NBER WORKING PAPER SERIES POLICY-DRIVEN PRODUCTIVITY IN CHILE AND MEXICO IN THE 1980S AND 1990S Raphael Bergoeing Parick J. Kehoe Timohy J. Kehoe Raimundo Soo Working Paper 8892 hp://www.nber.org/papers/w8892 NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachuses Avenue Cambridge, MA 02138 April 2002 We hank he paricipans in he Minneapolis Fed s Grea Depressions of he Twenieh Cenury projec, especially Ed Presco, for useful discussions. Jim MacGee provided invaluable research assisance. Bergoeing hanks he Hewle Foundaion, and Kehoe and Kehoe hank he Naional Science Foundaion, for research suppor. The views expressed herein are hose of he auhors and no necessarily hose of he Federal Reserve Bank of Minneapolis or he Federal Reserve Sysem. The views expressed herein are hose of he auhors and no necessarily hose of he Naional Bureau of Economic Research. 2002 by Raphael Bergoeing, Parick J. Kehoe, Timohy J. Kehoe and Raimundo Soo. All righs reserved. Shor secions of ex, no o exceed wo paragraphs, may be quoed wihou explici permission provided ha full credi, including noice, is given o he source.
Policy-Driven Produciviy in Chile and Mexico in he 1980s and 1990s Raphael Bergoeing, Parick J. Kehoe, Timohy J. Kehoe and Raimundo Soo NBER Working Paper No. 8892 April 2002 JEL No. E52, N6, O40 ABSTRACT Boh Chile and Mexico experienced severe economic crises in he early 1980s, bu Chile recovered much faser han did Mexico. Using growh accouning and a calibraed dynamic general equilibrium model, we conclude ha he crucial deerminan of his difference beween he wo counries was he faser produciviy growh in Chile, raher han higher invesmen or employmen. Our hypohesis is ha his difference in produciviy was driven by earlier policy reforms in Chile, he mos crucial of which were in banking and bankrupcy procedures. We propose a heoreical framework in which governmen policy affecs boh he allocaion of resources and he composiion of firms. Raphael Bergoeing Parick J. Kehoe Cenro de Economía Aplicada Deparmen of Economics Universidad de Chile Universiy of Minnesoa Saniago, Chile Minneapolis, MN 55455 raphaelb@dii.uchile.cl and Federal Reserve Bank of Minneapolis and NBER pkehoe@res.mpls.frb.fed.us Timohy J. Kehoe Deparmen of Economics Universiy of Minnesoa Minneapolis, MN 55455 and Federal Reserve Bank of Minneapolis kehoe@alas.socsci.umn.edu Raimundo Soo Insiuo de Economía Ponificia Universidad Caólica de Chile Saniago, Chile and Gerencia Invesigación Económica, Banco Cenral de Chile
I. Inroducion Governmen policy is he primary cause of economic depressions. Exogenous shocks like he deerioraions in he erms of rade and he increases in foreign ineres raes ha buffeed Chile and Mexico in he early 1980s can cause a decline in economic aciviy of he usual business cycle magniude. Governmen policy, however, can urn such a decline ino he severe and prolonged drop in economic aciviy below rend ha consiues a grea depression as defined by Timohy Kehoe and Edward Presco (2002). This view of depressions is suppored by a collecion of papers edied by Kehoe and Presco and published in he Review of Economic Dynamics (January 2002). These papers examine nine depressions using dynamic general equilibrium models. Their sriking finding is ha he main deerminans of depressions are no he drops in he inpus of capial and labor ha radiional heories of depressions sress, bu raher drops in he efficiency wih which hese inpus are used, measured as oal facor produciviy (TFP). Here we skech a heory of how governmen policy drives hese produciviy drops by examining he depressions in Chile and Mexico in he 1980s, which have been exensively sudied by Raphael Bergoeing, Parick Kehoe, Timohy Kehoe, and Raimundo Soo (2002). We hink he heory applies more broadly. Chile and Mexico, like mos Lain American counries, experienced severe economic crises in he early 1980s. Their recovery pahs differed markedly. In Chile, oupu per working-age person reurned o a 2 percen rend growh pah in abou a decade and since hen has grown faser han rend, as shown in Figure 1. In Mexico, oupu never fully recovered and wo decades laer is sill 30 percen below rend. Comparing daa from Chile and Mexico allows Bergoeing e al. (2002) o rejec wo popular explanaions for he economic performances of hese wo counries as explanaions for he difference: (1) Tha Chile s rapid recovery was driven by expor growh. During he lae 1980s and early 1990s, expors grew much faser in Mexico han in Chile, ye Chile grew while Mexico sagnaed. (2) Tha Mexico s sagnaion was due o a large exernal deb overhang ha discouraged new invesmen. The daa show ha Chile s raio of exernal deb o GDP was much higher han Mexico s, and we argue below ha Mexico s sagnaion was no caused by he lack of new invesmen. 1
Our explanaion for he difference in economic performance in Chile and Mexico is based on he differen iming of srucural reforms in he wo counries. Chile underook reforms in he 1970s and early 1980s in rade policy, fiscal policy, privaizaion, banking, and bankrupcy laws ha se he sage for he successful performance of he lae 1980s and 1990s. Mexico posponed hese reforms and sagnaed. Our numerical experimens indicae ha he only reforms ha can explain he difference in economic performance are hose ha show up primarily as differences in produciviy, no reforms ha show up as differences in facor inpus. This resul rules ou fiscal reforms, which primarily affec he incenives o accumulae capial and o work. Moreover, he iming is no righ for fiscal reforms as an explanaion: Boh Chile and Mexico reformed heir ax sysems in he mid-1980s, so hese canno accoun for he differen pahs. The maer of iming is crucial. We hypohesize ha reforms in rade policy and privaizaion were less imporan han hose in banking and bankrupcy law precisely because Chile had already reaped mos of he benefis of hese reforms while Mexico was saring o reap hem precisely when Mexico was sagnaing and Chile was growing. The crucial difference is ha Chile was willing o pay he coss of reforming is banking sysem and of leing inefficien firms go bankrup while Mexico was no. II. Growh Model Figure 2 plos TFP for Chile and Mexico. (See Bergoeing e al. (2002) for deails.) The similariies beween hese daa and hose for GDP per working-age person in Figure 1 srongly sugges ha TFP differences were primarily responsible for he difference in economic performance in Chile and Mexico. Wha abou changes in facor inpus? Some changes in facor inpus are induced by changes in TFP; ohers can be caused by fricions in facor markes. To disinguish beween hese wo ypes of changes facor inpus, we calibrae a growh model of a closed economy in which consumers have perfec foresigh over he TFP sequence. Our model includes a fiscal reform mean o capure some feaures of he reforms implemened in boh counries in he 1980s ha affeced facor inpus. To demonsrae ha differences in hese reforms did no drive he difference in he recoveries, we impose he same crude ax reform in boh counries. 2
In he model, he sand-in consumer chooses sequences of consumpion hours worked L o maximize (1) log C (1 )log( N L) 1980 C, capial K, and subjec o a budge consrain in each period C K 1 K wl (1 )( r ) K T and an iniial condiion on capial K 1980. Here N is hours available; w and r are facor prices; is he income ax rae on capial income; and T is a lump-sum ransfer ha in equilibrium is equal o ax revenue ( r ) K. The feasibiliy consrain is 1 C K 1 (1 ) K AK L. Figures 3 and 4 repor he resuls of numerical experimens of a model calibraed o 1960 1980 daa. We compare he daa wih equilibria in which consumers have perfec foresigh beween 1980 and 1988, are surprised by a ax reform in 1988 ha in Chile lowers from 0.51 o 0.12 and in Mexico lowers from 0.43 o 0.12, and have perfec foresigh from hen on. The ax parameers have been calibraed from he daa using he firs-order condiions from he consumer s problem, bu here is direc evidence ha he sors of ax reform ha we have imposed are sensible. In Chile, for example, he ax rae on firms income was lowered from 0.46 o 0.10 in 1984; i was increased o 0.15 in 1991. The resuls of our numerical experimens are sriking: Differences in TFP including induced effecs on facor inpus drive all of he differences in he economic experiences of Chile and Mexico. Tax reforms make Chile s recovery sronger and Mexico s depression less severe han hey would have been oherwise, bu do no drive he difference. There is no significan role lef o be played by fricions in facor markes ha do no show up in TFP. III. Explaining TFP Declines If TFP movemens drove boh he iniial downurns in Chile and Mexico and he difference in heir recoveries, wha drove he TFP movemens? Exernal shocks iniiaed he TFP drops. These drops were magnified by exising governmen policies ha made he financial secors in boh economies fragile. In Chile, rapid policy reform led o a recovery of TFP. In Mexico, he policy reacion o he iniial shocks increased disorions, resuling in a prolonged decline in TFP. 3
Economic heory suggess wo obvious mechanisms for he iniial TFP drops: (1) Higher real ineres raes make some previously profiable invesmen projecs unprofiable and make he capial involved unproducive. Our measure of TFP assumes full uilizaion of capial and, hence, aribues his drop in uilizaion o TFP. (2) A deerioraion in he erms of rade requires a counry o expor more o obain he same quaniy of impors, so ha more domesic resources are needed o produce he same final oupu. Hence, in a closed economy model, negaive erms of rade shocks manifes hemselves as declines in TFP. While hese mechanisms can accoun for par of he iniial TFP drops in boh counries, hey canno accoun for eiher he severiy of hese drops or he difference in he recovery pahs. In our view, misakes in governmen policies made he financial sysems in boh counries fragile and, hence, exacerbaed he effecs of he iniial exernal shocks. Bu wha drove he sriking difference in he paerns of recovery of TFP seen in Figure 2? Our view is ha higher produciviy growh in Chile was driven by he iming of banking and bankrupcy reforms. IV. Policy-Driven Produciviy Our hypohesis is ha, before reforms, governmen policy disored he allocaions of resources boh wihin and across secors in Chile and Mexico, pushing he economies inside heir aggregae producion possibiliy froniers. As reforms were implemened, firs in Chile and laer in Mexico, each economy moved closer o is producion possibiliy fronier. Here we propose a heory of how disorions in he financial sysem and poorly designed bankrupcy laws can have negaive effecs on produciviy. We describe firs he saic effecs of disorions on allocaions of resources and hen he dynamic effecs on he enry and exi decisions of firms. Saic Effecs of Policy A governmen s favored reamen of cerain secors can lead o a saic misallocaion of resources. Suppose ha he governmen favors some firms or secors wih low-ineres-rae loans. Wih he reurn on deposis held fixed, lower ineres raes for some secors mus be paid eiher by higher raes in unfavored secors or by ransfers o he financial sysem from he governmen. To see his in a simple model, suppress labor and le he producion funcions be AK i i, i 1, 2. Secor 1 receives a proporional subsidy 1 on he ineres rae ha i pays on loans, and 4
secor 2 pays a ax 2. We can hink of hese subsidies and axes as being accomplished hrough preferenial lending by banks. Any ne revenues required by banks are financed by he governmen using lump-sum axes or ransfers. The firs-order condiions for profi maximizaion imply ha he relaive allocaions of capial are given by (2) K K 1 1 1 1 1 A 1 1 2 A 1 2 2 1. In conras o he efficien allocaion, he disored equilibrium has a higher fracion of he capial sock allocaed o he subsidized secor han is warraned by relaive produciviies and, hence, lower aggregae oupu. If hese disorions decrease he incenives o make loans, as hey did in Mexico, hen hey also lead o a lower aggregae capial sock and have an addiional negaive effec on oupu. Despie he saic naure of he inefficiency, his mechanism has he poenial for explaining some of he difference in TFP performance: Financial reform was he major reform ha Chile had done poorly in he 1970s and had o redo in he early 1980s. In Mexico, however, unil a leas he early 1990s, he banking sysem remained naionalized, gave subsidized loans o some firms, and raioned loans o ohers. Dynamic Effecs of Policy One way disorions in he financial sysem and poorly designed bankrupcy rules can lead o lower aggregae produciviy is by discouraging poorly performing firms from exiing from producion. By keeping firms operaing ha oherwise would have exied, hese disorions and bankrupcy rules can preven new poenially producive firms from enering. The models developed by Andrew Akeson and Parick Kehoe (1995) and Tianshu Chu (2001) can be used o address such issues. In hese models, new firms ener wih he newes echnology, sochasically learn over ime, and hen exi when heir prospecs for furher produciviy improvemens are poor. In hese models, any policy ha inerferes wih he naural birh, growh, and deah of firms based on curren and prospecive produciviies can push he economy furher and furher inside he producion possibiliy fronier. Simulaions by Chu (2001) indicae ha hese dynamic disorions can have enormous effecs over ime. 5
To ge some idea of how his sor of model works, consider Akeson and Kehoe s (1995) model in which producion is done in a coninuum of plans. Each plan has is own level of produciviy A and is operaed by a manager. A plan wih produciviy A has he producion 1 1 funcion y A ( k l ). The manager s span of conrol parameer, v 1, deermines he degree of diminishing reurns a he plan level. A manager who decides o operae a plan chooses capial k and labor l o maximize saic reurns (3) d A A k l rk wl w, 1 1 ( ) max m ( ) kl, where w m is he manager s opporuniy cos of no working or saring anoher plan. Le he soluions be denoed k (A) and l A ). For a given disribuion (A) of produciviies across ( 1 1 plans, aggregae oupu is given by Y A K L, where A A ( da) is aggregae produciviy and K k ( A) ( da) and L l ( A) ( da) are aggregae capial and labor. A Tha is, aggregae produciviy is he mean produciviy of he plans ha are operaing. A Over ime, he produciviy of each plan evolves sochasically: A plan wih produciviy A a has random produciviy A a + 1, where is drawn from a probabiliy disribuion ( ). The crucial decision for he manager of wheher or no o operae a plan is dynamic and is described by he Bellman equaion o o 1 (4) V( A) max[0, V ( A)], where V ( A) d( A) V 1( A ) ( d ) 1 R. A Here R is he marke rae of ineres beween and + 1, V o ( A) is he value if he plan is operaed in he curren period, and V (A) is he maximum of he reurns from eiher closing he plan or operaing i. New plans can ener according o a similar process. The oucome of all he managerial decisions o operae or no is a new disribuion ( ) over produciviies in + 1. 1 A In his simple version of he model, he probabiliy disribuion ( ) ha generaes shocks o plan-specific produciviy does no change wih plan age. In he daa, he labor employed and oupu produced by a cohor of plans end o sar low when plans are young, grow for he nex 20 years or so, and hen gradually decline. To capure his sor of paern, Akeson and Kehoe (1995) add a fronier level of new echnology ha grows over ime, and hey allow he mean 6
value of shocks o produciviy o firs increase and hen decrease. In heir model, new plans ener wih he newes echnology, bu lile specific knowledge on how o use i. Over ime, hese plans build up heir specific knowledge and grow as hey draw shocks o produciviy from a disribuion wih an increasing mean. Afer 20 years or so, he mean of he shocks sars o fall, as he learning process slows. Plans hen decline in size and evenually exi. Now consider he effecs of a disorion which he governmen encourages banks o lend a a subsidized rae R R ( 1 ) o firms (idenified wih plans in he model) in secor 1 and o 1 1 lend o oher firms a a relaively high rae R R ( 1 ). The favored firms discoun he 2 2 fuure less han he unfavored ones. In addiion o he saic effec of misallocaing capial, his disorion resuls in differen soluions o he managers dynamic programming problems in secors 1 and 2. In paricular, favored firms will choose o coninue o operae in siuaions where unfavored firms would choose o exi. Even if a favored firm experiences a low produciviy A, i will be more prone han an unfavored firm o borrow o cover is losses and o coninue operaing, hoping for a favorable shock o increase is produciviy in he fuure. Consider nex poorly designed bankrupcy procedures ha make i difficul for firms o exi or subsidize he losses of firms o keep hem operaing. These procedures cause firms o keep operaing for longer han hey would have oherwise and impede he enry of new firms. Togeher, disorions in he financial sysem and poorly designed bankrupcy procedures change he mix of firms ha operae, leading o inefficienly many low-produciviy firms coninuing. This leads o a lower value of aggregae produciviy A. How would he removal of hese disorions affec he pah of produciviy over ime? Some effecs would be immediae. Upon removal, some previously favored firms ha would have coninued will exi, and some unfavored firms ha would have exied will coninue o operae. The subler, and poenially more imporan, effecs ake more ime o show up in aggregaes. The removal of disorions would encourage new firms o ener. These firms would have he newes echnologies, bu would build up heir organizaion-specific produciviy only slowly. Consequenly, removing hese disorions would show up wih a lag in he aggregae produciviy saisics. 7
REFERENCES Akeson, Andrew and Kehoe, Parick J. Indusry Evoluion and Transiion: Measuring Invesmen in Organizaion Capial. Federal Reserve Bank of Minneapolis, Research Deparmen Saff Repor 201, 1995. Bergoeing, Raphael; Kehoe, Parick J.; Kehoe, Timohy J. and Soo, Raimundo. A Decade Los and Found: Mexico and Chile in he 1980s. Review of Economic Dynamics, January 2002, 5(1), pp. 166 205. Chu, Tianshu. Exi Barriers and Economic Sagnaion. Manuscrip, Eas-Wes Cener, 2001. Kehoe, Timohy J. and Presco, Edward C. Grea Depressions of he 20h Cenury. Review of Economic Dynamics, January 2002, 5(1), pp. 1 18. 8
Figure 1. Real GDP per working-age (15-64) person in Chile and Mexico, derended by 2% per year. 130 120 110 Chile 100 90 80 70 Mexico 60 1980 1985 1990 1995 2000 year 9
Figure 2. Toal facor produciviy in Chile and Mexico, derended by 1.4% per year. 110 100 Chile 90 80 70 Mexico 60 1980 1985 1990 1995 2000 year 10
Figure 3. Model experimen: Real GDP per working age (15-64) person in Chile, derended by 2% per year. 130 120 110 model 100 90 80 70 daa 1980 1985 1990 1995 2000 year 11
Figure 4. Model experimen: Real GDP per working age (15-64) person in Mexico, derended by 2% per year. 110 100 model 90 80 70 60 daa 1980 1985 1990 1995 2000 year 12