ESSAYS ON THE EFFECT OF INFLATION VOLATILITY AND INSTITUTIONS ON GROWTH AND DEVELOPMENT NOHA EMARA

Transcription

1 ESSAYS ON THE EFFECT OF INFLATION VOLATILITY AND INSTITUTIONS ON GROWTH AND DEVELOPMENT by NOHA EMARA A Disseraion submied o he Graduae School New Brunswick Rugers, The Sae Universiy of New Jersey In parial fulfillmen of he requiremens For he Degree of Docor of Philosophy Graduae Program in Economics Wrien under he direcion of Ira Gang And approved by New Brunswick, New Jersey May, 2009

2 ABSTRACT OF THE DISSERTATION Essays on he Effec of Inflaion Volailiy and Insiuions On Growh and Developmen by NOHA EMARA Disseraion Direcor: Ira Gang The purpose of his disseraion is o analyze empirically and heoreically he impac of he decrease in inflaion volailiy versus he impac of he improvemen in insiuions on growh and developmen. The firs chaper of his disseraion esimaes he effecs of inflaion and inflaion volailiy on economic growh in he presence of differen degrees of legal and financial insiuions. The main conribuion of his chaper is o show ha while he level of inflaion does no have a significan effec on growh, which is in line wih previous sudies; inflaion volailiy does significanly impac growh even for counries wih moderaely high levels on inflaion. In addiion, improving eiher legal or financial insiuions has a saisically significan posiive impac on growh and helps o reduce he negaive impac of inflaion volailiy on growh. The second chaper analyzes he channel hrough which inflaion volailiy and financial insiuions affec a counry s abiliy o borrow on inernaional capial markes; ii

3 which affecs heir abiliy o inves and herefore grow. The findings of his chaper show ha reducing inflaion volailiy or improving financial insiuions will significanly improve a counry s sovereign deb raing leading o a drop in is cos of borrowing, which is o be quanified. One imporan conribuion of his chaper is o show ha i is inflaion volailiy ha is imporan in deermining a counry s sovereign deb raing raher han he level of inflaion which has been argued in he lieraure. The welfare implicaions of he decrease in inflaion volailiy versus he improvemen in insiuions are quanified in chaper hree. This chaper analyzes a micro-foundaion based small open economy model ha is used o help fully undersand he dynamics of a decrease in inflaion volailiy and an improvemen in insiuions for a developing economy. The sudy finds ha he welfare effec of improving insiuions and of reducing inflaion volailiy is large wih he larges effec being caused by an improvemen in financial insiuions. One policy implicaion of hese resuls is ha developing economies can ge larger welfare gains from improving heir insiuions han from reducing inflaion volailiy. iii

4 ACKNOWLEDGMENTS I graefully acknowledge Professor John Landon-Lane, my main disseraion advisor, who encouraged me o launch his disseraion projec, and guided me wih paience all along. This disseraion would have been impossible wihou his ongoing suppor. I owe many hanks o Ira Gang, Chairman of my disseraion, and o Michael Bordo and he ouside member, Rob Vos, for heir helpful commens as I developed he successive drafs of his disseraion. I would also like o hank Dorohy Rinaldi for her help, kindness, and her advice. Since my firs day in he graduae program a Rugers, Dorohy has been always supporive and encouraging. I would like also o hank all my friends for heir help and suppor. Among many, I would like specifically o hank Haruhiko Shimizu who spen long hours wih me o help me sudy and prepare for he qualifying and he field exams, Deme Tunali who shared wih me all my moral and emoional highs and lows hroughou he graduae school, Leah Traub for her valuable commens and advice a he various sages of developing his disseraion. I will never find words of hanks and graiude o my faher who made all of my life accomplishmens possible. To my moher for her coninuous prayers and encouragemen. To my sisers for heir love and sincere wishes. To my sons, Mousafa and Ismail, for puing up wih heir mom while spending mos of her ime away from iv

5 hem while sudying and working on his disseraion. To my husband, Ihab, for his help and emoional suppor. This disseraion would have no been possible wihou his love and encouragemen. v

6 DEDICATION To Papi and Mami Mousafa and Ismail And Ihab vi

7 TABLE OF CONTENTS Absrac ii Acknowledgemens.....iv Dedicaion vi Lis of Tables... x Lis of Figures... xiv Par I: Inroducion...1 Par II: Chaper 1: Inflaion Volailiy, Insiuional Environmen, and Economic Growh : Inroducion : Empirical Specificaion : Daa : Esimaion Resuls : Toal Effec of Inflaion Volailiy : Robusness Check : Reverse Causaliy Tes : Regular Panel Esimaion : Conclusion.. 42 Par III: Chaper 2: Inflaion Volailiy, Financial Insiuions, and Sovereign Deb Raing : Inroducion : Empirical Specificaion...54 vii

8 2.3: Daa : Esimaion Resuls : Esimaing he Impac of he Componens of Financial Insiuions : Calculaing he Toal Effecs : Toal Effec of Inflaion Volailiy : Toal Effec of Financial Insiuions : Robusness Check : Impac of he Increase in Sovereign Deb Raing on Long-Term Bond Yield : Conclusion.. 89 Par IV: Chaper 3: Welfare Implicaions of Inflaion Volailiy and Financial Insiuions in a Small Open Economy : Inroducion : Moneary Small Open Economy Model : Households : Firms : Open Economy Condiions : Defining he Equilibrium : Characerizing he Equilibrium : Seady Saes : Calibraion : Resuls : Impulse Response Funcions for a Permanen Improvemen viii

9 In Insiuions : Calculaing he Welfare Implicaions of he Reducion in Inflaion Volailiy and he Improvemen in insiuions : Transiion Pahs Deerminisic Model : The Three Experimens Along he Transiion Pah : Comparison of Experimens : Conclusion 131 Appendix (I).133 Appendix (II) Appendix (III)..166 Bibliography Curriculum Via ix

10 Lis of Tables Table 1.1: The Base Model of Economic Growh 23 Table 1.2: Economic Growh and he Ineracion beween inflaion Volailiy And Legal and Financial Insiuions Indices...28 Table 1.3: Toal Effec of Inflaion Volailiy on Growh (Given he LEGAL1 index)...33 Table 1.4: Toal Effec of Inflaion Volailiy on Growh (Given he LEGAL2 index)...34 Table 1.5: Lis of Counries included in he Sample..133 Table 1.6: Summary Saisics of main variables 133 Table 1.7: Correlaions beween he Main Variables..134 Table 1.8: Legal and Financial Insiuions indices 134 Table 1.9: Economic Growh and he Ineracion beween inflaion Volailiy And Economic Variables.138 Table 1.10: Economic Growh and he Ineracion of inflaion Volailiy wih LEGAL1 Indices Table 1.11: Economic Growh and he Ineracion of inflaion Volailiy wih LEGAL2 Indices Table 1.12: Toal Effec of Inflaion Volailiy on Growh (Given Corrupion) 142 Table 1.13: Toal Effec of Inflaion Volailiy on Growh (Given Law & Order) 143 Table 1.14: Toal Effec of Inflaion Volailiy on Growh (Given Bureaucracy) Table 1.15: Toal Effec of Inflaion Volailiy on Growh (Given Crediors Righs)..145 Table 1.16: Toal Effec of Inflaion Volailiy on Growh (Given Shareholders Righs) x

11 Table 1.17: Toal Effec of Inflaion Volailiy on Growh (Given Enforcemen).147 Table 1.18: Toal Effec of Inflaion Volailiy on Growh (Given Accouns) Table 1.19: Robusness Check: Economic Growh and he Ineracion beween inflaion Volailiy and Legal and Financial Insiuions Indices 149 Table 1.20: Robusness Check: Toal Effec of Inflaion Volailiy (Given LEGAL1ndex) Table 1.21: Robusness Check: Toal Effec of Inflaion Volailiy (Given LEGAL2 index). 150 Table 2.1: Lis of Counries included in he Sample Table 2.2: Definiion of Variables 60 Table 2.3: Firs Sage F-saisic of he TSLS...66 Table 2.4: Sovereign deb raing and Inflaion Volailiy...67 Table 2.4 : Robusness Check: Sovereign deb raing and Inflaion Volailiy 163 Table 2.5: Sovereign deb raing and Inflaion Volailiy Ineracions wih LEGAL2 Sub-indices...72 Table 2.6: Toal Effec of Inflaion volailiy on Sovereign Deb Raing (Given LEGAL2 index) Table 2.6 : Robusness Check: Using Panel Leas Squares..85 Table 2.6 : Robusness Check: Toal Effec of Inflaion volailiy on Sovereign Deb Raing (Given LEGAL1 index). 164 Table 2.7: Perceniles of he Daa on he LEGAL2 Index 75 Table 2.8: Toal Effec of LEGAL2 on Sovereign Deb Raing (Given Inflaion Volailiy)...79 xi

12 Table 2.8 : Robusness Check: Using Panel Leas Squares.. 85 Table 2.8 : Robusness Check: Toal Effec of LEGAL1 on Sovereign Deb Raing (Given Inflaion Volailiy) 164 Table 2.9: Perceniles of he Average of he Log of Inflaion Volailiy Daa over he Period Table 2.10: Sovereign deb raing and Inflaion Volailiy...83 Table 2.11: Descripive Saisics Table 2.12: Correlaion Marix Table 2.13: Regional Dummies..152 Table 2.14: Wald Coefficiens Tes 153 Table 2.15: Toal Effec of Inflaion Volailiy on Sovereign Deb Raing (Given Crediors righs) Table 2.16: Toal Effec of Inflaion Volailiy on Sovereign Deb Raing (Given Shareholders righs)..155 Table 2.17: Toal Effec of Inflaion Volailiy on Sovereign Deb Raing (Given Enforcemen) Table 2.18: Toal Effec of Inflaion Volailiy on Sovereign Deb Raing (Given Accouns) Table 2.19: Definiion of Moody s Sovereign deb raing..158 Table 2.20: Annual Yield in 2000 (firs quarer) and Moody s raing in Table 2.21: Correlaion Marix beween LEGAL1 and LEGAL Table 3.1: Values of he Calibraed Parameers for Mexico Table 3.2: Effec of he Drop in r* holding (a differen values) on he New Seady xii

13 Sae of Oupu per Worker (as a raio o is iniial level) Table 3.3: Effec of he Drop in holding r* (a 3.05%) on he New Seady Sae of Oupu per Worker (as a raio o is iniial level)..119 Table 3.4: Effec of he Drop in r* when coupled wih a drop in on he New Seady Sae of Oupu per Worker (as a raio o is iniial level) Table 3.5: Values of he Calibraed Parameers for Mexico for Barro e al. (1995) Model.172 Table 3.6: Effec of he Drop in r (holding a 0.20) on he New Seady Sae of Oupu per Worker (as a raio o is iniial level) in Barro e al. (1995) Model Table 3.7: Effec of he Drop in holding r (a 3.05%) on he New Seady Sae of Oupu per Worker as a raio o is iniial level in Barro e al. (1995) Model Table 3.8: Effec of he Drop in r* when coupled wih a drop in on he New Seady Sae of Oupu per Worker (as a raio o is iniial level) in Barro e al. (1995) Model.175 xiii

14 Lis of Figures Figure 1.a: Scaer Plo of he Relaion beween Inflaion Volailiy and Growh.135 Figure 1.b: Scaer Plo of he Relaion beween Inflaion Volailiy and Growh.136 Figure 1.c: Scaer Plo of he Relaion beween Inflaion Volailiy and Growh.137 Figure 1.1: Toal Effec of Inflaion Volailiy on Growh (Given LEGAL1) 141 Figure 1.2: Toal Effec of Inflaion Volailiy on Growh (Given LEGAL2) 141 Figure 1.3: Toal Effec of Inflaion Volailiy on Growh (Given Corrupion).142 Figure 1.4: Toal Effec of Inflaion Volailiy on Growh (Given Law and Order)..143 Figure 1.5: Toal Effec of Inflaion Volailiy on Growh (Given Bureaucracy)..144 Figure 1.6: Toal Effec of Inflaion Volailiy on Growh (Given Crediors righs) 145 Figure 1.7: Toal Effec of Inflaion Volailiy on Growh (Given Shareholders righs)..146 Figure 1.8: Toal Effec of Inflaion Volailiy on Growh (Given Enforcemen)..147 Figure 1.9: Toal Effec of Inflaion Volailiy on Growh (Given Accouns) 148 Figure 2.1: Toal Effec of Inflaion Volailiy on Raing (Given LEGAL2)...77 Figure 2.2: Toal Effec of LEGAL2 on Raing (Given Inflaion Volailiy)...81 Figure 2.3: The negaive relaionship beween Moody s raing in 2000 (firs quarer) And 5-year Annual Bond Yield in 2000 (firs quarer)..87 Figure 2.4: The Effec of Inflaion Volailiy on he Long Term Bond Yield.. 88 Figure 2.5: The Effec of LEGAL2 on he Long Term Bond Yield.88 Figure 2.6: Toal Effec of Inflaion Volailiy on Raing (Given Crediors righs)..154 Figure 2.7: Toal Effec of Inflaion Volailiy on Raing (Given Shareholders righs) 155 Figure 2.8: Toal Effec of Inflaion Volailiy on Raing (Given Enforcemen) 156 xiv

15 Figure 2.9: Toal Effec of Inflaion Volailiy on Raing (Given Accouns).157 Figure 2.10: Invarian LEGAL2 and varian LEGAL Figure 3.1: Inflaion Volailiy and LEGAL2 index Figure 3.2: Impulse Response Funcions for a 25% Permanen Drop in Figure 3.3: The Effec of Inflaion Volailiy on Welfare Figure 3.4: The Effec of LEGAL2 on Welfare..120 Figure 3.5: Transiion Pah for he Drop in r* from o (Holding Consan a 0.20)..124 Figure 3.6: Transiion Pah for he Drop in from 0.20 o 0.15 (Holding r* Consan a ) Figure 3.7 Transiion Pah for he Drop in r* from o Coupled wih he Drop In from 0.20 o Figure 3.7 Comparison of Experimens..129 xv

16 1 Par I INTRODUCTION A growing srand of lieraure focuses on he role of macroeconomic policies versus he role of insiuions in generaing growh in developing counries. While he policy view does no undermine he role of insiuions in economic growh, i argues ha he mos imporan facor for he divergence in growh beween counries of he world oday sems from he differences in he policies implemened by heir governmens. On he oher hand, he insiuions view argues ha he level of developmen in insiuions is he main facor explaining he divergence in growh beween counries. This view does no consider ha macroeconomic policy by iself is an imporan facor affecing growh. Insead macroeconomic policy is considered a channel hrough which insiuions affec growh. According o his view, bad macroeconomic policy is simply a resul of poor insiuions in he economy. This disseraion conribues o his debae of policies or insiuions by separaely quanifying he long-run welfare effecs of policy in paricular moneary policy- versus he long-run welfare effecs of insiuions. The resuls show ha alhough insiuions were found o conribue more o welfare in he long-run, he significan role of moneary policy canno be undermined. In addiion, he resuls show ha his significan role of moneary policy is no semmed from he underlying insiuions. The overall resul from his disseraion is ha inflaion volailiy, and no jus he level of inflaion, is an imporan facor in he deerminaion of growh and welfare for a developing economy bu ha he level of financial insiuions also is imporan and ha a developing economy

17 2 can ge bigger welfare gains by reforming heir financial insiuions a he same ime hey reduce inflaion volailiy in heir economy. Agains he above overview, he firs chaper of his disseraion analyzes he effec of inflaion volailiy on growh in he presence of differen degrees of insiuional developmen. A panel-daa growh regression using Sysem GMM procedure is used o conrol for endogeneiy and unobserved counry specific facors for a sample of 37 developed and developing counries over he period A non-linear growh regression specificaion ha ineracs inflaion volailiy wih legal and financial insiuions indices, developed by Chinn and Io (2005), is esimaed in his disseraion using he principal componen analysis for he sample and he ime period under sudy. The resuls of his chaper show ha inflaion does no have a saisical significan impac on growh. On he oher hand, i is he volailiy of inflaion ha has a saisically negaive significan growh effec. In addiion, and in conras wih he resuls of Acemoglu, Johnson, Robinson and Thaicharoen (2003) and Easerly (2004), he sudy finds ha inflaion volailiy does no ac as a proxy for insiuions. Improving insiuions will have a saisical significan posiive impac on growh which will help o reduce he negaive impac of inflaion volailiy. The magniude of he effec of insiuions on growh will be small and insignifican in insiuionally developed counries, bu relaively big and significan in insiuionally poor counries. The resuls of his chaper were robus o he use of differen esimaion mehodology. Having found a significan impac of inflaion volailiy on growh ha is separae from he significan impac of insiuions on growh, he second chaper of his disseraion analyzes he channel hrough which inflaion volailiy versus insiuions

18 3 affec growh. The chaper idenifies an imporan channel which is he invesmen channel, or he sovereign deb raing channel. To do ha, he chaper empirically analyzes of he impac of inflaion, inflaion volailiy, and financial insiuions on a counry s sovereign deb raing. Using a sample of 37 developed and developing counries over he period , he sudy esimaes a non-linear raing regression ha inerac inflaion volailiy wih he index for financial insiuions. The resuls sugges ha inflaion does no have a saisical significan effec on sovereign deb raing once inflaion volailiy has been included in he regression. In addiion, reducing inflaion volailiy can have a saisically and economically significan posiive effec on a counry s sovereign deb raing as compared o he level of inflaion. The resuls also show ha improving financial insiuions has a saisically significan posiive direc and indirec effec on a counry s sovereign deb raing. For insance he sudy finds ha for a developing counry, wih relaively poor financial insiuions and moderaely high inflaion volailiy, a one sandard deviaion decrease in he log of inflaion volailiy will lead o an increase in a counry s sovereign deb raing of abou wo classificaions. This increase in sovereign deb raing leads o a reducion in he average annual long-erm bond yield by abou 4.4%. On he oher hand, a one sandard deviaion increase in he financial insiuions index will lead o an increase in he raings class of abou one class, which in urn reduces he average annual long-erm bond yield by abou 4.27%. These empirical resuls were hen used in chaper hree o calibrae a moneary small open economy model of McCandless (2008) wih cash in advance consrain, capial adjusmen cos, foreign bonds, rade secor, and wih four ypes of shocks; moneary, foreign price, echnology, and insiuions shocks. The main goal is o quanify he

19 4 welfare implicaions under wo policy scenarios; 1) a one sandard deviaion decrease in inflaion volailiy and 2) a one sandard deviaion increase in he level of financial insiuions. The moneary small open economy model was calibraed using he quarerly daa for Mexico; a counry ha, by he measure used in he sudy has poor financial insiuions and moderaely high inflaion volailiy. The resuls of he model sugges ha a one sandard deviaion reducion in inflaion volailiy will increase welfare by a mos 11.4% while a one sandard deviaion improvemen in insiuions will lead o an increase in welfare in he range of 17% o 23%. The welfare implicaions of he moneary small open economy model were found o be much smaller when compared o he open economy model of Barro and Sala-i-Marin (1995) where he economy had neiher access o rade nor access o foreign bonds. Furher, simulaions were underaken in order o analyze he impulse response funcions of he economy o a permanen shock in insiuions. In addiion, he sudy analyzes he impac of he reducion in inflaion volailiy versus he impac of he improvemen in insiuions on he behavior of he endogenous variables along he ransiion pah. An imporan policy implicaion of his disseraion suggess ha developing economies can ge larger welfare gains from improving heir insiuions han from reducing inflaion volailiy. This policy implicaion does no undermine he role ha can be played by he moneary policy in growh, bu improving insiuions has a couple of posiive effecs; A direc effec on welfare hrough reducing he resource wase in he economy and an indirec effec on welfare ha acs hrough he financial insiuions impac on reducing he harmful impacs of inflaion volailiy on growh. Also, his policy

20 5 implicaion is inuiive in he sense ha reducing inflaion volailiy is cosly on oupu, while improving financial insiuions is a cos free policy.

21 6 Par II Chaper 1: Inflaion Volailiy, Insiuional Environmen And Economic Growh 1.1 Inroducion There are many sudies addressing he role of he economic policy versus he role of insiuions in growh. However, few of hese sudies have addressed he role of he second momens of he policy insead of he policy iself in affecing growh. In his sudy, he role of inflaion volailiy versus he role of insiuions in growh is sudied by adding he volailiy of inflaion and is ineracion erm wih legal and financial insiuions o he growh regression. The aim is o deermine how he oal effec of inflaion volailiy on growh is affeced by he level of insiuional developmen. The main conribuion of his chaper is ha i shows ha while he level of inflaion does no have a significan effec on growh, which is in line wih previous sudies; inflaion volailiy does significanly impac growh. Thus, inflaion volailiy does have an effec on growh even for counries wih moderaely high levels of inflaion. This chaper also finds ha improving eiher legal or financial insiuions will saisically significanly help o reduce he harmful effecs of inflaion volailiy on growh. In addiion and in conras o he resuls of Acemoglu, Johnson, Robinson and Thaicharoen (2003) and Easerly (2004), he chaper shows ha inflaion volailiy does no ac as a proxy for eiher ype of insiuions. Finally his chaper shows ha he conribuion of insiuions in reducing he harmful impacs of inflaion volailiy on growh will be

22 7 relaively small and insignifican in insiuionally developed counries, bu relaively big and significan in insiuionally poor counries. The Chinn and Io (2005) indices of legal and financial insiuions, LEGAL1 and LEGAL2 respecively, were esimaed in his chaper for he sample of counries and ime period under sudy. The LEGAL1 index was esimaed using he principal componen analysis from hree indices; bureaucracy, corrupion, and law and order. Similarly, he LEGAL2 index was consruced from four indices; he proecion of crediors righs, he proecion of shareholders righs, ransparency of he company s accoun, and srong enforcemen of laws. Using a cross-secion of 37 developed and developing counries over he period ( ), he resuls of his chaper show ha he impac of insiuions is significan up o he counries on he 80 h percenile for eiher LEGAL1 or LEGAL2. A one sandard deviaion reducion in inflaion volailiy will lead o an increase in growh of abou 0.93% and 0.90% in counries wih he wors level of legal insiuions and financial insiuions respecively. Also, a one sandard deviaion reducion in inflaion volailiy will lead o an increase in growh of only abou 0.13% and 0.08% for counries on he 80 h percenile of legal and financial insiuions respecively. Finally, for counries wih he bes level of eiher legal or financial insiuions, improving insiuions will no have any saisical significan effec on growh. The empirical model presened in his chaper is heoreically based on Choi, Smih and Boyd (1995). In heir paper a heoreical model was developed and he iner-linkages beween inflaion, marke fricions, inflaion volailiy, and growh were presened o prove ha high inflaion will lead o high inflaion volailiy which in urn leads o more

23 8 financial fricions in he marke and hereby inhibis growh. The ype of fricion assumed in his model is informaion asymmery. A high inflaion will reduce he real value of ineres raes making i cheaper for hose borrowers who were no borrowing a he high real ineres raes. Accordingly, here will be a higher demand for credis by his ype of borrowers. Credi owners on he oher hand would fear lending hem, and hey would prefer o raion credis. The more credi raioning he higher is he inflaion volailiy which in urn affecs he real aciviy. So he higher is inflaion, he more informaion asymmery, he more inflaion volailiy, and he lower is he real growh rae. The model used by Choi, Smih, and Boyd (1995) analyzes how he ransmission of he effec of inflaion o real growh hrough he financial marke depends on wheher we are o analyze a relaively high average inflaion counry vs. a relaively low average inflaion counry. A high average inflaion counry will relaively experience more financial fricions han a low average inflaion counry. For he former counry, a higher inflaion will exer a srong negaive impac on growh, while in he laer counry his negaive impac migh no appear a all. As noed in Choi, Smih, and Boyd (1995), he low average inflaion counry will experience a Mundell-Tobin effec, where a higher inflaion leads o more growh as credi raioning migh no appear a all in he financial marke of his counry. Accordingly, here is a hreshold limi for inflaion beyond which i will exer is negaive impac on growh. And his is because here is a hreshold limi for marke fricions beyond which i will adversely affec growh. The relaion beween inflaion and growh has been exensively sudied in many empirical papers. Alhough he general consensus among economiss is ha inflaion has a saisical significan negaive impac on growh, as in he work of Fisher (1993),

24 9 Moley (1994) and Barro (1995) have showed, oher economis believed ha his resul is no robus. The work of Levine and Renel (1992) showed ha inflaion growh relaion is very sensiive o he regression specificaion. In a cross secion sudy by Levine and Zervos (1993), hey found ha he growh-inflaion relaion depends on few ouliers in heir sample of counries; basically Uganda and Nicaragua. Also, in he sudy of Bruno and Easerly (1995), hey found ha he growh inflaion relaion depends on a hreshold level of inflaion of 40% where in counries wih annual inflaion below his percenage inflaion will have an insignifican impac on growh. In addiion, he sudy of Clark (1997) concluded ha alhough he resuls show a general saisical significan negaive relaion beween inflaion and growh, his relaion is however no robus o eiher he change in he sample of counries or o he change in he ime period under sudy. The harmful impacs of inflaion volailiy insead of he level of inflaion has been highlighed long ime ago in he work of Friedman (1977) where he found ha alhough he long erm moneary policy neuraliy holds in he level of he policy, i does no hold in is second momen. The work of Levine and Zervos (1993) sudied he separae effecs of he level and he volailiy of inflaion on growh. They concluded ha neiher he level nor he volailiy of inflaion is robusly correlaed wih growh. On he oher hand, using panel daa Orphanides and Judson (1996) found ha boh inflaion and inflaion volailiy were robusly negaively correlaed wih growh for he high inflaion counries and hey concluded ha a good policy is a one ha is able o reduce he level of inflaion and a he same ime sabilize inflaion. In addiion, using he square roo of he condiional variance of inflaion from a GARCH(1,1) model as he measure of inflaion uncerainy, Coulson

25 10 and Robbins (1985) found a posiive associaion beween his measure of inflaion uncerainy and US economic performance. Jansen (1989) on he oher hand found no significan relaionship beween he wo variables. Coulson and Robbins (1985) resuls were however no also confirmed by Grier and Perry (2000) and Grier, Henry, Olekalns, and Shields (2004) who repored a negaive relaion beween inflaion uncerainy and growh in he US. The lieraure on he imporance of srong insiuions, eiher legal or financial insiuions, is rapidly growing. This lieraure sared wih he work of La Pora, Florencio, Andrei and Rober (1997) where hey showed he imporance of legal and financial insiuions for firm s decisions. Rodrik (1999) showed ha counries wih he sharpes drop in growh afer 1975 are hose counries wih weak insiuions which was is measured by rule of law, democraic righs, and social safey nes. Acemoglu e al.(2003) showed ha insiuions maer more o growh han does he economic policy. Once insiuions variable has been included in he regression, he coefficien of he macroeconomic policy urns insignifican. These resuls were confirmed by Easerly (2004). In conras o hese resuls of Acemoglu e al. (2003) and Easerly (2004), in a cross secion of 91 counries Faas and Mihov (2005) sudied he effec of fiscal policy volailiy, insiuions and growh. The resuls of heir paper showed ha fiscal policy volailiy has a significan negaive impac on growh and in addiion hey showed ha insiuions affec growh only hrough is effec on he policy and paricularly he volailiy of he policy.

26 11 In line wih he resuls of Acemoglu e al.(2003) and Easerly (2004), using a panel daa on hundred counries over he period Veiga and Aisen (2006) found a posiive associaion beween greaer fragmenaion, polarizaion, and poliical insabiliy, as forms of marke fricions and inflaion volailiy. And ha hese forms of marke fricions consiue he main deerminans of inflaion volailiy. On he micro-economic level, Johnson, McMillan, and Woodruff (2002) showed from heir survey on new firms in pos communis counries ha srong propery righs affecs he firms decision o reinves heir profis. Also, using daa on housands of firms in 80 counries, Albero and Mark (2006) concluded ha volailiy has a negaive effec on he firm s growh and his negaive effec is magnified wih weak insiuions. In heir paper volailiy is defined as he volailiy in he firm s decisions o ener new projecs which in urn depends on he governmen inabiliy o creae a credible policy. And insiuions On he oher hand were defined as high enry barriers. Beck, Demirguc-Kun, Laeven, and Levine (2006) found ha he financial consrains faced by firms will depend on he firm s size. Large firms are expeced o face less financial consrains. In addiion, hey found ha firms locaed in counries wih well developed legal and financial insiuions are facing less financial consrains and hese insiuions will be basically he driving force for economic developmen. Agains he above background, in wha follows he empirical specificaion of a model is presened ha exends previous sudies and demonsraes ha i is he volailiy of he policy and no he level of he policy ha has more impac on growh. Nex he growh regression model is expanded by including he effec of insiuions and is hen used o

27 12 compue he oal effec of volailiy of inflaion on growh under differen degrees of insiuional developmen. The sudy consiss of he following secions; secion 1.2 presens he empirical specificaion of he model, secion 1.3 describes our daa, secion 1.4 discusses he esimaion resuls, secion 1.5 calculaes he oal effec of inflaion volailiy on growh, secion 1.6 presens wo robusness checks, secion 1.7 concludes he chaper. Finally, Appendix (I) is by he end of he disseraion. 1.2 Empirical Specificaion Mos of he researches in economic growh focusing on he endogenous growh models have sared a early 1990 s. These models basically assume ha growh arises from human capial accumulaion. The endogenous growh models were found o be imporan for analyzing he differen governmen policy measures on long run growh in he economy. Among many oher economiss, Lucas (1988), Romer (1990), King and Rebelo (1990), and Rebelo (1991) were he firs o analyze he long run implicaions of hese models on economic growh. In line wih he endogenous growh heory models and for he purpose of he analysis of his chaper, his secion empirically esimaes a growh model ha is a funcion of he iniial level of per capia GDP, he level of inflaion, volailiy of inflaion, invesmen in human capial, and insiuional developmen. In addiion, he empirical growh model of his secion includes a proxy for rade openness which was found by many endogenous

28 13 growh models as in Young (1991), Grossman and Helpman (1991), Eicher (1993), and Lee (1993) o posiively conribue o long-run growh. The main focus of he empirical model of his chaper is o analyze he effec of inflaion volailiy on growh in a model ha conrols for he level of legal and insiuional variables. Following Caselli, Equivel and Lefor (1996), Holz-Eakin, Newey and Rosen (1988) and Arellano and Bond (1991), he growh model is esimaed using dynamic panel Sysem GMM. Equaion (1.1) provides he general form of he empirical model; y i, yi, 0 y 1CVi, i d i,, (1.1) where he subscrips i and represen he counry and he ime period respecively. The variable y i, is he log GDP per capia. Hence, he lef hand side of he above Equaion represens he log difference of he real GDP per capia over a period of years, in oher words i represens he growh rae over he period, where = 3. The explanaory variables consis of y, which represens he beginning of he period GDP per capia, i CV i, is he se of conrol variables ha are measured eiher a he beginning of he period or as an average over he period. This se of conrol variables include level of inflaion a he beginning of he period, log inflaion volailiy a he beginning of he period, he average over he period of he curren accoun balance as a percenage o GDP as a proxy of rade openness, and he average over he period years of schooling as a proxy of human capial invesmen. This is in addiion o he counry specific effecs and ime period dummies ha are represened by i and d respecively.

29 14 Worh noing ha he counry specific effec can pick he impac of eiher economic or financial crises ha affeced abou seven of he counries included in he sample. Also, he ime period dummy can pick he impac of he conagion effecs of a crisis on financial markes in developed or developing counries. The Sysem GMM proposed by Arellano and Bover (1995), Blundell and Bond (1998), and Blundell, Bond, and Windmeijer (2000) over comes he bias problems of he difference GMM esimaor. I works by basically sacking ogeher Equaion (1.1) above wih Equaion (1.2) below, ( yi, yi,, ) ( yi, yi, 2 ) 0( y y 2 ) 1( CVi, CVi, 2 ) d ( i, i ). (1.2) The main reason for differencing Equaion (1.1) is o eliminae he counry specific or unobserved effecs following Arellano and Bond (1991). In addiion o he assumpions of he Difference GMM 1, he Sysem GMM assumes ha he firs difference of he dependen variable and he explanaory variables are uncorrelaed wih he counry specific effecs which provides he following wo exra momens condiions abou he correlaion beween he dependen variable and he error erm and he se of he independen variables and he error erm, E ] 0, For = 2, T [ y i, i, E ] 0, For = 2, T, (1.3) [ X i, i, 1 More deails on Difference GMM are available in Traub (2006).

30 15 where X i, is he se of all he explanaory variables of Equaion (1.1) which includes y i,, CV i,. In order o saisfy hese addiional momens condiions, he correc se of insrumens was chosen by regressing he dependen variable and each of he explanaory variables in levels, each variable in a urn, on all he possible lags of he dependen variable and he explanaory variables in firs difference saring from he second lag in addiion o ime dummies. The insrumens were chosen in such a way ha a unique se of insrumen for y, and X i, can be found. This se of insrumens usually consiss i of he second lag of y, and X i, plus any oher exra insrumens ha appear o be i also significan. For he se of insrumens o be valid, i has o be boh relevan and exogenous o he error erm. The relevance es checks and makes sure ha he firs sage F-saisic equals or exceeds 10 or in oher words ha he bias of he Two Sage Leas Squares is a mos 10% of he bias of he Ordinary Leas Squares mehodology 2. The exogeneiy es or Over Idenificaion Tes, or as know by Sargan es (Sargan 1988), ess ha he momens condiions of Equaion (1.3) above are saisfied. To find he relevan se of insrumens i is imporan o sar by a small model in which lags of y i, and he componens of he marix X i, are regressed, each one in a urn, on he y, and X i, saring from he second up o he fourh lag. The significance i of he las lag of y i, only is hen checked. If i shows o be significan, is fifh lag is o be added. Bu if i is no significan, he fourh lag is removed and he model is reesimaed again wih lags saring from he second up o he hird of i y, while keeping 2 The book of Sock & Wason (2006) provides good explanaion on he TSLS.

31 16 he lags of X i, unchanged (i.e. second up o he fourh lag). The process is repeaed unil he righ se of insrumens for he componens of he marix y i, is reached. Nex, he same process is repeaed for X i, each one in a urn, while keeping he se of insrumens found for he y i, model unchanged. Afer he process of finding he correc se of insrumens for all he explanaory variables of he marix X i, is over, i is very imporan o check again ha he final chosen se of insrumens is significan in he y i, model. If he las lag of y i, is confirmed o be sill significan his would mean ha he correc se of insrumens is finally found. If on he oher hand, he las lag of i y, urns ou o be insignifican, one lag has o be reduced and he model is o be re-esimaed. The process is repeaed unil he correc se of insrumens is found. The significance of each insrumen is judged by is -saisic and also by he F-saisic of he whole regression. According o his process, he mos relevan se of insrumen for he base growh regression is he second and he hird lag of; he beginning of he period GDP, he beginning of he period inflaion, and he beginning of he period log inflaion volailiy. This is in addiion o he second lag of he average rade volume as a percenage of GDP, and he second lag of he average over he period years of schooling. Following Chang e al (2005) and Traub (2006) he base model is expanded by including ineracion erms. The ineracion erms include he legal and financial insiuions indices and heir seven sub-indices each one is ineraced in a urn. The objecive is o esimae he oal effec of inflaion volailiy on growh when i is ineraced wih hese legal and financial insiuions variables (each one in a urn) and o analyze which ineracion erm conribues more and significanly o reducing he

32 17 negaive effecs of inflaion volailiy on growh. The ineracion erm is esimaed by adding 3( L1 i, * Infvol) o he righ hand side of Equaion (1.1) as shown below. y i, y i, y 0 3 CV 1 i, Infvol i, L1 i, * Infvoli, i d i, 2 (1.4) where Infvol, is he beginning of he period inflaion volailiy and i is aken ou of i he CV i, se and added o he regression above. The erm L i, 1 represens he legal insiuions index, or LEGAL1. Similarly, Equaion (1.1) will be expanded by adding he ineracion erm 3( L2 i * Infvol) which represens he indirec effecs of financial insiuions index LEGAL2 and he equaion is shown as follows; y i, y i, y 0 CV 1 3 i, Infvol 2 i, L2 i * Infvoli, i d i,. (1.5) The oal effec of inflaion volailiy on growh will hen be esimaed by adding he coefficien of inflaion volailiy 2 o he coefficien of he ineracion erm 3 imes he facor ineraced wih volailiy. Accordingly, in Equaion (1.4) he oal effec of inflaion volailiy equals 2 3 L 1i, equivalenly * when ineraced wih he LEGAL1 index and 2 3 * L2 i in Equaion (1.5) when ineraced wih he LEGAL2 index.

33 Daa The daa se was consruced as a panel of counry observaions from he daa base of he World Bank World Developmen Indicaors, and includes 37 counries over he period In order o analyze he long-erm effecs of inflaion volailiy on growh, he daa were averaged ino hree years ime periods. Daa are hus available for six ime series observaions for each counry. The se of conrol variables is seleced considering boh heir imporance as growh deerminans per se and heir poenial for affecing he growh response of inflaion volailiy. The conrol se includes variables ha vary boh across counries and over ime, as well as variables ha vary only across counries. Among he former, he se includes he beginning of he period inflaion rae, he beginning of he period inflaion volailiy, he average over he period curren accoun balance as a percenage of GDP as a measure of rade openness, he beginning of he period years of schooling o accoun for human capial invesmen, and he LEGAL1 index and is hree componens; bureaucracy, corrupion, and law and order. Among he variables ha vary only across counries, he se includes he LEGAL2 index and is four indices; crediors righs, shareholders righs, enforcemen, and accouns. The growh daa was compued as he log difference of GDP per capia normalized by he lengh of he period. Daa for he years of schooling were colleced from he Barro and Lee (2000) daa se. Following Coulson and Robins (1985), he daa on inflaion volailiy was calculaed as he log of he square roo of he condiional variance series of inflaion calculaed by GARCH(1,1) model. As noed in Clark (1997) Measuring

34 19 inflaion volailiy as he coefficien of variaion of he level of inflaion provides an assurance ha he level of inflaion is no correlaed wih is variance and hence will no pose any imperfec mulicollineariy issues in esimaing he growh regression when boh he level and volailiy of inflaion are included. As menioned previously, he Chinn and Io (2005) indices, LEGAL1 and LEGAL2 were esimaed for he sample and ime period under sudy in his disseraion. These wo indices were developed using he principal componen analysis. Concerning he firs ype of insiuions index, LEGAL1 3, i conains he measures relaed o he general developmen of he legal sysems. I was esimaed by he principal componen analysis of four indices namely corrupion, law and order and bureaucracy. The corrupion index ranges from 1 o 6, where a highly corruped counry will ake an index of 1 and he lowes corruped counry will ake an index of 6. For insance he Belgium has an index of 6 while Egyp has an index of 1. Nex, he law and order index ranges from 1 o 6, where he higher he index reflecs a sricer criminal jusice sysem. This index refers o a poliical plaform which suppors a sric criminal jusice sysem, especially in relaion o violen crime and propery crimes, hrough harsher criminal penalies. These penalies may include longer erms of imprisonmen, mandaory senencing, and in some counries, capial punishmen. Finally, he Bureaucracy index ranges from zero o 4, where he higher he index he less is he bureaucracy in he economy. The daa on he hree legal indices were colleced from La Pora e al. (1998). The LEGAL1 index combines hose 3 The firs eigenvecor of LEGAL1 was found o be (Bureaucracy, Corrupion, Law & Order) = (0.269, 0.041, 0.671, 0.689). This index is normalized, cenered a zero wih a sandard deviaion equal o one.

35 20 hree indices using he principal componen analysis and i ranges from up o 1.73 where again he higher he index he more developed he legal insiuions. The second ype of insiuions index, LEGAL2 4, i conains legal indices ha govern financial ransacions. From here afer i will be referred o as he financial insiuions index. Again his index was esimaed using he principal componen analysis of he four indices namely proecion of crediors righs, effeciveness of he legal sysem in enforcing conracs, proecion of shareholders righs, and comprehensiveness of company repors. The index of crediors proecion ranges from 0 o 4, where more proecion for crediors implies a higher index. I is composed of he variables ha incorporae he auomaic say proposiion on he asses of a failing firm, he coninuaion of he old managers in a reorganizaion process, resricions for going ino reorganizaion and he senioriy sysem of secured crediors. Nex, he index of he degree of law enforcemen ranges from 4.87 o 9.99 where again he higher he index he sricer is he sysem of law enforcemen. This index consiss of he average of he efficiency of judicial sysem, rule of law, risk of expropriaion, and risk of conrac repudiaion. Concerning he hird sub-index, he index of shareholders righs, i ranges from 0.05 o 5.10 where he higher he index he more proecion of shareholders righs. This index is composed of he sum of he one share-one-voe, proxy by mail, shares no blocked before meeing, cumulaive voing/proporional presenaion, oppressed minoriies, preempive righ o new issues and percenage of share capial o call an emergency shareholder meeing less han en percen 5. Finally, he accouns index ranges from 24 o 83 where a 4 The firs eigenvecor of LEGAL2 was found o be (Credior, Shareholder, Enforcemen, Accoun) = (0.269, 0.041, 0.671, 0.689). This index is normalized, cenered a zero wih a sandard deviaion equal o one. 5 More deails on hese sub-indices are provided in La Pora e al. (1998).

36 21 higher index implies more ransparency and comprehensiveness of he repors. The daa on hese four sub-indices of he LEGAL2 index were colleced from he Inernaional Counry Risk Guide- ICRG (counry daa). The LEGAL2 index ranges from up o I is imporan o noe ha he counries of he daa se were chosen on he basis of he availabiliy of he daa on he componens of LEGAL1 and LEGAL2. The resuls migh hen be seen as conservaive since he consraining he daa se o only 37 counries migh cause sample selecion bias. The lis of counries included in he sample is repored in Table 1.5 of Appendix (I). As shown in his able, he counries included in he sample consis of a se of low income developing counries and a se of developed counries. Mos of he OECD counries were included in he sample. Also many of he counries in he sample are Lain American counries. In addiion, he sample includes some counries which faced economic or financial crises in he pas like Argenina ( ), Brazil (1980 s), Mexico (1994), Uruguay (2002), Korea (1997), Thailand (1997), and Turkey (1980 s, 1994, ). Figure 1.a, 1.b and 1.c of Appendix (I) plos he scaer plo of he ime series daa of inflaion volailiy (on he y-axis) agains he daa of GDP growh (on he x-axis) over he period of he sudy. As can be noiced from he spread of he volailiy and growh combinaions on he graphs, he negaive expeced relaionship beween hese wo variables was obvious in mos of he counries bu was hard o be noiced in some ohers. Also, from hese graphs i can be noiced ha hree counries have experienced relaively high volailiy over he period of he sudy Argenina, Brazil, and Peru. These counries

37 22 were excluded from he sample based on a non-subjecive crierion by excluding he op 10% of he log inflaion volailiy disribuion. 1.4 Esimaion Resuls Using dynamic panel wo-sep Sysem GMM wih ime dummies in he level equaion and wih using he finie sample correcion for he variance marix proposed by Windmeijer (2005), he growh model is esimaed for he panel of 37 counries over he period 1989 o The esimaion resuls of he benchmark model are presened on Table 1.1 below. The Table consiss of five columns represening he resuls of five regressions. In each regression, he dependen variable is per-capia real GDP growh. As menioned previously, in esimaing he model he op 10% of he volailiy disribuion was discarded from he daa. This is basically o remove he impac of he ouliers on he esimaion bias in he resuls. The resuls presened in Column 1 shows an expeced saisical significan negaive effec of he iniial level of GDP a he beginning of he period on growh over he period. As he beginning of he period GDP increases by 1%, growh over he period decreases by abou 0.62%. This resul confirms he convergence hypohesis; counries ha are relaively advanced in heir developmen process will grow slower han hose counries lagged behind. Adding he beginning of he period inflaion o he regression as shown in Column 2 of he same able, he coefficien of inflaion does no show any sign of saisical significance. This resul goes in line wih he implicaion of he heoreical model of

38 23 Choi, Smih, and Boyd (1995) on which he empirical model of his chaper is based, ha here is a hreshold limi for inflaion beyond which i will exer is negaive impac on growh. Table 1.1 The Base Model of Economic Growh Cross-counry panel daa consising of non-overlapping 3-year averages ( ). Esimaion Mehod: Sysem GMM Esimaion [1] [2] [3] [4] [5] Consan 3.154*** (0.620) 2.796** (1.235) 2.762** (1.298) 2.936*** (0.819) 4.093*** (1.153) Iniial *** ** ** *** *** GDP/capia (0.135) (0.246) (0.257) (0.169) (0.332) Inflaion (0.091) (0.109) (0.072) (0.085) Volailiy *** (0.164) *** (0.013) *** (0.015) Openness 0.028* (0.015) (.029) Human Capial Invesmen 0.176** (0.072) Counries/Obser 34/161 34/161 34/161 34/161 34/161 vaions Sargan Tes, p-value [0.30] [0.35] [0.33] [0.49] [0.36] Noes: Numbers below coefficiens are he corresponding robus sandard errors. (***), (**) and (*) denoes saisical significance a he 1%, 5% and 10% levels respecively. Number beween [.] are he p-values of he Sargan Tes. In addiion he insignificance of he coefficien of inflaion confirms wih he findings of Bruno and Easerly (1995) ha excluding inflaion levels above 40% from heir sample, inflaion will no have any sign of saisical significance o growh. As noed in heir paper if we omi he se of counries ha had inflaion a some poin pass 40 percen, hen we fail o deec any significan associaion beween inflaion and growh. By looking a he daa on he level of inflaion for he 37 counries over he period of he sudy, i is found abou 93% of he inflaion observaions are below 40% which is considered a relaively no high level of inflaion ha could lead o harmful

39 24 effecs on growh. This also goes in line wih he resuls of Bullard and Keaing (1994) ha inflaion will exer is negaive impac on growh only in counries wih relaively high iniial level of inflaion. Also, he resuls of Clark (1997) ha inflaion is no robusly negaively affecing growh. And as noed in Levine and Renel (1992), he relaion beween inflaion and growh is no robus o he changes in; he regression specificaion, he sample of counries or he ime period under sudy. Adding inflaion volailiy o he model, Column 3 shows ha is coefficien is saisically significan wih an expeced negaive sign where a one percen increase in inflaion volailiy a he beginning of he 3 period will lead o a reducion in growh over he period by abou 0.042%. This resul is considered an imporan conribuion of his chaper, where while he level of inflaion does no have a significan effec on growh, which is in line wih previous sudies; inflaion volailiy does significanly impac growh. Column 4 shows he resuls of adding he proxy for openness o he growh regression, as can be noiced from he able above, he coefficien of he iniial level of GDP and he coefficien of he volailiy of inflaion boh remained significan and wih he expeced sign. In addiion, he coefficien of openness appeared o be also significan and wih he expeced posiive sign. A one percen increase in he percenage of he curren accoun o GDP a he beginning of he period will lead o abou 0.028% increase in growh over he hree years period. Finally, Column 5 shows he resuls of adding he beginning of he period years of schooling o accoun for human capial invesmen. As obvious from above, he