect that commonly affect wages of all drvers, and X t s a vector of explanatory varables that reflect ndvdual characterstcs that changes over tme, such as tenure, a number of chldren, and resdental area. d t s a dummy varable that capture polcy changes, whch s defned as follows: d = 1 f a certan type of product market deregulaton has been mplemented at t or t before t n the ndustry where ndvdual belong to, and d t = 0 otherwse. u t s the other factors that affect wages. Our emprcal model s slghtly dfferent from the standard one. Snce we use grouped data that s aggregated accordng to (job category) (age) (job experence), subscrpts should be nterpreted as a group ndcator. The estmaton perod s set to ten years, of whch eght years are before the polcy change and two years are after the polcy change. Fgure 4 llustrates the tmng of polcy change and the wage survey. The wage survey s mplemented every June. If the government changed ts polcy between June of year τ 1 and June of year τ, we estmate wages of the treatment group and the control group from June of year τ 8 to year τ + 1. The star on Fgure 4 ndcates that a polcy change occurred, and the dashed lne mples that the treatment group s 15

16 affected by the polcy change n ths perod. The polcy change dummy takes one for the treatment group n year τ and τ + 1, and zero otherwse. We measure the effect of polcy change only up to two years after the polcy change, because the effect wll dmnsh as workers shft from the low wage ndustry to a hgh wage ndustry. Fgure 4. Polcy Change and Estmaton Perod τ 8 τ 1 τ τ + 1 Treatment Group τ 8 τ 1 τ τ + 1 Control Group 4.3 Predcted Effects Tax We wll estmate the effect of nne polcy changes n four ndustres as s summarzed n Table 2. The deregulaton of tax ndustry n 1993 abolshed the dentcal fare n the same dstrct system. As a result, tax fare declned. Hence, we can apply the result of Proposton 2() n our theoretcal framework. That s, we predct that the wages of tax drvers wll decrease. [Insert Table 2] In May 1997, the mnmum vehcle holdng regulaton was relaxed, whch ncreases the number of frms n our theoretcal framework. Accordng to Proposton 2(), wages do not change n response to the ncrease n the number of frms. However, at the same tme, the startng fare regulaton was relaxed, whch decreases the fare. Hence, we can apply Proposton 2(), and predct that the wages of tax drvers wll decrease. In February 2002, the busness lcense system was replaced by the entry approval system. Ths mples that the entry regme was shfted from the DER to the IER. If the DER was effectve before the deregulaton, ths regme shft has the same effect as the ncrease n the number of frms under the DER regme. At the same tme, the prce approval regulaton was replaced by the prce-celng approval system. If the prce-celng s not bndng any more, the 16

17 regme s regarded as the NRR. Ths regme shft has the same effect as the decrease n the regulated prces under the RR regme. We should apply Proposton 1() for the predcton of the entry deregulaton, snce the new regme of rate regulaton s NRR. Smlarly, Proposton 4() should be appled to predct the effect of the rate deregulaton, snce the new regme of entry regulaton s IER. Therefore, we predct that wages wll decrease by the 2002 deregulaton. Chartered Bus In chartered bus ndustry, the mnmum vehcle holdng regulaton was relaxed n May Ths deregulaton had a large mpact on the number of frms n the ndustry, as Fgure 3 shows. Under the RR regme, however, there would be no effect on wages f the fare were fxed, as Proposton 2() states. In February 2000, the busness lcense system was abolshed, and entry approval system was ntroduced n February Ths mples that the entry regulaton regme was shfted from the DER to the IER. At the same tme, the prce range approval regulaton was replaced by the advance prce notfcaton system. The latter regulaton s so weak that we can consder that the rate regulaton was abolshed at that tme. That s, the rate regulaton regme was shfted from the RR to the NRR. We should apply Proposton 1() to predct the effect of entry deregulaton, snce the new regme of rate regulaton s NRR. Smlarly, Proposton 4() should be appled to predct the effect of the rate deregulaton, snce the new regme of entry regulaton s IER. Therefore, we predct that wages wll decrease by the 2000 deregulaton. Local Passenger Bus In local passenger bus ndustry, the subsdy to bus companes was reduced n Aprl The specfc subsdes to local passenger bus companes were extngushed 22. Ths wll have a negatve effect on wages by Proposton 2(). The busness lcense system was replaced by the entry approval system, and the prce approval system was replaced by the advance notfcaton under prce-celng approval n February If the prce-celng s not bndng, the regme s regarded as the NRR. Ths regme shft has the same effect as the decrease n the regulated prces under the RR. The wages wll decrease by Proposton 4(). The entry deregulaton also has a negatve effect on wages by Proposton 1(). 22 The fares of local passenger buses have not been changed from 2000 to The average fare of local passenger bus durng ths perod was yen per klometer and stable. See Mnstry of Land, Infrastructure and Transport (2005). 17

18 Truck Fnally, n truck ndustry, the busness lcense system was abolshed n December 1990, and replaced by the entry approval system. We consder the regme of entry regulaton vrtually shfted from the DER to the IER. At the same tme, the prce approval regulaton was replaced by the prce notfcaton regulaton. We consder the regme shfted from the RR to the NRR. The predcted effect on wages s negatve by Proposton 1() and Proposton 4(). The busness dstrct regulaton was abolshed n Aprl 2003, whch reduces the entry cost. We predct that wages declne by Proposton 3(). At the same tme, the prce notfcaton regulaton was abolshed. Ths would have lttle effect on wages, snce the prce notfcaton regulaton had not been an effectve regulaton. As we have dscussed above, however, the 1990 and 2003 entry deregulatons dd not substantally ncrease the number of frms n truck ndustry. Moreover, the rate regulaton was not rgd even before the deregulatons because there were so many small companes n ths ndustry and t s dffcult for the authortes to montor the fare. 23 Hence, the effect of these deregulatons on wages wll be weak. 5 Emprcal Results Tax Table 3 shows the effect of deregulaton n tax ndustry. The table gves the coeffcent of polcy dummy varable, γ. Group specfc effects and year specfc effects are not lsted on the table. [Insert Table 3] In models (1) and (6), the effect of the 1993 deregulaton on wages of tax drvers s examned wth prvate car drvers and prvate truck drvers as control groups. The wages of tax and prvate car drvers from 1986 to 1995 are used for the estmaton. d = 1 s assgned to the tax drvers n 1994 and Ths s because the 1993 deregulaton was mplemented n October 1993, whereas the Basc Survey on Wage Structure s mplemented n every June. We apply the weghted least squares method for the estmaton. That s, varables are weghted by the number of drvers n each group. We can see that the hourly wages of tax drvers decreased by about 11%, f we do not take nto account the effect of unemployment rate. t 23 Sato (2004) ponted out ths. 18

19 Model (6) controls the effects of unemployment rate that may dffer among the ndustres. The theory predct the unemployment rate has a negatve effect on wages, snce t reduce the threat pont of unons. Note, however, that the cross term of tax drver dummy and unemployment rate captures the relatve effect of unemployment rate on tax drvers wages compared to that on the wages of prvate truck drvers. The sgn of the coeffcent wll be negatve (postve) f wages of tax drvers are more senstve (less senstve) to the labor market condtons than those of prvate truck drvers. The estmaton result shows that the coeffcent of the cross term of tax drvers and unemployment rate s not sgnfcantly dfferent from zero. The effect of the deregulaton on wages s margnally smaller f we control the effect of the unemployment rate. Models (2), (3), (7) and (8) examne the effect of the 1997 deregulaton. Models (2) and (7) ncludes dummy varables for the 1997 deregulaton only, but models (3) and (8) nclude those for the 1993 deregulaton as well. Model (2) shows that the 1997 deregulaton decreased hourly wages by 19%. The effect becomes 9% f we control the effect of unemployment rate. The coeffcent of the cross term of tax drver dummy and unemployment rate s strongly negatve. Ths mples that the tax drvers wages are more senstve to the unemployment rate than those of prvate truck drvers n ths perod. The magntude of the polcy effects on wages becomes 13% and 10%, respectvely f we adjust the effect of the 1993 deregulaton. Model (4), (5), (9) and (10) examne the effect of the 2002 deregulaton. Models (4) and (9) ncludes dummy varables for the 2002 deregulaton only, but models (5) and (10) nclude those for the 1997 deregulaton as well. Models (4) and (9) show that the 2002 deregulaton decreased hourly wages of tax drvers by 9% and 7%, respectvely. The magntude of the effects declnes to 2% f we take nto account the effect of the 1993 deregulaton. The effect becomes postve f we control the effect of unemployment rate. Ths contradcts to the theory. Ths s because the unemployment rate was at ts peak n 2002 and stayed at the hgh level n 2003, and the whole wage declne was explaned by the ncrease n the unemployment rate. In sum, the 1993 and 1997 deregulaton had a sgnfcant mpact on a decrease n the wages of tax drvers, but the 2002 deregulaton dd not. Bus Table 4 shows the effect of deregulaton on wages of bus drvers. As we mentoned before, the regulaton for chartered buses was relaxed n 1997 and 2000, whle that for local passenger buses was relaxed n 2001 and Unfortunately, bus drvers n our data were not dvded nto the two bus servces. Therefore, the effect of product market deregulaton on the two bus ndustres could be mxed up. In addton, snce the deregulaton was mplemented n a row, t s 19

20 dffcult to measure the effect of each deregulaton separately. [Insert Table 4] Model (1) shows that the 1997 deregulaton reduced hourly wages and monthly wages by 5%. Model (8) shows, however, the effect becomes almost zero f we take nto account the effect of the unemployment rate, as predcted by the theory. Model (2) mples that the 2000 deregulaton reduced hourly wages by 8%, but the effect declnes to 2% and nsgnfcant f the unemployment rate s controlled. The magntude becomes to 5% and 2% respectvely f we take nto account the 1997 deregulaton. Model (4) shows that 2001 deregulaton for local passenger bus ndustry decreased hourly wages by 4%. The effect s almost zero f the unemployment rate s controlled. Models (5) and (12) contan the dummy varables for the 1997 and 2000 deregulatons. The dummy for the 2001 deregulaton s not ncluded to avod multcollnearty. The 2000 deregulaton dummy varable captures the effect of the seres of deregulatons from 2000 to Model (5) shows that the deregulaton reduced bus drvers wages by 5%. The effect s margnally smaller n model (8) that controls unemployment. The results are smlar n models (6), (7), (13) and (14). The seres of deregulatons for the bus ndustry from 2000 to 2002 reduced wages by 5%. In sum, the 1997 deregulaton dd not have a sgnfcant mpact on wages of bus drvers. On the other hand, the seres of deregulatons from 2000 to 2002 dd have a negatve effect on ther wages, though t s dffcult to dentfy the effects of each deregulaton polcy. Truck Table 5 shows the effect of deregulatons on wages of truck drvers. The estmaton perod of models (1) and (3) s nne years, rather than ten years, because the data of 1982 s not avalable. There are two types of truck drvers n our data base: large-szed truck drvers and small-szed truck drvers. We assume that the effects of deregulatons on the wages of these two types of drvers are the same. [Insert Table 5] The results show that deregulatons have lttle mpact on wages of truck drvers. Ths result could be explaned from the followng two reasons. Frst, the entry barrers n truck ndustres had already been suffcently low before In fact, there were more than 40,000 frms n 20

21 1990, most of whch are very small. As Fgure 3 shows, the annual ncrease rate of frms n truck ndustry s relatvely stable. There s no evdence that the 1990 and 2003 deregulatons ncreased market entry. Second, the prce regulaton n truck ndustry had not been as effectve as those n tax and bus ndustres even before the 1990 deregulaton. There were so many small frms n truck ndustry that t was dffcult for the authorty to montor prces that are determned by the negotatons between carrers and cargo owners. As a result, actual prces had reflected the demand and supply of the market even before the 1990 deregulaton. 6 Concludng Remarks Ths paper examned the effect of product market deregulaton on labor earnngs. In order to predct ts effect, we provded a smple theoretcal model. We then characterzed four regulatory regmes. Based on the theoretcal model, we obtaned the followng emprcal fndngs wth the use of the DID approach (See the left two columns of Table 2 for the summary of estmaton results). Frst, we ensured that both the entry and rate deregulatons had a sgnfcantly negatve mpact on wages of tax drvers except for those of the 2002 deregulaton. The 2002 deregulaton also has a sgnfcantly negatve effect on ther wages f we do not control the effect of the unemployment rate. Second, the entry deregulaton n chartered bus ndustry dd not have any effect on wages. Ths s consstent wth theoretcal predcton that the entry deregulaton does not reduce wages f the rate s rgdly regulated. Thrd, the seres of deregulatons n bus ndustry from 2000 to 2002 had a sgnfcant negatve effect on wages even f we take nto account the effect of unemployment rate. It s dffcult, however, to dstngush the effect of each deregulaton polcy. Fourth, the deregulatons n truck ndustry dd not have sgnfcant effect on wages. Ths s because both entry regulaton and rate regulaton n ths ndustry had been weak. The entry deregulaton had lttle effect on the number of frms, and the prce had been set by negotatons between carrers and cargo owners even before the deregulatons. Ffth, n many cases, the effect of deregulaton on wages substantally declned as the effect of unemployment was controlled. Sometmes, t totally dsappeared. Ths mples that the sharp wage declne n the drvers n regulated sectors n the 1990s and 2000s s partly explaned by the deteroraton of labor market condtons. That s, ther wages are more senstve to labor market condtons than drvers n other ndustres. 21

22 Needless to say, product market deregulaton has an mpact not only on wages but also on the level of employment. We could not, however, collect relevant data for the estmaton of the level of employment. These are left as tasks of the future. Appendx [1] Proof of Proposton 1 () Dfferentatng (8) wth respect to N and rearrangng terms, we have 2 [( N + 1) P' + Nl P" ] + [ P' l + P" l ] = 0 l N Hence, from Assumpton 1, we have l N = l ( N + 1) [ P' + P" l ] P' + Nl. (a1) < 0. (a2) P" Smlarly, dfferentatng (9) wth respect to N, we have w P' l = l + N N. (a3) 2 N Substtutng (a2) nto (a3) and rearrangng terms, we have w N = 1 2 ( N + 1) 2 l P' P' + Nl < 0. (a4) P" () Dfferentatng (8) wth respect to s and rearrangng terms, we have l s = ( N + 1) 1 P' + Nl > 0. (a5) P" Then, dfferentatng (9) wth respect to s and substtutng (a5) nto t, we have w s * P' l = P' N 2 N = 1 2 P' ( N + 1) where ( Q) QP" P' DN ( 1 δ ( Q ) P' + Nl + 1, P" δ s the elastcty of the slope of nverse demand functon and (a6) Q Nq = Nl s the total output n the equlbrum. Hence, we obtan the result stated n the text. Q.E.D. 22

23 [2] The dervaton of the symmetrc equlbrum n the drect entry regulaton wth rate regulaton regme (Regme D-R) Let us focus on the symmetrc equlbrum. Consder frm s problem when the rate regulaton P s mposed and all the other frms except frm sets frm s problem s formulated as follows. Maxln w, L * DR ( P( l + ( N 1) L ) ( w s) ) + 2lnl + ln( w u) * DR s. t. P( l + ( N 1) l ) P The assocated Lagrangean s as follows. DR l * * DR ( P( l + ( N 1) l ) ( w s) ) + 2lnl + ln( w u) + P P l + ( N 1) and DR w * * [ ( l )] DR Ω = ln λ.. Then, Note that respect to w does not enter n the constrant, whch mples that the frst-order condton wth w s not affected by the constrant. Then, rewrtng the constrant and the frst-order condton wth respect to the text. w gves the characterzaton of the symmetrc equlbrum gven n [3] Proof of Proposton 3 () It s apparent from (13) that a decrease n E nduces entry n the ndustry (.e., an ncrease n the number of actve frms) n the frst stage of the game. Hence, the equlbrum wage decreases as E decreases, snce an ncrease n the number of actve frms decreases the equlbrum wage (see (a4)). () Dfferentatng (13) wth respect to s and rearrangng terms, we have * N s IN * P' l = P' l IN + N + N l w l s + 1 s l w * l N N IN + + [ P ( w s) ] [ P ( w s) ] * l s * l N Substtutng (a2), (a4), (a5) and (a6) nto (a9) and rearrangng the terms, we have N s = 2 ( P' + P" l ) l 2E * l IN 2 * 2E( P' + P" l IN ) 2 * [ P' ] l IN Then, usng (a10) and (a6), we have dw ( N () s ; s) ds w = N = N N s * w + s P' Eδ IN IN IN. (a9). (a10) ( Q ) 2E( P' + l P" ) [ ' ]. 2 2 P l (a11) 23

24 From Assumpton 1, the denomnator s postve, so that we obtan a result stated n the text. Q.E.D. 24

25 References [1] Blanchard, O. and F. Gavazz (2003), Macroeconomc Effects of Regulaton and Deregulaton n Goods and labor Markets, Quarterly Journal of Economcs, 108, [2] Crémux, P. (1996), The Effect of Deregulaton on Employee Earnngs: Plots, Flght Attendants, and Mechancs, , Industral and Labor Relatons Revew, 49, [3] Dowrck, S. (1989), Unon-Olgopoly Barganng, Economc Journal, 99, [4] Gärlng, T., T. Latla, A. Marell, and K. Westn (1995), A Note on the Short-Term Effects of Deregulaton of the Swedsh Tax-cab Industry, Journal of Transport Economcs and Polcy, 30, [5] Haskel, J. and S. Szymansk (1993), Prvatzaton, Lberalzaton, Wage and Employment : Theory and Evdence for the UK, Economca, 60, [6] Hrsh, B. T. and D. A. Macpherson (1998), Earnngs and Employment n Truckng: Deregulatng a Naturally Compettve Industry, n Regulatory Reform and Labor Markets, Boston, MA: Kluwer Academc Publshers. [7] Layard, R, S. Nckel and R. Jackman (1991) Unemployment: Macroeconomc Performance and the Labour Market, New York, Oxford Unversty Press. [8] MacDonald, J.M. and L.C. Cavalluzzo (1996), Ralroad Deregulaton: Prcng Reforms, Shpper Responses, and the Effects on Labor, Industral and Labor Relatons Revew, 50, [9] Matsuzawa T. and N. Ishda (1993), Motor Transports (n Japanese) n Y. Kanemoto and H. Yamauch, eds. Transportaton System, Tokyo: NTT Publshers. [10] Mnstry of Land, Infrastructure and Transport (2005), Motors n Data (n Japanese), Tokyo: Automoble Busness Assocaton of Japan. [11] Nckell, S. (1999), Product Markets and Labour Markets, Labour Economcs, 6, [12] Ohtake, F. (2005), Inequalty n Japan (n Japanese), Tokyo: Nhon Keza Shnbunsya. [13] Okuno-Fujwara, M. and K. Suzumura (1993), Symmetrc Cournot Olgopoly and Economc Welfare: A Synthess, Economc Theory, 3, [14] Peoples, J. (1998), Deregulaton and the Labor Market, Journal of Economc Lterature, 12, No.3, [15] Romly, P. (2001), Subsdy and Local Bus Servce Deregulaton n Brtan: A Re-evaluaton, Journal of Transport Economcs and Polcy, 35, [16] Sato, M. (2004), Deregulaton and Structure of the Truckng Industry (n Japanese), IATSS Revew, 29, [17] Takeuch, K. and K. Terada (1993), Motor Freght Transports (n Japanese) n Y. 25