Working Paper No Labor Taxes, Employment and Growth: Chile by

Transcription

1 Woring Paper No. 131 Labor Taxes, Employmen and Growh: Chile by Raphael Bergoeing + Felipe Morandé * March 2002 Sanford Universiy John A. and Cynhia Fry Gunn Building 366 Galvez Sree Sanford, CA Cener for Applied Economics, Universiy of Chile * Professor of Economics, Universiy of Chile, former Chief Economis, Cenral Ban of Chile, and Visiing Fellow, CREDPR, Sanford Universiy (Winer )

2 Labor Taxes, Employmen and Growh: Chile Raphael Bergoeing Cener for Applied Economics, Universiy of Chile Felipe Morandé Universiy of Chile and CREDPR, Sanford Universiy March 2002 Absrac The per capia growh rae of Chile from 1984 o 1997 was among he highes in he world. During he las hree years however, per capia growh dropped significanly. This paper discusses he role of facor accumulaion and facor produciviy growh wihin a simple neoclassical growh framewor o explain he evoluion of oupu in Chile for he pas 20 years, including he deline in he rae of growh. In conras o he experience of he 1980s and early 1990s, in he las hree years he driving force has no been oal facor produciviy growh (or he lac hereof), bu a severe fall in employmen. This fall, in urn, could be due o a higher hiring cos of labor associaed wih a labor code reform, which was debaed during he period. Using a calibraed dynamic general equilibrium we conclude ha a 6.75% ax on labor hiring replicaes he observed fall in employmen. 1 The auhors are graeful for he efficien assisance of Facundo Piguillen. This paper was finished while Morandé was a Visiing Fellow a CREDPR, Sanford Universiy, during he Winer Quarer of

3 1. Inroducion From 1984 unil a couple of years ago, he Chilean economy was growing a a rae of 5.4% per capia, placing i among he world s mos successful economies in he pas weny years. The growh can undoubedly be aribued o he liberally oriened srucural reforms ha oo place during he 1970s and 1980s, and coninued ino he early 1990s. A firs, however, his pah was far from easy. In fac, he subsanial growh since 1984 began wih a profound crisis in he early 1980s ha led o an accumulaed decline in per capia oupu of around 20% for Chile hen grew seadily, and in 1990 i had recovered o is rend level. 3 In he years ha followed, he growh rae held seadily a around 6%, bringing oupu per capia 30% higher han is 1980 rend level by In he pas hree years, however, he growh of he Chilean economy has dropped. From 1998 o 2001, he per capia growh rae averaged around 1.2% per year. Differen hypoheses have arisen o explain his period of sagnan growh. In paricular, analyss spea of exernal facors associaed wih he decline in he erms of rade and reduced access o exernal capial flows. The recen recession affecing he world economy (which furher deerioraed afer he Sepember 11 h erroris aac) has also conribued o worsening he ouloo for he erms of rade and dampening invesor s appeie for ris. Ohers argue ha his fall could be he resul of he excessively resricive moneary policy sance adoped by he Cenral Ban since mid-1998 o miigae he impac of he Asian crisis emerging a he ime. The effecs of his policy, combined wih he direc impac of he Asian crisis iself, may have proven more lasing and harder o urn around han originally foreseen, even wih he openly expansive moneary approach applied for several quarers ill now. Meanwhile, ohers have argued ha he counry s difficulies in achieving growh raes of he pas decade go beyond he explanaions of a normal business cycle. Moreover, hey sugges ha recen resuls reveal a decline in he Chilean economy s poenial o grow a more han 4% annually. Furhermore, he economy has proven unable o creae new jobs a raes comparable o hose for oupu. Boh phenomena sagnan growh and lile job creaion have coincided no only wih a complex exernal environmen for emerging economies bu also wih a range of policy proposals affecing producion, some of which already implemened in he form of legal reforms. Among hese, he so-called labor reform sands ou. Discussions began in he poliical arena during he presidenial elecion in lae 1999 and i oo almos wo years of parliamenary debae before he law was finally passed in Ocober This debae concluded ha he reform would raise he cos of hiring labor. Earlier, owards he end of 2000, reforms o reduce ax evasion were passed, and a lile laer, in mid-2001, reforms o reduce he ax burden on individuals bu gradually increase he burden on companies were also approved. In an opposie direcion, in lae 2001 reforms o liberalize he capial mare were also passed, which should reduce invesmen and capial coss in he fuure. This aricle focuses on he hird of hese hypoheses, ha is, he decline in growh and problems wih job creaion are lined o changing producion coss, mainly higher labor coss. Wih Bergoeing e al. (2002) as our saring poin, we analyze he role of facor accumulaion and efficiency in uilizaion during he pas 20 years in Chile o undersand oupu rends in he conex of a simple neoclassical growh. The analysis suggess ha during he crisis in he early 1980s and he subsequen recovery and growh phase hrough 1998, he efficiency of facor uilizaion was he main deermining facor of economic aciviy. In addiion, he pas hree years, he fall in employmen has been he main elemen behind declining growh. 2 The Chilean crisis of is considered one of he wors in he 20 h cenury. Kehoe and Presco (2002) provide evidence for his. 3 For he purposes of his sudy we will use oupu per populaion of woring age o analyze growh processes in he Chilean economy and he 2% annual rae as rend. Oupu per populaion of woring age, ha is, populaion from 16 o 64 years of age, is he appropriae indicaor for per capia oupu in he conex of he neoclassical growh used in his sudy. Finally, he 2% rae used as a proxy for rend growh corresponds o average annual growh in his variable from in Chile. 3

4 2. Growh Accouning In he conex of he neoclassical, slower growh can be due o a decline in labor facor accumulaion, or changes in implici or explici axes ha mae labor more expensive o hire and affec he relaive prices of producion facors. Evidence in Kehoe and Presco (2002) shows ha mos crises during he 20 h cenury were he consequence of falling efficiency of facor uilizaion or labor conribuion. In Chile from 1981 o 1998, he main source of growh was he efficiency wih which labor and capial were used; since hen, rends in aciviy levels have resuled fundamenally from changes in employmen. To deermine he conribuion of facor accumulaion and how efficienly he facors are used o he changes in oupu per woring age populaion, we brea down he changes in he laer by changes in oal facor produciviy (TFP), he capial o oupu raio, and hours wored per person of woring age. This brea down is based on a Cobb-Douglas aggregae producion funcion, Y = A K α L 1-α (1) where Y is oupu, K is capial, L oal hours wored and A oal facor produciviy. In his conex, A = Y / (K α L 1- α ) (2) Labor and oupu series are direcly available from naional accouns. To obain A, however, we mus choose a value for he capial share in oupu, α, and generae aggregae capial series, K. Informaion from naional accouns indicaes ha he labor compensaion share of Chile s oupu is almos This, in a compeiive conex, corresponds o 1-α, so he capial share is This fracion is sable over ime and among many developing counries. In developed counries, however, labor s share is much higher, wih α flucuaing a around Gollin (2002) shows ha if we correc for labor s share in developing counries, o allow for he fac ha independen worers and heir families are underesimaed, labor s conribuion rises significanly, converging o levels observed in developed counries, ha is A second reason for using his labor s share and no he informaion from naional accouns is ha in his case, he growh predics a marginal produciviy for capial ha is unrealisically high. 4 In any case, and as he sensiiviy exercise ha appears in Appendix 2 shows, he resuls of his sudy would no be subsanially differen if we assumed ha α is close o he value arising from naional accouns, for example, The fracion of oupu accorded o he labor facor only affecs he disribuion of changes in oupu beween TFP and capial, bu does no affec he labor facor s conribuion, which is he main elemen behind oupu s behavior in he pas hree years. Because of his, and given ha his aricle ceners precisely on he changes o producion coss due o legal reforms, paricularly o labor laws, from now on we assume ha α = 0.3. Using logarihms for he producion funcion, we have: Y 1 α K L = + + log log A log log N 1 α 1 α Y N (3) 4 If α=0.45, for example, he before-ax rae of reurn on capial would average 23% from Wih α=0.30, however, his rae is 15%. 4

5 where L / N is he number of hours available for wor per person of woring age. 5 We can brea his expression down o separae ou changes in real oupu per woring age populaion for he period and +s, his way: Y s Y K + + log N log N log s A s A + 1 log + log α Y + = + s 1 α s 1 α L s L + log N log s N + + s s s K log Y s (4) The firs erm on he righ hand side of he equaion represens TFP s conribuion o growh, while he second erm is he conribuion from changes in he capial-oupu raio, and he hird erm is he conribuion from changes in hours wored per person of woring age. In he long run, empirical evidence reveals ha boh he capial-oupu raio and employmen remain consan. In he shor run, however, facor accumulaion can be very imporan for growh. Table 1 provides he breadown described above of oupu per woring age populaion referred o, from now on, as per capia oupu for he Chilean economy from 1980 o dae. These daa reveal ha for he pas hree years, unlie during he period of susained growh from 1983 o 1998, employmen has been he mos relevan facor behind he level of economic aciviy. 6 This variable explains an average annual decline in per capia oupu of around 2.8%. 7 Per capia oupu, however, rose 0.85% per year on average during his period, due o he fac ha oal facor produciviy growh was 2.07%, wih he capial-oupu raio conribuing 1.57% during his period. In previous years, however, TPF appears o have been he main deerminan of growh. As menioned above, consisen wih he siuaion observed in Chile, Kehoe and Presco (2002) provide evidence ha during he 20 h cenury, mos major depressions were explained by he decline in eiher he efficiency of facor uilizaion or oal hours wored. Table 1 Growh Accouning in Chile Period Change in Y/N Conribuion from PTF Conribuion from K/Y Conribuion from L/N Deerminisic Growh Model In his secion, we will use a deerminisic version of he neoclassical growh. This considers a single good ha is consumed or used in invesmen. The represenaive household solves he following problem: 5 N is obained by muliplying he populaion aged from 16 o 64 years by he number of hours available for wor in he year, assuming 100 hours per wee for 52 wees. L corresponds o he number of people woring in Chile for he average number of hours wored in Greaer Saniago. This breadown is based on Hayashi and Presco (2002). The complee descripion of he daa used and sources appears in Appendix 1. 6 During he crisis in he early 1980s, employmen and oal facor produciviy accoun for similar percenage drops in per capia oupu. 7 Noe ha we are using a logarihmic approximaion for growh. This allows us o carry ou an addiive decomposiion of growh facors. 5

6 max s.a. β [ γ log C + ( 1 γ ) log( N L )] = 1980 (5) C + K + 1 K = ( 1 l τ ) w L + ( 1 τ )( r δ ) K + T where C is consumpion, N - L is leisure, r is he real reurn on capial before axes, w is he real wages, τ l is he labor ax rae, τ is he ax on ne capial minus depreciaion and T is a ransfer ha he governmen pays he consumer. Moreover, β (0,1) is he discoun facor and δ is he depreciaion rae. The represenaive firm solves he problem: α 1 α max Π = A K L r K w L (6) The governmen s problem is o balance is budge, ha is, l + ( r δ ) K T = τ w L τ (7) Finally, he equilibrium requires mare clearing: α 1 α ( 1 δ ) K = A K L Y (8) C K = The consumer problem is characerized by a condiion requiring ineremporal opimizaion for consumpion and an inraemporal consumer-leisure opimizaion condiion, represened, respecively, by he following equaions: ( 1 τ )( r δ ) c+ 1 = Bc C ( 1 γ ) γ l = w (1 τ )( N L ) (9) (10) The problem of firms is characerized by condiions of equaliy beween marginal produciviy and facor prices, r = αa K α 1 1 α L = Y α K (11) w = Y 1 (12) L α α ( α ) A K L = ( 1 α ) Equaions (7)-(12) are necessary and sufficien o compleely characerize he equilibrium. To simulae he, we mus provide he required parameers. The parameric specificaion used is given by β = 0.98, δ = 0.05 and γ = The discoun facor and he depreciaion rae have been specified using he values ypically assigned in he lieraure. The parameer for labor disuiliy, γ, was calibraed according o equaion (13), assuming zero labor ax and considering an average value for he period consisen wih daa for consumpion, employmen and oupu. Therefore his parameer also implicily includes disorions associaed wih he labor mare. This parameer is 6

7 consisen, moreover, wih hose repored by McGraan (1994) for he Unied Saes and Bergoeing e al. (2001) for Chile. In he nex secion, in order o evaluae he plausibiliy of an increase in disorions in he consumpion-leisure decision, associaed wih he labor reform, he labor ax for equaion (13) is calibraed so as o replicae employmen s behavior during he period in Chile. Finally, he capial ax is calibraed in equaion (14), given β and δ. γ = β = C C + w 1 C + 1 l ( N L )( 1 τ ) C ( 1+ (1 τ )( r δ )) (13) (14) Finally, noe ha in our, C corresponds o oal privae and governmenal consumpion and expors. 4. Simulaions This secion uses he growh described above o analyze how he relevance of changes in facor prices resuling from disoring ax policies were o deermining Chile s economic growh over he pas hree years. To do so, we carry ou five simulaion exercises. Each consiss of simulaing he from 1980 o infiniy using acual values for oal facor produciviy and differen values for axes, associaed wih unexpeced reforms. Thus, we repor he impac of oal facor produciviy, he capial-oupu raio, and he employmen-populaion-of-woring-age raio on growh for he period, in a manner consisen wih he breadown presened in he previous secion. 8 Specifically, he firs simulaion exercise consiss of solving for an equilibrium in he absence of axes. The second simulaion incorporaes a capial ax of 49%, for he enire period under analysis. The hird simulaion aes ino consideraion he income ax reforms implemened in Chile in he mid- 1980s. This is simulaed assuming he capial ax falls from 49% o 15% saring in These values have been calibraed for he periods and , respecively, according o he consumpion-invesmen decision implici in he daa using equaion (14). By assuming ha he decline in he capial ax rae is unexpeced, he equilibrium for he firs six years of simulaion remains unchanged. I is ineresing o poin ou ha income ax raes acually implemened in Chile consised of reducing his ax from 45% o 10% in 1985 and han raising i o 15% in As a resul, he capial ax raes calibraed from he daa using equaion (14), alhough hey represen he se of disorions implici in he consumpion daa, are surprisingly similar o he raes acually observed during his period. The fourh exercise is perhaps he mos ineresing for he purpose of his aricle: i assumes ha he debae abou changes o labor legislaion ha sared in 1999 increased he lielihood of labor becoming more expensive, which in he is expressed as a hiring ax. This ax is calibraed so as o replicae he decline in employmen s conribuion o growh as observed in he previous hree years and is mainained from hen on. A final exercise consiss of calibraing he capial ax for he period so as o replicae he observed decline in employmen. Resuls indicae ha: (1) he axless and he unreformed capial ax significanly underesimae oupu growh from 1983 o 1998 and over-esimae i for he nex hree years. The main reason for his underesimaion from 1983 o 1998 is ha he drop in he capial-oupu raio and 8 From 2002 on, i is assumed ha oal facor produciviy grows a he same average rae as i did from Alhough he reform sared in 1985, i wasn compleely implemened unil

8 employmen is overesimaed. During he pas hree years, however, he opposie occurs: he underesimaes he increase in capial and he drop in he fracion of oal hours wored. The drop in employmen is so dramaic ha i ends up overesimaing oupu growh by abou 1.6% during his period; (2) by incorporaing he capial ax reform, resuls for he period improve significanly. Now, capial falls by almos he same amoun as he daa and employmen is underesimaed by 40% less han before. However, for he period he increase in oupu is overesimaed, mainly because he doesn capure he fall observed in employmen. Because of his, exercises 4 and 5 apply increases in employmen and capial axes, respecively, o replicae he behavior observed in employmen. These exercises suppor he hypohesis ha higher producion coss may be responsible for he lower growh observed in Chile during he pas hree years. These simulaions demonsrae ha a 6.75% (up from zero) labor ax or a 32.6% (up from 15%) capial ax can produce his effec. The second ax, however, while i does provide a closer approximaion of oupu s performance does so a he cos of worsening he overall predicion. In paricular, he unreformed from 1999 overesimaes he fall in he capial conribuion, so he higher capial ax worsens he simulaion even more in his regard. Because of his, he labor ax appears o be he mos probable explanaion for wha has occurred in Chile in recen years. This labor ax could be he resul of higher hiring coss perceived by economic agens, as a resul of he debae concerning labor reforms. By raising he relaive price of labor, his percepion, while i may be misaen in pracice, is enough o generae a significan drop in shor-erm growh in Chile. 10 Table 2 Growh accouning in Chile: Simulaions wih α = 0.30 Daa Model 1 Model 2 Model 3 Model 4 Model Change in Y/N Due o PTF Due o K/Y Due o L/N Change in Y/N Due o TFP Due o K/Y Due o L/N Change in Y/N Due o TFP Due o K/Y Due o L/N Noes: Model 1 corresponds o he economy wihou axes presened in secion wo of his sudy. Model 2 inroduces a fixed 49% capial ax. Model 3 includes a capial ax reform reducing i o 15% as of Model 4 adds o Model 3 a labor ax of 6.75% as of Finally, 5 adds o Model 3 a capial ax increase o 32.6% as of The figures a he end of he ex show annual rends affecing he elemens resuling from he breadown of oupu in our daa and simulaions using a fixed capial ax, ax reforms in 1987, and a labor ax as of 1999 for he whole period under analysis. The simplifies realiy in several dimensions, one of which is poenially relevan o our analysis. By using a closed economy for our, we do no ae ino accoun he effec of changes in he erms of rade or oher exernal variables ha may be relevan in he case of a small, open economy lie Chile s. These variables, however, mainly affec wha is referred o here as oal facor produciviy, ha is, he residual ha remains afer considering he accumulaion of labor and 10 Beyer (2001) finds ha he expeced cos of lay-off associaed wih he new labor srucure would rise by abou 16%. 8

9 capial. 11 Wha he daa for he pas hree years show, however, is ha he decline in employmen, raher han oal facor produciviy, was he dominan elemen behind rends in per capial oupu. Moreover, our exercises include acual TFP, hus capuring he impac of he erms of rade on oupu. In his conex, he relaionship beween growh and employmen is no dependen on he assumpion of a closed economy. 5. Conclusions This sudy suggess ha he decline in economic aciviy in Chile during he pas hree years may have been he resul of he hiring expenses of labor perceived by economic agens as a resul of a proposed labor code reform and he debae surrounding i. This reform was finally approved by Congress in Ocober of In our, he increased cos of hiring can be approximaed using a higher labor ax of 6.75% promulgaed as of This, however, does no mean ha he labor reform acually made hiring labor more expensive. Wheher or no his is rue could only be deermined hrough a specific sudy of he reform iself. Therefore, he main conclusion of his sudy is ha such insiuional changes as new rules and regulaions can seriously affec shor-erm economic growh, hrough changes in he relaive prices of producion facors. If hese changes persis, hey may evenually affec long-erm economic rends. References Bergoeing, R., P. Kehoe, T. Kehoe and R. Soo (2002). A Decade Los and Found: Mexico and Chile in he 1980s, Review of Economic Dynamics, 5 (1), Beyer, H. (2001). Una Noa sobre las Reformas Laborales, Cenro de Esudios Públicos, mimeo. Gollin, D. (2001). Geing Income Shares Righ, forhcoming in he Journal of Poliical Economy. Hayashi, F. and E. Presco (2001). The 1990s in Japan: A Los Decade, Review of Economic Dynamics, 5 (1), Kehoe, T. and E. Presco (2002). Grea Depressions of he 20 h Cenury, Review of Economic Dynamics, 5 (1), McGraan, E. (1994). A Progress Repor on Business Cycles, Federal Reserve Ban of Minneapolis Quarerly Review 18 (4). Mendoza, E. (1995). The Terms of Trade, The Real Exchange Rae, and Economic Flucuaions, Inernaional Economic Review 36, Mendoza (1995) demonsraes ha in a neoclassical using a closed economy, such as he one used in his sudy, flucuaions in he erms of rade are capured by he oal facor produciviy parameer, A. If fac, by comparing wih he pas hree years, we see ha TFP s conribuion o per capia growh fell by almos 50%. This fall is undoubedly he resul, among oher facors, of he lower erms of rade apparen since

10 Appendix 1: Descripion and daa sources The Gross Domesic Produc series were obained from he Cenral Ban of Chile. The invesmen series used came from gross capial formaion and invenory changes in he Inernaional Moneary Fund s Inernaional Financial Saisics. Capial was generaed using he invesmen series correced for he assumed depreciaion rae. The woring age populaion corresponds o people from 16 o 64 years of age, as repored by he World Developmen Indicaor. Employmen series are from he Naional Saisics Bureau (INE). Finally, oal hours wored were calculaed using employmen per average hours wored in urban Saniago, according o resuls from he employmen and unemploymen survey carried ou by he Universiy of Chile s economics deparmen. For all series used, figures for 2001 came from acual hird quarer figures and a linear inerpolaion for he fourh quarer. 10

11 Appendix 2: Alernaive Simulaion Table 3 provides he resuls of growh accouning for he daa and for each of he five simulaion exercises presened in Table 2, assuming α = Simulaions were carried ou using β = 0.98, δ = 0.05 and γ = Capial ax raes, calibraed in equaion (14), were in his case τ = 0.72 unil 1986 and τ = 0.51 from 1987 on. The labor ax ha replicaes employmen s conribuion (fall) in he pas hree years is τ l = Finally, he capial ax ha replicaes employmen s conribuion during his period is τ = As wih he previous case, simulaions 1 and 2, ha is, hose wihou ax reforms, considerably underesimae growh in oupu per adul of woring age during he period of susained growh and overesimae his oupu during he period beginning in The capial ax reform ha began in 1987 allows us o replicae he facor accumulaion process observed in he daa wih greaer accuracy. Finally, he labor ax rae necessary o replicae acual employmen rends from is almos equal o he resul of he simulaion exercise repored in Table 1. From he qualiaive poin of view, herefore, he resuls repored in Table 3 do no differ from hose presened in Table 2. The sole difference lies in he relevance of capial and oal facor produciviy in each case. Noneheless, and as menioned above, α=0.45 isn jus implausible from he empirical perspecive according o Gollin (2001), bu moreover suggess an annual before-ax reurn on capial averaging 23% from This rae of reurn is oo high. Table 3 Growh accouning in Chile: Simulaions wih α = 0.45 Daa Model 1 Model 2 Model 3 Model 4 Model Change in Y/N ,15-9,20-9,20-9,20-9,20 Due o PTF -12,60-12,60-12,60-12,60-12,60-12,60 Due o K/Y , Due o L/N -8,38-6,36-7,01-7,01-7,01-7, Change in Y/N 4,85 3,20 2,34 4,33 4,33 4,33 Due o PTF 3,82 3,82 3,82 3,82 3,82 3,82 Due o K/Y ,24-1, Due o L/N 1, ,12 1,12 1, Change in Y/N ,36 1,47 2,26 1, Due o PTF Due o K/Y 3,00 1,62 1,63 2,63 3,30 2,61 Due o L/N -2, ,01-2,79-2,79 Noes: Model 1 corresponds o he economy presened in Secion 2 of his sudy, wihou axes. Model 2 incorporaes a fixed capial ax of 71%. Model 3 includes a capial ax reform reducing i o 51% in Model 4 adds a 6.65% labor ax o Model 3 as of Finally, Model 5 adds o Model 3 a 60.5% reform o he capial ax as of

12 Figures: Y/N: Model 2 Y/N: Model 3 Y/N: Model daa 110 daa 110 daa K/Y: Model 2 K/Y: Model K/Y: Model daa daa 2.0 daa L/N: Model 2 L/N: Model 3 L/N: Model daa daa 0.22 daa