Federal Reserve Bank of Minneapolis June 2004 Modeling he Dynamic Effecs of Trade and Foreign Invesmen Liberalizaion Timohy J. Kehoe* Universiy of Minnesoa and Federal Reserve Bank of Minneapolis ABSTRACT Polices governing rade and foreign invesmen can have imporan effecs ha are no capured by saic applied general equilibrium (GE) models. Mos imporanly, such policies can affec changes in efficiency, measured as oal facor produciviy (TFP). Numerical experimens using a simple dynamic applied GE model indicae ha flucuaions in TFP are he primary deerminans of macroeconomic flucuaions in Mexico over he period 1980-2002. Furhermore, rade and foreign invesmen liberalizaion can change incenives for boh domesic savings and foreign invesmen, hereby changing he rae of capial accumulaion. Each of hese wo ses of effecs can aler boh he rae of economic growh in an economy and relaive prices and can dwarf he effecs analyzed by saic applied GE models. * This paper summarizes a presenaion a he Empirical Trade Analysis Conference: Srenghening Analyical Capabiliies o Suppor Trade Negoiaions held a he Woodrow Wilson Inernaional Cener for Scholars, 22-23 January, 2004. A preliminary version was circulaed in January 2004. I hank Kim Ruhl for excellen research assisance and he Naional Science Foundaion for financial suppor. The views expressed herein are hose of he auhor and no necessarily hose of he Federal Reserve Bank of Minneapolis or he Federal Reserve Sysem.
1. INTRODUCTION Trade and foreign invesmen policies can have large impacs ha are no capured by saic applied general equilibrium (GE) models. Mos imporanly, hese policies can affec he rae of growh of produciviy. Furhermore, hey can affec incenives for domesic savings and foreign invesmen, hereby changing he rae of capial accumulaion. Each of hese wo ses of effecs can aler he rae of growh in an economy and dwarf he effecs analyzed by saic applied GE models. Kehoe (1994) sresses his poin in analyzing he impac of he Norh American Free Trade Agreemen (NAFTA) on he Mexican economy. Saic applied GE models capure he efficiency gains from he reallocaion of resources across secors in response o changes in relaive prices. Even here, however, here are poenially large impacs of policy changes ha are no capured by saic models. The large capial inflows ha ofen accompany liberalizaion produce an iniial appreciaion of he real exchange rae followed by a laer depreciaion. These exchange rae movemens are changes in relaive prices generaed by he iniial reallocaion of resources from he raded goods secor o he nonraded goods secor and he laer reallocaion back o he raded goods secor. 2. BIG QUESTION: WHAT DRIVES CHANGES IN PRODUCTIVITY? The key o modeling no jus economic growh bu large economic flucuaions like depressions is modeling changes in produciviy. Bergoeing, Kehoe, Kehoe, and Soo (2002) sudy he depressions ha began in Chile and Mexico in he early 1980s and he differen pahs ha hese wo economies followed aferward, wih Chile growing rapidly and Mexico mired in crisis or sagnaion unil 1995. Bergoeing e al. conclude ha he differences in he recovery pahs of Chile and Mexico were due o differences in he pahs of oal facor produciviy (TFP) raher han o differences in he inpus of labor or capial. They argue ha hese differences in produciviy were due o Chile s earlier reforms in banking and bankrupcy procedures, which generaed a disribuion of firms wih higher produciviy han ha of he disribuion of firms in Mexico. In boh counries, fiscal reforms in he mid o lae 1980s led o increases in invesmen, bu his increased boh recovery pahs raher han causing he wo pahs o differ. We focus he relevance of he research of Bergoeing e al. for applied GE analysis very ighly as a challenge o modelers of NAFTA. Using a simple dynamic applied GE model, we argue ha, by exogenously incorporaing he evoluion of TFP, we capure he deerminans of 1
mos of he macroeconomic flucuaions ha occurred in Mexico over he period 1980-2002. The changes in rade flows and foreign invesmen associaed wih NAFTA are relevan o he exen ha hey deermine produciviy, no aggregae employmen or invesmen. To make he poin blunly, and perhaps a lile crudely: if NAFTA was no imporan for TFP in Mexico, hen i was no imporan in deermining macroeconomic flucuaions here. I is worh noing ha Trefler (2001) finds ha he major impac of he Canada-U.S. FTA on Canada was in changing he disribuion of firms in erms of size and produciviy. Trefler argues ha he change in he disribuion of firms ha occurred in Canada did no mach he predicions of applied GE models of NAFTA ha relied on he Dixi-Sigliz (1977) heory of indusrial organizaion. We modify he closed economy model of Bergoeing, Kehoe, Kehoe, and Soo (2002) o include flucuaions in he rade balance. The feasibiliy consrain in his economy is Here C + K 1 (1 δ ) K + X = Y= AK L. α 1 + C is aggregae consumpion, boh privae and public; K is capial; K 1 (1 + δ ) K is gross invesmen; δ K is depreciaion; X is he rade balance; Y is GDP; and L is he labor inpu measured in hours worked per year. Following Bergoeing e al., we se δ = 0.050 and cumulae invesmen o calculae he capial sock, and hen se α = 0.300 o calculae TFP, K = 1 I + (1 + δ ) K, A = Y K L. α 1 α We model flucuaions in A as exogenous. The poin is no ha we as applied GE modelers should wan o ake changes in produciviy as exogenous, bu precisely he opposie: if a model wih changes in TFP reaed as exogenous accouns for mos macroeconomic flucuaions, hen we know ha i is changes in TFP ha we need o explain! The sand-in consumer chooses sequences of consumpion, capial, and hours worked o maximize he uiliy funcion = 1980 β γ log C + (1 γ)log( hn L) 2
subjec o he budge consrain in each period, C + K + 1 K = wl + (1 τ )( r δ ) K + T X and an iniial condiion on capial, K 1980. Here h is he number of hours available, assumed o be 100 hours per week, 52 weeks per year for working age (15-64) persons; N is he working age populaion; ( hn L ) is leisure; r and w are he marginal producs of producion funcion A K L α 1 wih respec o K and L ; τ is he ax rae on capial income; and T is a lump-sum ransfer ha, in equilibrium, equals ax revenue τ ( r δ ) K. Using he firs-order condiions for he labor-leisure decision from he consumer s problem, we follow Bergoeing e al. in using 1960-1980 daa ha is, daa are no from he period ha we sudy o esimae γ = 0.304. Seing β = 0.985, we use he firs-order condiion for he consumpion-invesmen decision o esimae a consan ax disorion τ = 0.455. A, and Figure 1 presens he resuls of four numerical experimens in which he sequences of X are reaed as exogenous. In each, N, N and A grow a heir rend growh raes 1980-2002 afer 2002, X shrinks o 0 a a consan rae afer 2002, and he equilibrium is assumed o converge o is balanced growh pah by 2030. In he base case, we se he rade balance and TFP equal o heir values in he daa over 1980-2002, and we impose a ax reform ha unexpecedly lowers he ax rae o τ = 0.133 in 1988. The base case experimen does well in racking he performance of he Mexican economy 1980-2002 because capures well he endogenous responses of he economy o he exogenous flucuaions ha we have imposed in he rade balance, he ax rae, and TFP. The hree oher experimens examine he relaive imporance of incorporaing each of hese hree exogenous sources of flucuaions by eliminaing each one separaely, one a a ime. In he experimen consan rade balance, we mainain he rade balance consan a is average value over 1980-2002. Conrasing he resuls of his experimen wih hose of he base case, we observe ha incorporaing he evoluion of he rade balance in he model even is violen flucuaions in 1981-1984 and 1994-1996 has relaively lile impac on he evoluion of GDP. Tha flucuaions in he rade balance have lile impac on he equilibrium of our one- 3
secor model does no imply ha he corresponding flucuaions in foreign invesmen played no role in deermining economic flucuaions in Mexico over 1980-2002. Flucuaions in foreign invesmen can have large effecs on relaive prices and he allocaion of resources across raded and nonraded goods secors. I is jus ha whaever impac hese flucuaions have a a macroeconomic level works primarily hrough flucuaions in produciviy raher han hrough flucuaions in aggregae employmen or invesmen. In he experimen no ax reform, we do no lower he ax rae in 1988. Conrasing he resuls of his experimen wih hose of he base case, we see ha incorporaing he fiscal reforms of 1987 and 1989 in he model has a relaively large impac on he equilibrium. We follow Bergoeing e al. in modeling hese reforms as a fall in he ax rae in 1988, using he firs-order condiions for he consumpion-invesmen decision and 1988-2002 daa o esimae a new value for τ. Observe ha, if liberalizaion of foreign invesmen had an impac on incenives o accumulae capial, i may have been responsible for some of he esimaed fall in he effecive ax on capial income. In he experimen consan TFP growh, we model TFP as growing a is 1960-1980 rend. Noice ha, if we do no incorporae he TFP pah in he daa, we lose almos all abiliy o accoun for flucuaions even hough we have incorporaed changes in he rade balance and in axes. The message is clear: if changes in rade flows and capial flows ino Mexico had imporan effecs on macro aggregaes, hen hese effecs had o operae hrough produciviy and no jus by loosening he feasibiliy consrain or alering aggregae employmen or invesmen. 3. WHERE DO WE GO FROM HERE? Trade flows and capial flows affec produciviy by reallocaing resources across secors and by reallocaing resources across firms wihin a secor. Fernández de Córdoba and Kehoe (2000) use a model wih a raded goods secor and a nonraded goods secor o analyze Spain s 1986 enry ino wha was hen he European Communiy. Similar models have been used by Bajona and Chu (2003) o analyze China s accession o he World Trade Organizaion and by Bems and Jönsson (2002) o analyze rade and foreign invesmen liberalizaion in he Balic counries. In response o liberalizaion, capial flows ino he counry, and resources flow from he raded goods secor o he nonraded goods secor, causing he relaive price of he nonraded good o rise a real exchange rae 4
appreciaion. Laer, as he economy sars o run rade surpluses o repay he res of he world for is earlier rade deficis, resources flow from he nonraded goods secor o he raded goods secor, causing he relaive price of nonraded goods o fall a real exchange rae depreciaion. Fernández de Córdoba and Kehoe argue ha a calibraed model needs o include fricions on he movemens of facors from he one secor o anoher o produce he sors of changes in relaive prices observed in he daa. Combining hese fricions wih sudden changes in he rade balance, like ha in Mexico in 1995 which force resources o move rapidly from one secor o anoher has he poenial o explain sudden drops in aggregae produciviy. An even more promising, bu more difficul, approach o modeling he impac of changes in rade and foreign invesmen policies on produciviy would incorporae heerogeneiy among firms wihin differen secors, following Lucas (1978) and Hopenhayn (1992). Policy changes ha induce resources o flow from inefficien firms o efficien firms increase TFP. Some firs seps in consrucing rade models wih heerogeneous firms have been aken by Bernard, Eaon, Jensen, and Korum (2003) and Meliz (2003). A crucial ingredien in his sor of model is he decision by firms wheher or no o expor. Rolleigh (2003) finds ha a model wih firms wih heerogeneiy in produciviy and in he fixed coss of exporing can do beer han saic applied GE models in accouning for he increase in rade afer he implemenaion of NAFTA and he disribuion of his increase across secors. Ruhl (2003) uses a dynamic sochasic model wih heerogeneous firms ha solve dynamic programming problems in which he crucial decision is wheher or no o pay he fixed coss of exporing. He finds ha firms respond differenly o business cycle shocks ha cause flucuaions in he real exchange rae han hey do o policy changes ha permanenly change he coss of rading. This finding poenially explains why saic models wih Armingon elasiciies esimaed using flucuaions in relaive prices a he business cycle frequency failed miserably in predicing he huge increase in rade volume ha accompanied he policy changes in NAFTA. Reflecing on he poor performance of he saic applied GE models used o sudy he impac of NAFTA, Kehoe (2003) conjecures ha he bigges effec of liberalizaion of rade and foreign invesmen is on produciviy hrough changing he disribuion of firms and encouraging echnology adopion raher han he effecs analyzed by saic applied GE models. A lo more research is needed. 5
REFERENCES Bajona, C. and Chu, T. (2003), Economic Effecs of Liberalizaion: The Case of China s Accession o he World Trade Organizaion, Eas-Wes Cener, Universiy of Hawaii. Bems, R. and Jönsson, K. (2002), Trade Deficis in he Balic Saes: How Long Will he Pary Las? Sockholm School of Economics. Bergoeing, R., P. J. Kehoe, T. J. Kehoe, and R. Soo (2002), A Decade Los and Found: Mexico and Chile in he 1980s, Review of Economic Dynamics, 5, 166 205. Bernard, A. B., J. Eaon, J. B. Jensen, and S. Korum, (2003), Plans and Produciviy in Inernaional Trade, Naional Bureau of Economic Research Working Paper 7688. Dixi, A. K., and J. E. Sigliz (1977), Monopolisic Compeiion and Opimum Produc Diversiy, American Economic Review, 67, 297 308. Fernández de Córdoba, G. and T. J. Kehoe (2000), Capial Flows and Real Exchange Rae Movemens Following Spain s Enry ino he European Communiy, Journal of Inernaional Economics, 51, 49 78. Hopenhayn, H. (1992), Enry, Exi, and Firm Dynamics in Long Run Equilibrium, Economerica, 60, 1127-1150. Kehoe, T. J. (1994), Towards a Dynamic General Equilibrium Model of Norh American Trade, in J. Francois and C. R. Shiells, ediors, Modeling Trade Policy: Applied General Equilibrium Assessmens of Norh American Free Trade, Cambridge Universiy Press, 328-347. Kehoe, T. J. (2003), An Evaluaion of he Performance of Applied General Equilibrium Models of he Impac of NAFTA, in T. J. Kehoe, T. N. Srinivasan, and J. Whalley, ediors, Froniers in Applied General Equilibrium Modeling: Essays in Honor of Herber Scarf, Cambridge Universiy Press, forhcoming. Lucas, R. E. (1978), On he Size Disribuion of Business Firms, Bell Journal of Economics, 9, 508-523. Meliz, M. J. (2003), The Impac of Trade on Inra-Indusry Reallocaions and Aggregae Indusry Produciviy, Economerica, 71, 1695-1725. Rolleigh, M. (2003), Plan Heerogeneiy and Applied General Equilibrium Models of Trade: Lessons from he Canada-US Free Trade Agreemen, Universiy of Minnesoa. Ruhl, K. J. (2003), Exporing Under Uncerainy: Reconciling Esimaes of he Armingon Elasiciy, Universiy of Minnesoa. Trefler, D. (2001), The Long and Shor of he Canada-U.S. Free Trade Agreemen, Universiy of Torono. 6
Figure 1 120 110 Real GDP per Working Age Person in Mexico base case index (1980=100) 100 90 daa consan rade balance 80 70 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 120 110 base case index (1980=100) 100 90 daa 80 no ax reform 70 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 200 index (1980=100) 180 160 140 120 100 80 consan TFP growh base case daa 60 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002