essays on financial liberalisation, financial crises and

Essays on Financial Liberalisation,Financial Crises and Economic Growth

A thesis submitted to The University of Manchester for the degree ofDoctor of Philosophy

in the Faculty of Humanities

2014

Zeeshan Atiq

School of Social SciencesEconomics

Contents

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Copyright . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1 Thesis Introduction 101.1 Financial Development and Economic Growth: The Role of Financial

Liberalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.2 What Makes Financial liberalisation Work: The Impact of Sequencing

on Financial Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 Financial Development and Economic Growth: The Role of Financial Liber-alisation 182.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.2.1 Financial Development and Economic Growth . . . . . . . . . . 212.2.2 The Relationship between Financial Liberalisation and Economic

Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.2.3 The Impact of Financial Liberalisation on Finance-Growth Rela-

tionship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.3.1 Description of the Data . . . . . . . . . . . . . . . . . . . . . . . 282.3.2 Summary Statistics and Correlations . . . . . . . . . . . . . . . . 372.3.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462.4.1 Estimation Results of Baseline Model . . . . . . . . . . . . . . . 462.4.2 Regression with Interaction . . . . . . . . . . . . . . . . . . . . 492.4.3 Tests for Instruments . . . . . . . . . . . . . . . . . . . . . . . . 53

2.5 Robustness Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

2

CONTENTS 3

2.5.1 Exclusion of Outliers . . . . . . . . . . . . . . . . . . . . . . . . 552.5.2 Alternate Measures of Financial Development . . . . . . . . . . . 572.5.3 Reconstruction of the Liberalisation Index . . . . . . . . . . . . . 59

2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

3 What Makes Financial liberalisation Work: The Impact of Sequencing onFinancial Crises 643.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3.2.1 What Causes Systemic Fragility and Financial Crises? . . . . . . 673.2.2 What should Optimal Financial Reforms Achieve? . . . . . . . . 693.2.3 Sequencing of Financial Reforms and Crises . . . . . . . . . . . 70

3.3 Sequencing of Reforms: Choice of Benchmark . . . . . . . . . . . . . . 733.4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

3.4.1 The Dependent Variable: Financial Crises . . . . . . . . . . . . 783.4.2 The Main Explanatory Variable of Interest: Distance . . . . . . . 793.4.3 Control Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 81

3.5 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 833.5.1 Main Methodology . . . . . . . . . . . . . . . . . . . . . . . . . 833.5.2 Alternative Specification . . . . . . . . . . . . . . . . . . . . . . 84

3.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 863.6.1 Main Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 863.6.2 Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

4 Thesis Conclusion 1004.1 Main Findings and Policy Issues . . . . . . . . . . . . . . . . . . . . . . 1024.2 Areas for Further Research . . . . . . . . . . . . . . . . . . . . . . . . . 105

Appendix A 108

Appendix B 125

Total Word Count: 29116

List of Tables

2.1 Principal Component Analysis for Financial Development Index for theYear 1973 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2.2 Financial Liberalisation Index . . . . . . . . . . . . . . . . . . . . . . . 39

2.3 GDP/Capita Growth and Financial Development . . . . . . . . . . . . . . 40

2.4 Financial Liberalisation and Financial Development . . . . . . . . . . . . 41

2.5 GDP/Capita Growth and Financial Liberalisation Index . . . . . . . . . . 42

2.6 Baseline Regresssion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

2.7 Financial Liberalisation, Financial Development and Economic Growth,One-Step System GMM . . . . . . . . . . . . . . . . . . . . . . . . . . 50

2.8 Financial Liberalisation, Financial Development, and Economic Growth,Two-Step System GMM . . . . . . . . . . . . . . . . . . . . . . . . . . 54

2.9 Financial Liberalisation, Financial Development, and Economic Growth,One-Step System GMM without Oultiers . . . . . . . . . . . . . . . . . 56

2.10 Financial Liberalisation, Financial Development, and Economic Growth,One-Step System GMM with PVT/Y . . . . . . . . . . . . . . . . . . . . 58

2.11 Financial Liberalisation, Financial Development, and Economic Growth,One-Step System GMM with LLR . . . . . . . . . . . . . . . . . . . . . 60

2.12 Estimates using Financial Liberalisation Index . . . . . . . . . . . . . . . 61

3.1 The Sequence of Full liberalisation in Sample Countries . . . . . . . . . 74

3.2 Frequency of Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

3.3 Marginal Effects using Probit for Liberalisation Dummy . . . . . . . . . 87

3.4 Probit Model for Chile using Sum of Square Deviations . . . . . . . . . . 90

3.5 Probit Marginal Effects for All Countries With and Without Country Dum-mies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4

LIST OF TABLES 5

3.6 Results for Alternative Specification . . . . . . . . . . . . . . . . . . . . 94

3.7 Marginal Probit Effects for Group of Countries . . . . . . . . . . . . . . 95

3.8 Robustness with Different Specifications and Definition of Our Indepen-dent Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

A.1 Literature on Relationship between Financial Liberalisation, FinancialDevelopment and Growth . . . . . . . . . . . . . . . . . . . . . . . . . . 108

A.2 List of Sample Countries . . . . . . . . . . . . . . . . . . . . . . . . . . 112

A.3 Principal Component Analysis for Financial Development . . . . . . . . 113

A.4 Principal Component Analysis for Financial Development Vector 1 . . . . 114

A.5 Cross-Correlation Table for Components of FD . . . . . . . . . . . . . . 115

A.6 Cross-Correlation Table . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

A.7 Data Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

A.8 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

A.9 Cross-Correlation Table . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

A.10 Financial Liberalisation, Financial Development, and Economic Growth,Two-Step System GMM without Outliers . . . . . . . . . . . . . . . . . 120

A.11 Financial Liberalisation, Financial Development, and Economic Growth,Two-Step System GMM with PVT/Y . . . . . . . . . . . . . . . . . . . 121

A.12 Financial Liberalisation, Financial Development, and Economic Growth,One-Step System GMM with PVT/Y without Outliers . . . . . . . . . . . 122

A.13 Financial Liberalisation, Financial Development, and Economic Growth,Two-Step System GMM with LLR . . . . . . . . . . . . . . . . . . . . . 123

A.14 Financial Liberalisation, Financial Development, and Economic Growth,One-Step System GMM with LLR without outliers . . . . . . . . . . . . 124

B.1 Number of Reversals after Full liberalisation . . . . . . . . . . . . . . . . 125

B.2 Average Index of Financial Reforms . . . . . . . . . . . . . . . . . . . . 126

B.3 List of Sample Countries . . . . . . . . . . . . . . . . . . . . . . . . . . 127

B.4 Data Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

B.5 Criteria to Define Liberalisation Periods . . . . . . . . . . . . . . . . . . 129

Abstract

This thesis investigates the impact of financial liberalisation policies on finance-growth relationship and financial crises. Analysis of recent trends and economic perfor-mance of financially developed and stable economies raises at least two very importantquestions that seem to have strong analytical connections. The first question is associatedwith the link between financial development and economic growth and the second ques-tion focuses the possible association between the policies of financial liberalisation andfinancial vulnerability. In this thesis we aim to shed light on some of the aspects that havegained so much attention from academics and policy makers during the last two decades.

First we address whether excessive liberalisation has caused financial developmentto lose its effectiveness in generating economic growth. We employ a dynamic panel dataanalysis for 88 countries over the period of 1973 to 2005. Our index for the financialsector liberalisation covers seven aspects: credit controls and reserve requirements, inter-est rate controls, entry barriers, state ownership, policies on securities markets, bankingregulations and restrictions on capital market. We use a comprehensive financial de-velopment indicator constructed through principal component analysis of five differentindicators: bank private credit to GDP ratio, liquid liability to GDP ratio, deposit moneybank assets to total bank assets ratio, deposit money bank assets to GDP ratio, and bankcredit to bank deposit ratio. The results indicate that the positive effect of financial de-velopment on long-run growth continues to decline as the financial sector becomes moreliberalised. Our results are robust to changes in the financial development indicators andthe dis-aggregation of the financial liberalisation index.

Second, we examine the possibility for an optimal sequence of financial sector re-forms that may reduce an economy’s vulnerability to financial crises. We construct adistance measure from the countries that followed a more gradual approach and liber-alised their capital account at a later stage. Our analysis shows that the experience of thecountries that delayed or followed a very gradual approach for the liberalisation of theircapital accounts have high level of implications to those countries that allowed for shockapproach or liberalised their capital account before bringing reforms in other sectors.Key words: Financial system; Financial liberalisation; Financial Development; EconomicGrowth; Financial CrisesJEL classification: E44,G28,O16, O11, G01

6

Declaration

I declare that no portion of the work referred to in the thesis has been submitted insupport of an application for another degree or qualification of this or any other universityor other institute of learning.

7

Copyright

i. The author of this thesis (including any appendices and/or schedules to this thesis)owns certain copyright or related rights in it (the Copyright) and s/he has given TheUniversity of Manchester certain rights to use such Copyright, including for admin-istrative purposes

ii. Copies of this thesis, either in full or in extracts and whether in hard or electroniccopy, may be made only in accordance with the Copyright, Designs and Patents Act1988 (as amended) and regulations issued under it or, where appropriate, in accor-dance with licensing agreements which the University has from time to time. Thispage must form part of any such copies made

iii. The ownership of certain Copyright, patents, designs, trade marks and other intellec-tual property (the Intellectual Property) and any reproductions of copyright works inthe thesis, for example graphs and tables (Reproductions), which may be described inthis thesis, may not be owned by the author and may be owned by third parties. SuchIntellectual Property and Reproductions cannot and must not be made available foruse without the prior written permission of the owner(s) of the relevant IntellectualProperty and/or Reproductions.

iv. Further information on the conditions under which disclosure, publication and com-mercialization of this thesis, the Copyright and any Intellectual Property and/or Re-productions described in it may take place is available in the University IP Policy(see http://documents.manchester.ac.uk/DocuInfo.aspx? DocID=487), in any rele-vant Thesis restriction declarations deposited in the University Library, The Univer-sity Library’s regulations (see http://www.manchester. ac.uk/library/aboutus/regulations)and in The University’s policy on presentation of Theses.

8

Acknowledgments

First of all I thank to Almighty Allah who gave me skills and blessed me with all Hisbounties. I also express my deep gratitude to the supervisory committee that includes: Dr.M. Emranul Haque and Dr. Indranil Dutta who guided me all the way to complete thiswork. I also extend my gratitude to all my seniors, faculty and class fellows for the co-operation they extended in completion of my work. My special thanks go to Dr. KyriakosC. Neanidis, Dr. Matias Cortes, Dr. Kobil Ruziev (University of the West of England),Jonathan Gibson, Dr. Abdilahi Ali, and Dr. Mohammad Mehboob for their valuableinputs and help in my thesis. I owe special thanks to my family back home and my wife,my son, who bore all pain of being alone and tolerated my absence during this time. Ispecially thank to my wife, Sehar Zeeshan, who stood with me all the time and especiallyduring depressing moments and gave me courage and celebrated even my not-so-specialachievements. I can not return what my mother and late father have done over the yearsfor me. Their prayers and constant support with invaluable guidance has brought me tothe place where I am. I also thank from my depth of my heart to my siblings. I owe,special thanks to my eldest sister, Lubna Atiq, and her husband, Dr. Nadeem Khan whosesupport has been invaluable throughout my work and stay in the United Kingdom.

9

Chapter 1

Thesis Introduction

This thesis investigates the impact of financial liberalisation policies on the finance-growth relationship and financial crises. Analysis of recent trends and economic perfor-mance of financially developed and stable economies raises at least two very importantquestions that seem to have strong analytical connections. The first question is associatedwith the link between financial development and economic growth and the second ques-tion focuses on the possible association between the policies of financial liberalisationand financial vulnerability.

The most obvious reason for these questions is the number of recurrent crises inthe most developed countries in the world. These crises have raised questions about theability of the modern financial system, post-liberalisation, to manage its inherent risks. Inthis context the literature favouring the positive role of financial development in promot-ing economic growth seems very limited and fragmented. The recent crisis of 2007-08has proved the non-linearity of the relationship between finance and development and hasshown that the policy of excessive liberalisation does not necessarily always affect therelationship between growth and finance positively. Now there is rising evidence of in-significant (Aghion et al., 2004) and negative (Arcand et al., 2012) relationship betweenfinance and growth.

The recent literature has taken up the issues related to the roots, causes, and im-plications of the financial crises very seriously. However, there is still inadequate litera-ture available that studies the link between financial liberalisation and the finance-growthnexus. Moreover, in view of the role of short term capital flows in the emergence of recentcrises, there is a need to inquire extensively how capital controls and sequencing of thefinancial reforms play role in emergence of financial crises. The literature available for

10

Chapter 1. Thesis Introduction 11

sequencing anlayses the issue of sequencing using anecdotal evidence without introduc-ing a quantitative measure of sequencing. This seems to be less efficient in analysing thetrue impact of sequencing on financial crises.

To deal with the highlighted issues in the literature this thesis provides a systematicinquiry about the effect of excessive financial liberalisation on finance-growth relationshipusing comprehensive measures of financial development and financial liberalisation. Italso inquires about the role of the sequencing of financial reform using a quantitativemeasure as a proxy on emergence of financial crises.

Broadly speaking, the financial system is comprised of: various types of financialmarkets, financial intermediaries and financial infrastructure. All major money markets,capital markets, banks, insurance companies and the system available for facilitating thetransfer of money and transactions fall under the umbrella of financial system. In themodern world all transactions between people, industrial and commercial organisationsand countries directly or indirectly take place through one or the other financial subsidiary.

The current financial system has gained so much attention from policy makers andacademics due both to the benefits it entails and the societal fear it precipitates. Many fearthe financial system because of its ability to bring halt to the whole economic activity ifit faces problems1. On the other hand many academics appreciate and support strongerfinancial system because of its role in the overall welfare of the society (see for instance,Ebbinghaus and Manow (2004)). A good financial system can improve the efficient al-location of capital and can smooth the working of commerce, trade and industry throughfacilitating transactions and making liquidity available in difficult times. Having said this,any analysis of economic growth that does not take into account the role of financial sys-tem will be seriously flawed (Ang, 2008). Therefore, the role of financial sector and thepolicies for liberalisation have been an integral part of economic growth literature sincethe pioneering studies by Goldsmith (1969) and McKinnon (1973).

Frequent crises following the liberalisation periods, however, have undermined theconfidence of countries in role of the financial sector and the policies of liberalisation ineconomic growth. Intriguingly, the countries that have been struck with crises during thelast few decades were very popular among economic commentators and were perform-ing very well in terms of macroeconomic fundamentals. For instance, the debt crisis of1980s struck Chile that was McKinnon’s favourite during that time2, the East Asian crisis

1For instance, Stern and Feldman (2004) explain the importance and the costs associated with the prob-lem of too-big-to-fail. However, Mishkin et al. (2006) emphasise that with available regulatory and super-visory infrastructure Gary Stern and Ron Feldman overstate the importance of size of the financial sector.

2McKinnon (1982) even described Chile as “a norm or standard of reference”


battered World Bank’s (1993) miracle economies3 during late 1990s and the most recentcrisis hit USA, the world’s most powerful economy. This inexplicable behaviour of theeconomies has attracted a great deal of attention of the policy makers and academics alike.

Although countries all over the world have introduced rules to improve the perfor-mance of financial system, there are still some loose ends, contradictions and confusionspresent in these policies that cause failure of the whole system. Aghion and Durlauf(2005, p.974) count weak macroeconomic framework and low level of prudential regu-lations in the liberalised financial sector as the main reason for this puzzling behaviourof economies. Christine Lagarde, the managing director of International Monetary Fund,speaking at a London conference on inclusive capitalism observes that: “The behaviour

of the financial sector has not changed fundamentally in a number of dimensions since

the crisis. While some changes in behaviour are taking place, these are not deep or broad

enough. The industry still prizes short-term profit over long-term prudence, today’s bonus

over tomorrow’s relationship” (May 2014).

The great depression, following the collapse of banking sector during 1930s, was thefirst major setback to the role of financial system in allocation of credit and increasing theemployment through improved investment opportunities, which its propagators promisedin economic growth and development. A large number of banks failed in the US during theyear 1930-33 with very large losses to the depositors. Following the great depression, theUSA introduced key regulatory policies such as regulation Q which focused on limitinginterest payments from banks, excluded banks from underwriting, and created federaldeposit insurance (Crafts and Fearon, 2013).

Disturbances in economic and financial system in 1970s and 1980s called an end tothe era of strict regulatory policies in many countries4. A new wave of market orientedreforms aiming to liberate the financial sector from highly repressive policies were intro-duced. A widespread realization about the role of markets increased in many countriesdue to increasing financial globalisation, innovation, competition, and efficient financialmarket theory. As a result many countries all over the world adopted the policies of liber-alisation following the Latin American and East Asian countries.

Over the years countries have introduced, modified, and re-modified many reformsin financial sector on several occasions. The main focus of these reforms was to increase

3Emergence of “miracle economies” and , especially, “four little tigers” changed the views of manyacademics and policy makers about the potential of Asian economies (see (Harrison and Huntington, 2000)).The Asian economies attained 10% growth rate on average for the decade before they plunged into crisis.

4see Crotty (2002) & Crotty (2009) for further explanations on how the practices in the financial sectorchanged over the years.


role of market in allocation of credit with an expectation that the reforms would improvesavings, enhance efficiency and will direct capital to the most productive uses (Caprio Jret al., 1994). These reforms were introduced in three different directions: independenceof central banks; financial deregulation comprising financial liberalisation of key financialmarkets and abolition of interest rate controls; and supervisory reforms (Hall, 2003).

The rate of the financial reforms rose sharply during 1980s and 1990s. After cominginto full motion, during the first half of the 1990s, the policies of financial liberalisationfostered growth (Fernández-Arias and Montiel, 1997) and caused rise in the capital flowsin many Asian and Latin American countries (Griffith-Jones and Tobin, 1998). However,faith in these reforms shattered badly during the second half of the 1990s when the coun-tries that opted for liberalisation plunged into a series of crises. There were many commonfactors that were faced by these crisis hit countries before the emergence of turbulence inthose economies. Among these factors liberalisation of financial markets was the mostcommon after their rigid exchange rate policies, as well as erratic foreign exchange andcurrency markets behaviours (Glick et al., 2001).

This thesis consists of two independent but closely linked studies covering the is-sues related to the impact of financial liberalisation on finance-growth relationship andfinancial crises for a group of countries representing different regions of the world. Inthe first chapter, we examine the impact of financial liberalisation policies on the rela-tionship between financial development and economic growth. We study the relationshipbetween finance and growth for varying levels of financial liberalisation. The purpose ofthe inquiry is to investigate why high levels of financial development failed to generateeconomic growth during periods of financial turbulence? In the second chapter, we inves-tigate why financial reforms fail and how this failure affects systemic vulnerability of thecountries. The inquiry further deepens with the investigation of whether this system couldbe managed better in a different way to avoid crises. For this purpose we look into theimpact of different sequencings of financial reforms on the likelihood of financial crises.In the following discussion we provide details about these two chapters.

1.1 Financial Development and Economic Growth: TheRole of Financial Liberalisation

If the basic theory of financial liberalisaton is anything to go by, then the most liber-alised countries should have been able to achieve higher levels of financial development


and economic growth. Financial crises and consistently prolonging recessions into themost financially liberalised and financially developed economies defy this logic. In thesecond chapter we inquire whether the level of financial liberalisation explains the recentbreakdown of the relationship between financial development and economic growth. Inthis chapter, we postulate that the effectiveness of financial sector development in eco-nomic growth can be maintained through a rational liberalisation process. We definerational liberalisation process as a set of policies that gives chance to demand and supplyforces to determine market outcomes along with realization of regulation as part ratherthan converse to the concept of financial liberalisation (see McCleskey (2010, p.xv)).

A large literature provides empirical evidence on the positive link between financialdevelopment and economic growth. However, during recent economic and financial criseswe have observed that growth rates in many countries declined and their economies wentinto recessions despite having sophisticated and developed financial systems. This phe-nomenon has compelled many academics to revisit evidence related to the link betweenfinancial development and growth. The post 2007-08 crisis literature finds that althoughthere is a positive relationship between financial development and economic growth, thestrength of this relationship recently has decreased significantly (Rousseau and Wachtel,2011).

The dwindling impact of liberalisation might be due to the conflicting nature ofinterest of market forces. The free markets are mainly driven by the motive of profitmaximisation; they care less about the possible impacts of their decisions on sustainabilityof financial system, value of the equity of the individual investors, and volatility in themarkets. The free market structure has stimulated growth and set the stage for those levelsof development that have never been attained in the past. But, the problem with the marketis that it lacks the ability to regulate itself (Akerlof and Shiller, 2010), which results inreckless behaviour on the part of institutions working in the markets and make problemssevere in the absence of the role of a prudential regulatory authority or a watchdog.

Using dynamic GMM analysis we provide systematic evidence related to the im-pact of financial liberalisation on the relationship between financial development and eco-nomic growth for a sample of 88 countries over the period of 1973-2005. To measure theextent of reforms we use a financial reform index covering reforms in the areas of creditcontrols and reserve requirements, interest rate controls, entry barriers, state ownership,policies on securities markets, banking regulations and restrictions on capital market. Ourmeasure of financial development extensively covers different aspects of development fi-nancial sector related to bank private credit to GDP ratio, liquid liability to GDP ratio,


deposit money bank assets to total bank assets ratio, deposit money bank assets to GDPratio, and bank credit to bank deposit ratio.

Our general findings in this chapter suggest that, although the effect of financial de-velopment on economic growth is positive, its impact diminishes as the financial systembecome more liberal. We show that liberalisation of the markets increases financial de-velopment and therefore growth to a certain limit, however, beyond this limit higher levelof liberalisation cause disruptions in the economy.

1.2 What Makes Financial liberalisation Work: The Im-pact of Sequencing on Financial Crises

An explication of the puzzle discussed in the foregoing introduction of the sec-ond chapter concerning the rising economic problems in financially developed and stablecountries lies in the way countries manage their financial liberalisation process5. In thethird chapter we examine the impact of the sequencing of financial reform of the internalfinancial sector on the likelihood of financial crises. More specifically, by studying theinterrelationships between the liberalisation of different sectors of the financial system,we determine whether the sequence in which a country liberalises its financial sector hasa bearing on the chance of crises in those countries. For the sake of this investigation wedefine internal financial system as a combination of domestic market, stock market andcapital account and financial crisis as banking, currency and twin crises.

The issue of sequencing of financial sector reform was highlighted for the first timeafter the policy of abrupt reform resulted in economic problems in Argentina, Chile andUruguay collectively known as Southern Cone countries during late 1970s. The economicproblems in these countries6 due to abrupt financial reforms triggered a debate betweenthe “big-bang” and gradual approach (see for instance, (Bhagwati, 1982, Edwards, 1990,Popov, 2000)). Under the big-bang approach most of the reforms are implemented inone go within quick time period, whereas, under the gradualist approach the reforms areimplemented one by one over the extended period of time (Wei, 1997).

The proponents of big-bang approach pointed out the inability of policy makers tosuggest any particular sequence for liberalisation of the financial system to renounce the

5see for instance, Johnston and Sundararajan (1999)6Edwards (1990), however, notes difference of experience of these countries in terms of success of

reforms. Despite the early problems he termed Chilean experience as success, whereas, Argentinian expe-rience as complete failure and Uruguay’s experience as mixed.


idea of gradualism. Whereas the followers of gradual approach referred to the associatedcost of dislocation of GDP and conflicting nature of financial reform as the biggest prob-lems in the following of big-bang type of approach (Edwards, 1990, Agénor and Montiel,2008).

The debate about the optimal sequencing again picked up momentum during 1990swhen the experience of liberalisation of capital account failed in emerging markets. Manylow income and developing countries liberalised their capital account in the hope thatavailability of capital in the open economies would overcome the problem of capitalshortage, would promote competitiveness among financial institutions, and would makeallocation of financial resources efficient. The developing countries took the decision ofliberalising of capital account without realizing the fact that–with repressed and undevel-oped domestic financial sector– allowing free movement of capital could increase overallrisks in their economies and could loosen their grip on internal financial decision making.

Failures of the economies apparently due to the policy of capital account liberalisa-tion shifted the debate about sequencing to the choice of order between liberalisation ofcapital accounts and liberalisation of the other sectors of the financial system7. The eco-nomic and financial turbulence following the capital account liberalisation gained supportof policy makers and academics for stricter capital controls, especially, in the start of theliberalisation process (Johnston, 1998, Edwards, 2009). Bretton Woods institutions, thebiggest advocates of capital account liberalisation, halted their opposition of capital con-trols in the wake of 2008-09 crisis (see Arora et al. (2013)). According to Forbes et al.(2013) implementation of capital controls, in good times, can help prevent instabilities inbad times. The evidence related to international trade has also shown recently its supportto the capital controls in place to get benefit from international trade (see for instance,Lipinska and De Paoli (2013)).

In this chapter we contribute to the literature discussing the issue of the role of se-quencing of the financial sector reforms. Our particular focus in this chapter is to seek anoptimal sequence for the liberalisation of the financial sector between domestic market,stock market, and capital account. We seek a sequence that increases resilience to finan-cial turmoil and crises. Our findings in this chapter suggest that financial liberalisationprocess should allow domestic sector to be reformed and should be made able to stand inunexpected turn of the events before allowing free flow of foreign capital in the domesticeconomy.

In summary, the overall purpose of this thesis is twofold: analysis of the impact7see (Johnston, 1998, Johnston and Sundararajan, 1999)


of financial liberalisation on finance-growth relationship and secondly, we explore thepossibility for an optimal sequencing of the liberalisation reform that may reduce thelikelihood of financial crises.

Chapter 2

Financial Development and EconomicGrowth: The Role of FinancialLiberalisation

2.1 Introduction

The recent departure1 from the long held belief about the positive role of financialdevelopment in economic growth is due to the fact that financially developed countriesare in the midst of a major economic turmoil even after the six years of emergence offinancial crisis of 2007-08. Although a large literature suggests a strong positive, short-run and long-run relationship between finance and growth2, recent evidence suggest thatthis relationship is losing its strength (Rousseau and Wachtel, 2011). The post-crisisliterature discussing the reasons of the meltdown of the financial system finds root of therecent financial crisis in the deficiencies in financial regulation and architecture (Laevenet al., 2010).

The series of financial crises has brought widespread realisation of the flaws of thefinancial liberalisation policies and has called for a reassessment of the current finan-cial architecture. This chapter advances the existing literature by explicitly investigatingwhether the effect of financial development on economic growth varies with the extentof financial liberalisation. Our investigation provides direct evidence about whether ex-cessive financial liberalisation has been responsible for the breakdown of the relationship

1see for instance Rousseau and Wachtel (2011), Demetriades and Rousseau (2011)2For example, see King and Levine (1993a), King and Levine (1993b)

18

Chapter 2. Financial Development and Growth 19

between finance and growth. We focus mainly on the impact of liberalisation reforms thathave taken place in the form of lessening reserve requirements, controls on interest rates,entry barriers, state ownership, banking regulation and restrictions on capital markets oneconomic growth. This analysis is of great interest in the context of the recent economiccrises and future actions required on the part of countries to deal with these problems.

The complete removal of the policies restricting financial sector activities took placein the late 1980s and early 1990s following the economic and financial disturbances inearly 1970s and 1980s (Crotty, 2009) and due to the emphasis from endogenous growthmodels on the role of financial intermediation in attaining steady state growth (Pagano,1993, Greenwood and Jovanovic, 1989, Ang and McKibbin, 2007, Bencivenga and Smith,1991). Theoretical justification for these actions lies in the benefits that countries gainfrom financial liberalisation. Economic theory asserts that financial liberalisation allowsefficient allocation of financial resources, allows greater risk diversification, increasesreturns on savings and makes terms on borrowing competitive by allowing greater com-petition3.

This work is related to three different strands of literature. First, a number of studiesassociate financial development with economic growth, investment, productivity, innova-tion and entrepreneurship4. Second, there is a large literature studying the relationshipbetween financial liberalisation and economic growth5. Third, a small but still consider-able and growing body of literature studies the link between financial liberalisation andfinance-growth relationship6.

In this work, we focus on the impact of the policies of financial liberalisation onfinance-growth relationship. The proponents of liberalisation believe that repressive poli-cies, such as setting deposit and lending interest rates lower than the equilibrium rate,cause savings and eventually investments to decline (see for instance, Pagano (1993),Roubini and Sala-i Martin (1992)). The followers of this thought believe that financialliberalisation has profound effects on growth because of its positive impact on financialdeepening that increases availability of credit and improves allocation of credit (see forinstance, Arestis and Caner (2004)). This understanding has roots in the basic economictheory that relates the higher growth rate to the saving rate that in turn is increasing func-tion of real interest rate. Under financial repression the interest rates are administratively

3See Klein and Olivei (2008) and Edison et al. (2002)4These include King and Levine (1993b), Carlin and Mayer (2003), Arestis and Demetriades (1997),

Bagehot (1878), Hicks (1969)5For instance, Bekaert et al. (2005), Ayhan Kose et al. (2009)6See for example, (Arestis, 2006)


fixed below the equilibrium level that has adverse affects on saving rates and growth rate(Arestis and Caner, 2004). Consequently, according to this transitional link between fi-nancial liberalisation, real interest rates, savings and growth, the policies financial liberal-isation, especially in developing countries, can improve efficiency of the existing marketsand can pave way for new markets (Gibson and Tsakalotos, 1994).

There is also literature available showing the wariness against the role of financialliberalisation policies in economic growth due to their alleged involvement in economicand financial crises (Weller, 2001, Laeven et al., 2010). Notably, in the wake of recent se-ries of crises in developing and developed countries the opposition to the role of financialliberalisation has risen considerably (see Laeven et al. (2010)).

Contrary to the studies available related to the relationship between finance andgrowth, the literature related to the impact of liberalisation policies on the relationshipbetween finance and growth is scarce. Among several reasons, one reason might be theunavailability of proper data set taking into account actual changes in the financial sectorpolicies (Abiad, Detragiache and Tressel, 2008). The other reason might be the beliefthat the financial liberalisation policies themselves cannot be a cause of the financial meltdown (Demirgüç-Kunt and Detragiache, 1998b).

Empirically, the issue of the exact impact of financial liberalisation on the relation-ship between finance and growth is still unsettled. This study fills the gap by analysingthe impact of financial liberalisation on finance-growth relationship. Our goal is similar toArestis (2006) and Demetriades and Rousseau’s (2011) work7, but it improves the analy-sis in following counts. First, the structure, scope and technique of our empirical modelare wider than Demetriades and Rousseau’s (2011) work. Our model explicitly takes intoaccount the interactions between financial development and our measure of financial lib-eralisation. Second, our financial development measure is more comprehensive as it isan index of financial development (FD) using principal component analysis (PCA) of sixstandard measures of financial development. Our financial development measure takesinto account the variation effect that has taken place since the time of liberalisation thesisof McKinnon and Shaw in 1970s. Our PCA based measure of financial development cap-tures the maximum variance across popular components of financial development acrosstime in our sample period that provides us a better view of financial development process

7Demetriades and Rousseau (2011) examines the impact of no banking supervision taking value 0 whenthere is no banking supervision and +3 when banking supervision is at its best using simple ordinary leastsquares (OLS) on the relationship between finance and growth. Whereas, Arestis (2006) provide a theo-retical account related to the link between finance and growth in the presence of the policies of financialliberalisation. This study differs from both in terms of scope and technique.


and coverage of the sub-dimensions of the financial development. Finally, in particular,we take into account the issues of parameter heterogeneity, outliers, omitted variables andendogeneity in growth equations.

Our findings suggest that financial development has a positive impact on growth.However, we find evidence that this positive effect of financial development on long-rungrowth continues to decline as the financial sector becomes more liberalised. Our resultswith overall level of financial liberalisation and its components suggest that the impactof financial development varies both with the overall level of financial liberalisation andwhen there is change in the components of financial liberalisation policies.

The remaining chapter is organized as follows: Section 2.2 provides the review of lit-erature related to the finance-growth relationship. We explain the data set and the method-ology used in this study in Section 2.3. The main results of the estimation are discussedin Section 2.4. The robustness analysis is provided in Section 2.5. In the end Section 2.6provides conclusion of the study.

2.2 Literature Review

To place our contribution in the literature, in this section, we review empirical andtheoretical literature related to the linkages between financial development and growth,financial liberalisation and economic growth, and linkage between financial liberalisationand finance-growth relationship (also see table A.1 in the Appendix for a summary of therelated literature). Furthermore, through different channels we search for the explanationthat why the liberalisation process may fail to generate consistent positive impacts oneconomic growth. In the following we divide the literature in three different strands asmentioned above:

2.2.1 Financial Development and Economic Growth

The first and the most popular strand of literature studies the relationship betweenfinancial development and economic growth. The literature on the relationship betweenfinance and growth has a consensus that financial development affects growth positivelythrough effective mobilization of saving, capital accumulation, increased investment op-portunities, and the rise in total factor productivity. Historically speaking, the groundbreaking study by Schumpeter (1934) paved the way for the role of the financial sectorin economic performance of countries. Highlighting the role of financial sector, Keynes


(1930) also share the view of Schumpeter by defining banking credit as a “pavement” onwhich production travels.

The biggest support for the Schumpeterian view came from ‘financial liberalisa-tion thesis’ of McKinnon (1973) and Shaw (1973). They argued that repressive policieslike setting interest rates affect the efficiency of the financial sector that affects economicgrowth adversely. Consistent with the Schumpeterian view, King and Levine (1993a),using the Barro’s (1991) cross-sectional regression framework, provided cross-countryempirical evidence on the finance-growth relationship for a sample of 80 countries overthe period 1960-89. They found a positive relationship between financial developmentand higher rates of economic growth, physical capital accumulation, and economic effi-ciency. Along with the association of financial development with the contemporaneousgrowth rate, King and Levine (1993a) further conclude an association of financial devel-opment with future rates of long-run growth, physical capital accumulation and economicefficiency improvements.

King and Levine (1993b) in their endogenous growth model assume the role of thefinancial system in the evaluation of projects and diversification of the risks associatedwith innovation. Their analysis of the experience of five liberalised countries revealsan important role of the financial sector in the acceleration of economic growth throughits impact on innovation. Similarly, Levine and Zervos (1998) in their study related tothe role of the financial sector in economic performance conclude that banks and stockmarkets accelerate economic growth. Empirical studies also attribute the differences inthe stages of growth of different countries to the level of financial development. Levine(1999) argues that since the level of financial development has a causal impact on thegrowth rate, the differences in growth rates among countries is due to the differences inthe level of financial development. Therefore, a large literature finds that the financialsector is an inextricable part of the process of economic growth (Levine, 1997, Green-wood and Jovanovic, 1989). The general conclusion from these studies is that financialdevelopment unambiguously impacts both contemporaneous and future economic growthrates positively.

Following King and Levine numerous other studies also provided theoretical andempirical evidence for the role of financial structure and characteristics of industries ininvestment, industrial growth and economic performance (see for instance, Carlin andMayer (2003)). Rise in the saving rate and the ability of financial system in channelizingit to into the effective investment and increasing the social marginal productivity of invest-ment were the main channels Pagano (1993) emphasised in the effective role of financial


intermediation in the economic growth process. Rajan and Zingales (1996) examined theeffect of finance on growth for a cross country cross industry data set. Using firm leveldata for listed firms and the US as a benchmark country in a sample of 44 countries, theyconclude that financial sector development reduces the differential cost of external financethat directly benefits the firms that are short of funds given investment opportunities.

Along with a large support equally from economists and international financial in-stitutions8 that the finance-growth relationship has received over the years, there are alsostudies available that find either no support or a very weak support for the role of financialdevelopment in the growth structure of countries. For instance, Lucas (1988) believe thatthe growth finance literature “over-stresses" the importance of financial structure in theeconomic development process .

On the middle ground, few favour the role of financial development in economic per-formance but assume the financial development as a necessary but not sufficient condition(Holzmann, 1997). Moreover, a large part of the literature conditions the positive impactof financial development to other factors, for example, legal origin (Porta et al., 1996),institutional quality (Tressel and Detragiache, 2008), functioning of political institutions(Roe and Siegel, 2008), level of income (Odedokun, 1996), etc. Cojocaru et al. (2011),using data over 1990-2008, provide evidence of the positive impact of financial develop-ment for the communist countries of Central and Eastern Europe and the commonwealth.However, they observe that this positive relationship becomes insignificant when the levelof inflation is high.

There is also wariness found in the literature on the “one size fits all" assumption andthe monotonic relationship between finance and growth (Rioja and Valev, 2004, Deiddaand Fattouh, 2002, Arcand et al., 2012). Deidda and Fattouh (2002), using thresholdmodel over the data of King and Levine (1993) find a non-linear relationship betweenfinancial development and economic growth. They find an insignificant relationship fordeveloping countries, whereas, the relationship is significant and positive for developedcountries. Favarra (2003) argues a heterogeneity of the relationship between financialdevelopment and economic growth across countries. He finds no indications that financespurs growth.

Rioja and Valev (2007) test the theoretical findings of Acemoglu et al. (2006), forthe sample of 74 countries, divided into three regions, over the period 1961-1995. Amongtheir defined regions of financial depth, at low levels they find no statistically significant

8The world development report 1989 includes a separate chapter describing the importance of financialsector in economic development


relationship between finance and growth, for medium level there is strong and positiverelationship, and for high level they find weak but positive relationship. Arcand et al.(2012) using the private credit to GDP ratio as a measure of financial depth, concludethat countries with a small and medium sized financial sector benefit from increased fi-nancial depth, whereas, the effect of the size of the financial sector vanishes as the sizeof financial sector reaches to 80-100% of GDP. De la Torre et al. (2011, p.25) explain theproblem of “too much finance" under what they term as “dark side" of financial develop-ment. They argue that unchecked financial development at higher levels causes financialinstability, where, the marginal cost of maintaining financial stability becomes more thanthe marginal benefits of financial development.

The issue of causality is another source of disagreement among economists regard-ing the relationship between finance and growth. Robinson (1952) note that economicgrowth does not follow finance rather finance follows economic growth. Demetriadesand Hussein (1996) also find little support for the leading role of finance in economicgrowth. They find bi-directionality and in some cases reverse causation. Levine et al.(2000) examine the causality issue between financial development and economic growth.Using a GMM panel estimator to control simultaneity bias they conclude that financialdevelopment has a first order effect on long-run growth. However, a re-examination ofthe (Levine et al., 2000) analysis by Favarra (2003), using cross sectional and panel dataanalysis, show no evidence of causality running from financial deepening to GDP growth.Despite some exceptions, in summary, foregoing review of literature suggests a strong andpositive relationship between financial development and short-run and long-run economicgrowth.

2.2.2 The Relationship between Financial Liberalisation and Eco-nomic Growth

The second strand is related to the role of financial liberalisation in stimulating eco-nomic growth. Since the time of McKinnon and Shaw a large debate in the literature iswhether financial liberalisation has significant impact on economic growth. Theoretically,financial liberalisation can influence economic growth directly or indirectly, but there islack of open admittance amongst policy makers and academics about its role in economicgrowth9. Especially, the evidence related to the role of international financial flows as aresult of capital account liberalisation is at best mixed and at worst confusing.

9Further discussion can be found in the works of Kose et al. (2006), Ayhan Kose et al. (2009)


An extensive review of literature reveals contradictory views on the effects of fi-nancial liberalisation on economic growth. While on the one hand many studies favourliberalisation as a necessary and sufficient condition for its positive impact on economicgrowth through increased financial development (Stulz, 1999, Bekaert et al., 2011). Onthe other hand literature also links the international financial flows to the financial insta-bility (Bhagwati, 1998, Stiglitz, 2004).

On the positive side, Stulz (1999) argues that financial liberalisation increases levelof investment by allowing risk diversification and lowering down agency costs. Alongwith output growth liberalisation of financial markets also affects the productivity growth(Ayhan Kose et al., 2009) and the efficiency of the banking sector (Jayaratne and Stra-han, 1996). Ayhan Kose et al. (2009), contrary to a substantial literature on non-robustrelationship between financial openness and total factor productivity (TFP) growth, finda clear and robust association between de-jure measures of financial openness and totalfactor productivity growth for a sample of 67 countries over the period 1966-2005. Theyargue that this positive association is found due to the role of financial openness in real-location of output and inputs across individuals. However, they find less clear evidencefor the effect of de-facto measures of financial openness on TFP growth10. Similarly,Bekaert et al. (2011), dividing growth into capital stock growth and total factor produc-tivity growth find a positive impact of financial openness on both channels of growth.Nevertheless, they find a greater impact of the openness on total factor productivity ascompare to investment.

The point of departure in the literature on financial liberalisation is that threat to sta-bility outweighs the positive impact of these policies on growth. Caprio et al. (2005) showthat the liberalisation of the financial sector is followed by instability in the financial sec-tor that causes economic growth to fall. Jarrow (2014) shows that emergence of problemsin financial markets lead to contraction of economic growth through channels of creditrisk that links the financial market with real economy. Mainly, the idea that financialcrises are associated with the real economy has come from the large literature studyingthe linkages between financial markets and business cycles (see for instance, Bernanke(1983), Brunnermeier and Pedersen (2009), Cooley et al. (2004)).

Crotty (2009) finds radical financial deregulation along with the New Financial Ar-chitecture (NFA)11– that is based on very weak theoretical foundations– responsible for

10Ayhan Kose et al. (2009) explain de-jure measure of capital account openness as no constraints oncapital account transactions and de-facto measured as stock to foreign assets and liabilities to GDP ratio.

11New Financial Architecture refers to the integration of modern day financial markets with the era′slight government regulation.


the 2007-08 financial crisis. He concludes that the biggest reason for the recent crisisis the emergence of financial boom following the progressive deregulation along withsuccessive bailouts of the financial institutions with the label of “too big to fail".

Crotty (2009) further notes that since the NFA defines less regulations for com-mercial banks, even lesser for investment banks and hardly any for the ‘shadow bankingsystem’–hedge and private equity funds and bank-created Special Investment Vehicles(SIVs), these developments have caused a remarkable increase in complex, opaque andilliquid financial assets, which have given rise to system wide leverage.

Similarly, Grant (2009) also finds the roots of the recent financial crisis in the recentFinancial Services Modernization Act of 1999 which is also known as Gramm-Leach-Bliley Act12. Diaz-Alejandro (1985) a long time ago pointed out that the central banksin general have failed to introduce prudential regulations perhaps due to their belief thatthe banks are like other ordinary businesses or perhaps they lacked staff trained enoughto handle the enormously large and complex financial sector. Levine (1999) in his studyabout the role of financial sector development in economic growth emphasises on theeffectiveness of regulatory policies and legal system that could ensure the enforcementof the rights of creditors and shareholders and could also encourage the development offinancial sector. They further emphasise that rather than debating the role of the com-ponents of financial sector, there should be more emphasis on the regulatory and legalenvironment surrounding the financial sector.

Perhaps the mixed evidence on contrasting impacts of financial liberalisation ispartly due to the complexity of the financial system and partly due to the other set ofconditions with which the performance of the financial sector is linked. For instance,there is a belief that the stage of development at which financial liberalisation policies areintroduced matter (Loayza and Ranciere, 2006) along with effective financial institutionalstructure and macroeconomic stability (Caprio and Summers, 1993).

Due to the large contradictory results of the literature related to the impact of finan-cial liberalisation on financial development and economic growth, the literature advisesto be cautious when explaining the role of finance in economic growth (Demetriades andHussein, 1996, De Gregorio and Guidotti, 1995).

12This act called an end to the decades long regulations of the financial sector that were in place sincethe great depression of 1930s


2.2.3 The Impact of Financial Liberalisation on Finance-Growth Re-lationship

Our work is related to the third strand of literature that examines the impact of fi-nancial liberalisation policies on finance-growth relationship. Ex-post crisis literaturerelated to the inquiry about the factors that have led to the emergence of crises counts thelax regulatory environment to be liable for failure of the financial sector (Acharya andRichardson, 2009, Sinha et al., 2012, Crotty, 2009). The positive relationship betweenfinance and growth that has taken a central place in finance-growth literature seems to belosing its strength (Rousseau and Wachtel, 2011).

For the explanations of this changing phenomenon, we need to analyse channelsthrough which financial sector is linked to the real economy. The first explanation forthe lessening impact of finance on growth lies in the excessive financial deepening as aresult of financial liberalisation. Excessive financial deepening in form of credit expansionresults in credit boom and bust cycles, which weaken the banking structure and createinflationary pressures13.

The second explanation for emerging financial crisis in financial sector and fallingimpact of financial development on growth lies in the imperfect behaviour driven by the“animal spirit”14 of the financial institutions (Móczár, 2010). A large body of literaturesuggest that financial institutions are more prone to risk taking behaviour under the lib-eralised financial system (Diaz-Alejandro, 1985, Hellmann et al., 1995, Easterly et al.,2001). This excessive risk taking behaviour has direct consequences on the financialfragility that leads to turmoil in the financial sector and results in the fall of economicgrowth (Bernanke, 1983, Calomiris and Mason, 1994, Keeley, 1990).

The third important factor that affects the impact of financial liberalisation on finance-growth relationship is the uncertainty of response in terms of saving from the householdsand the firms. Bandiera et al. (2000) shows that the ambiguity of the effect of financialliberalisation on private saving arises because of the ambiguities in the relationship be-tween interest rates and savings and also due to the nature of the financial liberalisationprocess that also involves reversals.

13see Rousseau and Wachtel (2011) for further debate14Keynes defines animal spirit as “of a spontaneous urge to action rather than inaction, and not as the

outcome of a weighted average of quantitative benefits multiplied by quantitative probabilities. Enterpriseonly pretends to itself to be mainly actuated by the statements in its own prospectus, however candid andsincere. Only a little more than an expedition to the South Pole, is it based on an exact calculation ofbenefits to come. Thus if the animal spirits are dimmed and the spontaneous optimism falters, leaving us todepend on nothing but a mathematical expectation, enterprise will fade and die”.


Bandiera et al. (2000) show that financial liberalisation leads to a fall in savings,contrary to the belief that the saving rate rises because of financial liberalisation. Theresults obtained by Jappelli and Pagano (2000) suggest that the financial deregulation ofthe 1980s in OECD countries has caused a decline in national savings and growth rate.They suggest that savings declined due to the opening of markets for consumer financesand housing loans that created a disincentive for the households to save. Therefore, asa result, financial liberalisation expanded the prospects for development of the financialsector in terms of more consumer financing projects and house loans. Van Wijnbergen(1983) opines that the amount of savings depend on the functioning and extent of theinformal markets. Moreover, Devereux and Smith (1994) show that it also depends on theattitudes of consumers towards precautionary savings.

The fourth factor that may downplay the role of financial liberalisation on finance-growth relationship is related to the risky investment behaviour as a result of financialliberalisation and existence of unstable financial structure (Grabel, 1995). Financial lib-eralisation can potentially give rise to such investment activities which are speculativein nature that may result in ‘speculation led development’ (Grabel, 1995). The viewspresented by Grabel (1995) are in line with the post-Keyneisan view that emphasises thatfinancial liberalisation induces such investment practices that are risky in nature and makefinancial structures shaky that lead to the lower rates of real sector growth.

The foregoing literature review suggests individually positive relationship betweenfinancial development and growth and financial liberalisation and financial development.However, the transition between these three together depends on the extent of level offinancial liberalisation. This study systematically inquires about how the impact of finan-cial development on economic growth varies, in the aftermath of financial liberalisation,and if this process continues to rise unabated.

2.3 Data and Methodology

This section explains the data sources and proposed methodology used in this study.

2.3.1 Description of the Data

This section provides details about the sample of countries, the measures of financialliberalisation, measures of financial development and other co-variates used in this chap-ter. Appendix shows the description and sources of these variables and provides summary


statistics.

The Sample

We investigate the impact of financial reforms on the relationship between finan-cial development and economic growth using panel of 88 countries (see table A.2 in theAppendix for the list of countries) over the period 1973-2005. Since our study mainlyfocuses on the impacts of financial liberalisation on the level of financial developmentand economic growth, we chose the countries for the panel on the basis of availability ofthe data on financial reforms index and financial development indexes.

The panel of countries has diverse mix of developing, developed and emergingeconomies. Our sample has representative countries from South Asia and East Asia, LatinAmerica and the Caribbean, sub-Saharan Africa, Middle East or North Africa, WesternEuropean countries, Soviet Union countries and Australia, New Zealand and the U.S.The sample comprises 22 developed countries, 12 emerging Asian economies, 17 LatinAmerican economies, 12 countries from sub-Saharan Africa, 18 transitional economies,7 from Middle East and North Africa. The period 1973-2005 covers the major financialreforms brought all over the world following the global derive of the financial reforms af-ter the emphasis of the role of financial reforms from McKinnon and Shaw on economicdevelopment.

We divide data into six data points following standard practice in long-run growthliterature to control for business cycle effect, using five year averages over the period1973-2005. The division of the data into five year averages provide rolling panel suitablefor the dynamic specification of the growth equation and solve missing data problems iswidely used in dynamic growth model (Khadraoui and Smida, 2012).

Measure of Financial Development

A good financial development indicator should include information on different as-pects such as depth, access, efficiency and stability for both financial institutions and itsfinancial markets in the financial system. Moreover, a good indicator for financial de-velopment should also provide information about the ability of the financial system tochannel funds from depositors to investors (Ang and McKibbin, 2007). Therefore, it isreally difficult to have a single measure of financial development that could highlights ofall the aspects of financial system (Huang, 2011).


We use principal component analysis (PCA) on frequently used banking sector indi-cators15 of financial development to get new summary aggregate index that would captureall the possible aspects of financial developments discussed above. The studies coveringthe data since 1970s for financial development mainly use the banking sector indicatorsbecause of the problem of data availability for stock market indexes (see for instance,(Tressel and Detragiache, 2008)). Therefore, we also choose the banking sector indicatorsas measures of financial development. Keeping other aspects of the financial developmentin view, these measures of financial depth as explained by World Bank (2012) are ana-lytically incomplete but empirically ubiquitous for measuring the functioning of financialsystems.

The use of PCA for the aggregate index of financial development is gaining popular-ity in growth finance literature to construct a summary index of financial development andother dimensions of financial systems (Huang, 2011, Ang and McKibbin, 2007). Method-ologically, the PCA produces an orthogonal summary index using N number of differentindicators that are highly correlated. These principle components theoretically can cap-ture the highest amount of variance among different indicators, capturing different di-mensions of the dataset. Our PCA draws information from six financial developmentindicators. We perform the PCA analysis over time on our representative indexes of fi-nancial development16. We perform the PCA analysis over the years for our group ofcountries to capture the impact of transition in the financial system across the countriesover the time. Our summary index measures important dynamics of the changes in thefinancial development due to the coverage of the data set.

Generally, studies use M2/ GDP ratio or private credit/ GDP ratio as measures offinancial development. However, as Ang and McKibbin (2007) argue that these measuresonly show the extent of transaction services provided by the financial system rather thantheir ability to channelize funds. Therefore, for principle components analysis we use,log of Bank private credit to GDP ratio (PVT/Y), Log of Liquid Liabilities to GDP Ratio(LL/Y), log of Deposits money bank assets to deposit money bank assets and central bank(BA/BCBA), log of Deposits money bank assets to GDP ratio (DBA/Y), log of Centralbank assets to GDP ratio (CBA/Y), and log of bank credit to bank deposit ( BC/BD). Thesemeasures of financial development are usually strongly correlated (Ang and McKibbin,

15The data for all indicators are taken from the World Bank’s Financial Structure and Economic Devel-opment Database (2008).

16Table A.3 in the Appendix shows the percent variance for all principal components–six in this case–for all years from 1973–2005. Similarly, Table A.4 presents the results for the first vector corresponding tothe first component of principal component analysis.


2007), which validates the use of a single representative index for the measure of financialdevelopment17. In the following we explain the measures we use for constructing ourfinancial development index:

Private credit/GDP ratio: It is one the most widely used indicators of financialdevelopment in finance-growth literature to measure financial development (Levine et al.,2000, Rioja and Valev, 2004, King and Levine, 1993a). This measure separates out thecredit provided to the non financial private sector from the credit provided to the financialand government sector. This measure shows that the greater the share of private sector inthe financial sector, the higher are the chances for the private sector to obtain funds andinvolve in growth enhancing activities. Therefore, higher values of the private credit toGDP ratio shows higher value of financial development (King and Levine, 1993b).

Liquid Liabilities/GDP ratio: The second component of our index is the log ofliquid liabilities to GDP ratio which measures currency held outside of the banking sys-tem plus demand and interest-bearing liabilities of banks and non bank financial inter-mediaries. This measure is commonly used in the literature (Goldsmith, 1969, King andLevine, 1993b) as the size of the financial sector in relation to GDP. The main reasonto include this measure is the association between the size of the financial sector andavailability of the financial services.

Bank assets to deposit money bank assets and central bank assets: The thirdcomponent of our aggregate summary index is the log of deposits money bank assets todeposit money bank assets and central bank assets. Since commercial banks have directlink to the savers and the investors, they play more important role in facilitating investmentopportunities. This measure explains the share of the assets of commercial banks to thecombined assets of commercial and central banks. Therefore, the higher value of theassets of commercial banks means the higher level of financial development.

Deposit money banks assets/ GDP ratio: The fourth component of our aggregateindex of log of deposit money banks assets to GDP ratio provides broad measures ofmoney supply excluding the currency in circulation.

Central bank assets to GDP ratio : Another traditional measure along with depositmoney bank assets to GDP ration with is central bank assets to GDP ratio of size offinancial sector, which together provide measure of the financial services provided bythe deposit money banks and central banks in the economy (Demirguc-Kunt and Levine,

17Table A.5 in the Appendix shows correlation matrix for the components of financial development.With only few exceptions, all variables are significantly correlated with each other with sufficiently largemagnitudes.


2004).

Bank credit/ bank deposit ratio: The last measure of financial development is thebank credit to bank deposit that is used here as a measure for the development of bankingsector. Supposedly, the higher value of this ratio indicates higher level of banking sectordevelopment.

Table 2.1: Principal Component Analysis for Financial Development Index forthe Year 1973

PCA 1 PCA 2 PCA 3 PCA 4 PCA 5 PCA 6

Eigenvalues 3.23756 1.62338 .756813 .291184 .0728865 .0181775

% of variance 0.5396 0.2706 0.1261 0.0485 0.0121 0.0030

Cumulative% 0.5396 0.8102 0.9363 0.9848 0.9970 1.0000

Variable Vector 1 Vector 2 Vector 3 Vector 4 Vector 5 Vector 6

LL/Y 0.4688 0.3425 -0.2591 -0.1435 0.7490 0.1193

BA/BCBA 0.3937 -0.4045 -0.2712 0.7796 -0.0054 -0.0025

PVT/Y 0.5385 0.1035 0.1670 -0.1642 -0.2359 -0.7673

CBA/Y -0.0940 0.6521 0.5004 0.5602 0.0412 -0.0017

DBA/Y 0.5233 0.2202 0.0204 -0.1448 -0.5445 0.5998

BC/BD 0.2292 -0.4843 0.7620 -0.0993 0.2919 0.19291 LL/Y=Log of Liquid Liabilities to GDP Ratio2 BA/BCBA =Log of Deposits money bank assets to deposit money bank assets and centralbank3 PVT/Y=Log of Bank private credit to GDP (%)4 CBA/Y=Log of Central bank assets to GDP (%)5 DBA/Y=Log of Deposits money bank assets to GDP (%)6 BC/BD=Log of bank credit to bank deposit

For the purpose of illustration we present results for a specific year, 1973, obtainedfrom PCA. Table 2.1 displays the results obtained from the PCA analysis for the year1973. The eigenvalues indicated in the table explains about 53.96% of the standardisedvariance, the second principal component explains 27.06%, the third explains 12.61%, thefourth 4.8%, the fifth 1.21% and the last principal component accounts for 0.3%. Fromthe given amount of variation explained by principal components, we observe that the firstcomponent that explains the most of the variation is the best measure of financial devel-opment. By using the weights given by the first eigenvector the first principal componentis computed a linear combination of the six measures of financial development. The in-dividual contributions of LL/Y, BA/BCBA, PVT/Y, CBA/Y, DBA/Y, and BC/BD to the


standardised variance of the first principal component comes out to be around18 22.76%,19.11%, 26.14%, -4.5%, 25.41%, and 11.13% respectively. We use these individual con-tributions as the basis of weighting to develop our financial development index.

Measure for Financial Liberalisation

To measure overall changes in the policies of liberalisation in our sample countries,we use an index developed by Abiad, Detragiache and Tressel (2008) that measures over-all change of level of liberalisation in a sample of 91 countries. This is the most updateddata set available that takes into account the overall level of financial liberalisation indifferent countries. Compared to earlier databases by Edison and Warnock (2003) andKaminsky and Schmukler (2008) this data set covers wider number of countries and pro-vides graded index for more dimensions of financial reforms. Recently many studies haveused this data set to evaluate the impacts of financial reforms on the financial development(Demetriades and Rousseau, 2011, Tressel and Detragiache, 2008).

Abiad, Detragiache and Tressel (2008) provide data on 91 economies over the period1973-2005, capturing the multifaceted dimensions of the financial liberalisation. Six outof seven components of the financial reform index are coded in such a way that highervalue of these dimensions on the scale of 0-3 shows the higher level of liberalisation.Abiad, Detragiache and Tressel (2008) provide an aggregated index combining all impor-tant major dimensions of financial reforms i.e., credit controls and reserve requirements,interest rate controls, entry barriers, state ownership, policies on securities markets, bank-ing regulations and restrictions on capital account.

The aggregated index along with sub indexes is constructed after assigning raw scoreto each dimension and normalizing it to 0-3 scale. On this scale fully liberalised takes thevalue of 3, partially liberalised takes the value of 2, partially repressed take the value of1 and fully repressed takes the value of 0. The dimensions of the dataset comprise manyimportant sub-dimensions that covers further important aspects of the financial reformsbrought in those countries. Abiad, Detragiache and Tressel’s (2008) definitions of thesesub-dimensions and codes are reported below19.

(i) Credit controls and reserve requirements. This dimension is the sum of the scoresgiven for three questions. The first question asks whether the reserve requirementsare restrictive. This assumes code 0, 1, and 2 if the reserve requirements are more

18Figures are rounded off to the second decimal place.19For detailed explanation of the index formation please refer Abiad, Detragiache and Tressel (2008).


than 20%, between 10-20 %, and less than 10% respectively. The second question isrelated to the minimum amounts of credits that must be channeled to certain sectors.This question is coded 0 and 1if the credit allocations are determined by the centralbank and if the mandatory credit allocations do not exist respectively. The thirdquestion is related to whether the credits are being supplied to certain at subsidisedrates. This question uses code 0/1 if the banks have to supply credit to certainsector and 1 if they do not. The sum of the scores against these questions determinewhether the credit controls and reserve requirements are fully liberalised, largelyliberalised, partially repressed or fully repressed.

(ii) Interest rate liberalisation. This is coded as 0 for binding ceilings, 1 if fluctuatingwithin a band and 2 if freely floating. For this dimension fully liberalised showsscore of 4 showing freely floating for both deposit and lending rates. Largely liber-alised shows the score of 3 showing either deposit rates or lending rates are free butother is subject to band. Partially repressed takes score of 2 showing one of the ratesis free and other is subject to ceiling/floor; or both deposit rates and lending ratesare subject to band or partially liberalised. Fully repressed shows both deposit andlending rates set by the government.

(iii) The entry barrier. This dimension covers the policies of the government for al-lowing foreign banks and competition in the domestic banking market. The entrybarriers codes are drawn on the basis of the following questions: whether foreignbanks are allowed or not? Does the government allow the entry of new domesticbanks? Are there restrictions on branching in place? Does the government allowbanks to engage in a wide range of activities? The scores of these questions areadded to create the normalised codes on the scale between 0-3, which representscale of liberalisation.

(iv) Restrictions on capital account. The codes for restrictions on capital account arebase on the following questions with scores of 0 or 1: is the exchange rate systemunified? Does a country set restrictions on capital inflows? Does a country set re-strictions on capital outflows? The addition of the 0/1 codes for these questionsdetermine what stage the capital account liberalisation is on the scale of liberalisa-tion.

(v) Privatization of the financial sector. This dimension is coded as fully liberalisedif no state bank exists or state owned banks own less than 10% portion of the banks;


largely liberalised code is given to those countries in which most of the banks areprivately owned or the percentage of public banks is between 10 and 25 %. Partiallyrepressed show many banks are privately owned but major banks are state-ownedor the percentage of public bank assets is 25-50%; fully repressed show that majorbanks are owned by state and/ or the percentage of public bank asset is from 50% to100%.

(vi) Securities market policy. The dimension of policies on securities markets is thesum of codes resulting from the questions as has a country taken measures to developsecurities markets? Is a country’s equity market open to foreign investors?

(vii) Banking sector supervision The codes for dimension of banking sector supervi-sion are generated using the following questions: Has a country adopted a capitaladequacy ratio based on the Basel standard? Is the banking supervisory agency in-dependent from executives’ influence? Does a banking supervisory agency conducteffective supervision through on-site and off-site examinations? Does a country’sbanking supervisory agency conducts effective and sophisticated examinations?

Only this component out of seven components is coded in reverse order, whichshows that the higher the level of the banking supervision the greater is the ex-tent of financial reforms. This weakness of the index can be addressed by reversingthe banking supervision component and re-calculating the index, which materiallywill not change the empirical results as this is the only one component out of sevencomponents. Still to confirm and for robustness exercise we reverse the bankingsupervision component from 0-3 scale to 3-0 scale, where, 0 shows the highest and3 shows the lowest level of banking supervision. Along with fully re-calculatedfinancial liberalisation index by reversing the banking supervision component forrobustness of our results, we also estimate our main regression results on the dis-aggregated components of financial reform. This exercise will help us understandthe impact of financial development on growth when individual components of theoverall liberalisation index change.

These seven aspects of financial liberalisation are aggregated to calculate ‘a singleliberalisation index’ that takes values from 0 to 21 because each dimension values between0 and 3. The assigned values for the different dimensions of financial reforms are furtherused to construct an aggregated normalised index ranging between ‘0’ and ‘1 with equalweights to all dimensions. Higher value on index means the country is more liberal.


Table A.6 in the Appendix shows that all the sub-indices of financial reform index arepositively and significantly correlated. The positive significant correlations show that thereforms in one sector are positively and significantly associated with the reforms in theother sectors that bring overall change in the whole financial sector. The countries wherethe emphasis is say for example on the directed credit, they may have also the emphasison the other factors as credit ceilings, interest rate controls, etc.

Control Variables

We use standard control variables as used by majority of the studies employinggrowth equations (see for instance, Barro (1996), Levine and Zervos (1998), Rioja andValev (2004)). In the following we provide list of our control variables. Table A.7 in theAppendix provides the definitions and sources of all variables used in this study.

• Initial level of GDP per capita: Following Barro (1996) we use initial level ofGDP per capita as a measure of conditional convergence. We expect negative signfor our initial level of GDP. The negative sign for the coefficient of initial level ofGDP/capita is in line with neoclassical model showing differential growth rates forpoorer and richer countries. According to the conditional convergence hypothesisGDP/capita for poorer countries grow at faster rate than their richer counterparts(see Barro (1996), Jones and Manuelli (2005)).

• Openness to trade: Volume of imports and exports to GDP as a measure of tradeopenness of the countries. The openness to trade indicator is widely used for effectof trade liberalisation on growth (see for instance, Wacziarg and Welch (2008)).Theoretically, the policies of trade openness get so much support from classical,neoclassical and endogenous growth models. However, empirically the evidenceshows different signs and significance level for the impact of trade liberalisationon economic growth. For instance, respectively Greenaway et al. (1997), Michaelyet al. (1991), and Barro (1996) show negative, positive and no impact of trade lib-eralisation on economic growth.

• Investment: We use investment growth rate as a measure of physical capital ac-cumulation in economy. Hill (1964) a long ago explained physical capital to bean important determinant of economic growth. Following Kormendi and Meguire(1985) we expect positive significant effect of investment in physical resources onlong run economic growth .


• Size of government: Government expenditures as a ratio of GDP as a measure ofsize of government. The government size is expected to have negative effect ongrowth as the ‘supply side’ theories suggest that the taxes levied to support thegovernment expenditures decrease efficiency of allocation or resources (Kormendiand Meguire, 1985).

• Inflation: Inflation measured as log of one plus consumer price index. The empir-ical evidence of the impact of inflation on growth is mixed. The Tobin-Mundellhypothesis explain positive relationship between inflation and economic growth.This effect states that when inflation rises, it increases the capital investment thatin turn affect economic growth positively. Fischer (1979) analysing the impactsof anticipated changes in money supply concludes positive relationship betweengrowth rate in money supply and economic growth. Theoretical justification alongwith some empirical evidence for the impact of low inflation on economic growthis tenuous at best (Judson and Orphanides, 1999).

Davis and Kanago (1996) find inflation reduces real GNP growth rate but they fur-ther emphasise that this effect is temporary. However, they also refer studies thatfind the relationship between inflation and growth as negative and significant (seeLevi and Makin (1980), Mullineaux (1980))

• Human capital: Investment in human capital measured as log of average yearsof secondary schooling. We expect the investment in human capital as a positivecontributor to growth. However, literature is mainly inconclusive about the exactimpact of human capital on economic growth. Benhabib and Spiegel (1994) findno significant effect of human capital on GDP/capita, whereas, Becker et al. (1994)find positive impact of human capital on growth.

2.3.2 Summary Statistics and Correlations

Table A.8 in the Appendix provides summary statistics for six of our financial de-velopment indicators, summary measures of financial reform index and control variables.Our data comprises on the sample of 88 countries over the period 1973-2005. The rangeof the data for different variables in table A.8 shows a wide variation among the valuesof the variables. For example, GDP per capita growth ranges from -45% to around 25%.Financial reform index is normalised index that ranges between zero and one. We alsoobserve a wide variation in the measures of financial development. The most widely used


measure of financial development of private credit/ GDP ratio varies between 3.94% forUganda to 146.81% for Hong Kong. In the same way the bank credit to bank deposit ratiovaries between 14.16% for Albania to 424.46% for Vietnam; the liquid liabilities to GDPratio from 5.78% for Zimbabwe to 199.21% for Hong Kong; the central bank assets toGDP ratio from 0.03 % for Lithuania to 66.2% for Nicaragua; deposit money assets toGDP ratio from 4.75% for Zimbabwe to 193.67% for Japan; and deposits money bankassets to deposit money bank assets and central bank assets from 36.01% for Ghana to99.9% for Lithuania. Table A.9 shows correlation results, that shows positive and signifi-cant correlation between between financial development and financial reform index.

Table 2.2 shows frequency histogram of the liberalisation index for the period 1973-2005. We divide countries in five different groups based on the average level of liberalisa-tion in the countries over the period of 1973-2005. Not surprisingly, we observe that thedeveloped countries tend to have more liberalised financial structure than the developingcountries20 . The countries that stand out as the most liberal are United States, Nether-lands, and Switzerland, while Ethiopia and China are the least financially liberal economyfor our sample. The surprising fact is that the countries that have been hit during 2007-08financial crisis and some of them are still facing looming threats of economic recessionare amongst the most financial liberalised economy. For instance following the recessionin the U.S., eurozone crisis since early 2009 has affected the financial structures of manycentral and eastern European economies.

An important relationship that has a central place in finance-growth relationship isthe positive relationship between financial development and growth. Table 2.3 organisesthe countries in the data set in a matrix, categorizing them by quintile according to theaverage level of level of GDP per capita growth rate in those countries and the averagelevel of financial development defined as private credit/ GDP ratio (PVT/Y). Most of oursample countries lie on the diagonal or near the diagonal that shows a well establishedstrong positive association between financial development and economic growth. Thestrong positive correlation between growth and financial development shows that finan-cial development in the countries in the most of the cases has strong positive impact ongrowth. The surprising fact, on the other hand, is that despite the high level of financialdevelopment–mainly as a result of high level of financial liberalisation– many developedcountries have recently experienced recession, which raises serious questions about theestablished link between financial development and growth in the presence of liberalisedfinancial systems.

20for further reference see Byrne and Davis (2003, pg.31)


Table 2.2: Financial Liberalisation Index

5th Quintile 4th Quintile 3rd Quintile2nd Quintile 1st Quintile0.71-1.00 0.55-0.71 0.44-0.55 0.30-0.44 0-0.30Belgium Australia Albania Colombia AlgeriaCanada Bulgaria Argentina Dominican RepublicBangladeshDenmark Chile Austria El Salvador BrazilEstonia Czech Republic Azerbaijan Guatemala CameroonGeorgia Finland Belarus Indonesia ChinaGermany France Bolivia Jamaica Costa RicaHong Kong Italy Greece Kenya EcuadorHungary Japan Israel Madagascar EgyptIreland Jordan Kazakhstan Morocco EthiopiaLatvia Kyrgyz RepublicKorea Nicaragua GhanaLithuania Malaysia Mexico Peru IndiaNetherlands New Zealand Nigeria Senegal MozambiqueSingapore Norway Paraguay Sri Lanka NepalSpain Poland Philippines Thailand PakistanSweden Romania Portugal Tunisia TanzaniaSwitzerland Russia Ukraine Turkey UzbekistanUnited KingdomSouth Africa Uruguay Uganda VietnamUnited States Venezuela ZimbabweSource: Abiad, Detragiache and Tressel (2008)1. The financial liberalisation index is an average of seven indexes: credit controls and reserve re-quirements, interest rate liberalisation, entry barriers, restrictions on capital account, privatizationof the financial sector, and banking sector supervision.2. The countries are divided in 5 quintile on the basis of the level of liberalisation. The least liberaleconomies on the scale between 0 and 1 lie in the first quintile and the most liberal economies liein the 5th quintile

Table 2.4 shows another matrix that arranges countries in groups in quintile on thebasis of average level of financial liberalisation21 and average level of level of financialdevelopment measured as the average of private credit/ GDP ratio in our sample countries.The table shows the most of the financially developed countries have the most liberalisedfinancial structures. Here also the majority of the countries lie near the diagonal or on thediagonal that also depicts a fairly strong positive correlation between the countries’ levelof financial liberalisation and the level of financial development.

This suggest that there is indirect relationship between financial liberalisation andgrowth through positive impact on financial development that is directly associated withGDP/capita growth rate. However, this smooth transition between financial liberalisation,

21The average financial liberalisation index comprises of: credit controls and reserve requirements, in-terest rate liberalisation, entry barriers, restrictions on capital account, privatization of the financial sector,and banking sector supervision


Table 2.3: GDP/Capita Growth and Financial Development

GDP/ Capita Growth Rate(Decreasing→)

FD(Decreasing ↓)

5th Quint 4th Quint 3rd Quint 2nd Quint 1st Quint

5th QuintChina Austria Canada Switzerland -Ireland United Kingdom France

Malaysia Italy GermnaySingapore Jordan NetherlandsThailand Japan

PortugalSpain

4th Quint

Chile Finland Australia Czech Republic South AfricaKorea United States Belgium New Zealand

Norway DenmarkTunisia GreeceVietnam Israel

Sweden

3rd Quint

Egypt Bulgaria Brazil Algeria BoliviaEstonia Bangladesh Columbia El Salvador Republic Jamaica

Indonesia Dominican Republic Phillipines NicaraguaPoland Hungry Uruguay Senegal

Morocco

2nd Quint

Latvia Pakistan Costa Rica Argentina KeyneaIndia Turkey Cameroon Madagascar

Sri Lanka Ecuador VenezuelaGuatemalaLithuaniaMexicoNepal

Paraguay

1st Quint- Belarus Albania Nigeria Azerbaijan

Uganda Kazakhstan Romania EthiopiaMozambique Tanzania Georgia

Kyrgyz RepublicPeru

RussiaUkrain

Zimbabwe1 Source: :Abiad, Detragiache and Tressel (2008) & World Bank′s Financial Structure and Economic Devel-opment Database (2008)2 The financial development indicator is the average level of Prviate Credit/GDP ratio in sample countries.GDP/capita shows average average of GDP/capita growth rate.

financial development and economic growth in the presence of many episodes of reces-sions and economic turmoils in liberalised and financially developed economies seem tobe not straight forward.

To inquire further, we investigate how financial liberalisation and economic growthare interrelated directly. In the ex-post crisis literature the investigation about the as-sociation between financial liberalisation and economic growth rate, has received specialattention both from policy makers and academics. For this purpose in table 2.5 we arrange


Table 2.4: Financial Liberalisation and Financial Development

Financial reforms(Decreasing→)

FD(Decreasing ↓)


5th Quint

United Kingdom France Austria Thailand ChinaCanada Italy Portugal

Germany JapanIreland Jordan

Netherlands MalysiaSingapore

SpainSwitzerland

4th Quint

Belgium Australia Greece Tunisia VietnamDenmark Chile IsraelSweden Czech Republic Korea

United States New ZealandNorway

South Africa

3rd Quint

Estonia Bulgaria Bolivia Columbia AlgeriaaHungary Poland PhilippinesDominican Republic Bangladesh

Uruguay El Salvador BrazilIndonesia EgyptJamaica

NicaraguaSenegal

2nd Quint

Latvia - Argentina Guatemala CameroonLithuania Mexico Kenya Costa Rica

Paraguay Madagascar EcuadorVenezuela Sri Lanka India

Turkey NepalPakistan

1st QuintGeorgia Romania Albania Peru Ethiopia

Russia Azerbaijan Uganda GhanaNigeria Zimbabwe MazambiqueUkraine Tanzania

1 Source:Abiad, Detragiache and Tressel (2008) & World Bank′s Financial Structure and Economic De-velopment Database (2008).2 The financial development indicator is the average of Private Credit/GDP ratio. Financial liberalisationindex is the average of the indexes: credit controls and reserve requirements, interest rate liberalisation,entry barriers, restrictions on capital account, privatization of the financial sector, and banking sectorsupervision.

countries from our dataset in a matrix in quintiles on the basis of their average GDP/capitagrowth rate and average level of liberalisation. We observe that a large number of coun-tries are off the diagonal showing weak relationship between GDP per capita growth rateand level of financial reforms. The extreme cases include countries like China, India twoemerging economies showing the highest rate of growth rate despite the least level of fi-nancial liberalisation over our sample period. While on the one hand these countries havebeen able to achieve high level of financial development despite low level of financial


Table 2.5: GDP/Capita Growth and Financial Liberalisation Index

GDP/Capita Growth Rate(Decreasing→)

FL(Decreasing ↓)


5th QuintEstonia United Kingdom Belgium Lithuania Gerogia

Hong Kong Hungry Canada SwitzerlandIreland Spain DenmarkLatvia United States Germany

Singapore NetherlandsSweden

4th Quint

Chile Bulgaria Australia Czech RepublicKyrgyz RepublicMalaysia Italy France New Zealand RussiaNorway Finland Romania South AfricaPoland Japan

Jordan

3rd Quint

Korea Austria Albania Argentina AzerbaijanBelarus Greece Mexico BoliviaPortugal Israel Nigeria Ukraine

Kazakhstan Paraguay VenezuelaPhillipinesUruguay

2nd Quint

Indonesia Dominican Republic Columbia El Salvador JamaicaSri Lanka Morocco Guatemala KenyaThailand Turkey MadagascarTunisia Uganda Nicaragua

PeruSenegal

Zimbabwe

1st QuintChina Bangladesh Brazil Algeria EthiopiaEgypt Pakistan Costa Rica Cameroon GhanaIndia Mozambique Ecuador Uzbekistan

Vietnam NepalTanzania

Source: & World Bank′s Financial Structure and Economic Development Database (2008)

liberalisation, and on the other hand have also been able to achieve higher growth rates.

Allen et al. (2005) recognises China’s performance as an interesting case studyshowing results contrary to the findings in law-finance-growth literature. They argue thatdespite undeveloped financial sector, China has been able to grow at the fastest pace be-cause of the available financing channels other than formal sources. Guariglia and Poncet(2008) explain despite negative relationship between China-specific measures of state in-terventionism and growth and its sources, it has been able to achieve higher growth ratesbecause of high level of FDI. India, another emerging economy is also among the group ofthe countries with least financial liberalisation and highest GDP per capita. India despiteintroduction of several reforms in their financial sector structurally remained repressivedue to its policies of subsidised and directed credit to particular sectors. Xu (2000) also


find negative relationship between financial development and economic growth for 14countries out of 41 countries sample, including India. On the other extreme there arecountries like Georgia, Switzerland showing the least amount of GDP per capita growthdespite the highest level of financial reforms. The other major countries showing unevenrelationship between level of financial reforms and GDP per capita growth include UnitedKingdom, United States, Canada, and New Zealand.

2.3.3 Methodology

A large number of studies use cross sectional analysis for estimating the growthregression (Baumol, 1986, Barro, 1991, Barro et al., 1992). However, the restricted as-sumption of uncorrelated individual effects in standard cross sectional specification doesnot hold in dynamic specification of growth models (Caselli et al., 1996). Moreover, thecross sectional regression analysis according to Rodrik (2005) assume the reforms as ex-ogenous, which can not be justified econometrically or theoretically. Another issue withthe cross sectional regressions is its inability to control country specific effects, as allcountries in the group are at different stages of the policy variables.

To control the country specific effect many studies use the fixed effect regressionmodels. However, both cross sectional and fixed effect models do not take into accountthe issue of endogeneity present in the growth equations (Caselli et al., 1996). To deal withthe problem of country specific effects and the problem of endogeneity we use generalisedmethod of moments introduced by Holtz-Eakin et al. (1988); Arellano and Bond (1991)and Arellano and Bover (1995) and popularised by Caselli et al. (1996), Levine et al.(2000) and Beck and Levine (2004) in the growth finance literature. We use one-stepand two-step versions of system GMM. The system GMM deals with the endogenouscomponents by using lag of the variables as instruments. The one-step estimators assumeerror term εit to be i, i,d, whereas, the two-step estimators allows the error term εit to beheteroscedastic. The two-step procedure in the first step assume independent error term,and in the second step relaxes the assumption by establishing variance covariance matrixusing error terms from the first step.

Rousseau and Wachtel (2000) use the same technique to study the relationship be-tween stock market, banks and economic growth. Roodman (2006) also emphasises theuse of these techniques for the models with a dynamic dependent variable, which dependson its past realizations, have independent variables that are not strictly exogenous, havefixed effects, and heteroscedasticity and autocorrelation within individuals but not across


them.

Equation (2.1) shows regression equation for growth for our cross country model22.

yi,t− yi,t−1 = αyi,t−1 +β′Xi,t +ηi + εi,t (2.1)

Where, yi,t is real per capita GDP; Xi,t shows vector of explanatory variables except ini-tial level of per capita GDP and includes our indicators of financial reforms; ηi,t showscountry specific fixed effects and; εi shows idiosyncratic error, and the subscripts i and t

show country and time period, respectively. To measure the time specific effects, we alsoinclude time dummies.

According to Arellano and Bond (1991) differencing the equation 2.1 yield,

(yi,t− yi,t−1)− (yi,t−1− yi,t−2) = α(yi,t−1− yi,t−2)+β′(Xi,t−Xi,t−1)+(εi,t− εi,t−1)(2.2)

The model explained in equation 2.2 removes the country specific effects. Arellanoand Bond (1991) show that this specification introduces another bias due to correlationbetween new error term and lagged dependent variable. With the following two assump-tions explained by Arellano and Bond (1991) we can over come this new bias.

E[yi,t−s(εi,t− εi,t−1)] = 0 f or s≥ 2; t = 3, .....,T (2.3)

E[Xi,t−s(εi,t− εi,t−1)] = 0 f or s≥ 2; t = 3, .....,T (2.4)

Bond et al. (2001) argue that the first difference GMM estimators can be misleadingin presence of persistent variables, as lagged levels of the series provide weak instrumentsfor subsequent first differences. Therefore, we use system GMM as suggested by Arel-lano and Bover (1995) and Blundell and Bond (1998) to deals with this problem moreeffectively. According to Bond et al. (2001) the system estimator is useful even for seriesthat are persistent. We have the following stationary conditions.

E[yi,t+pηi] = E[yi,t+qηi] and E[Xi,t+pηi] = E[Xi,t+qηi] for all p and q (2.5)

22The specification is defined by Beck and Levine (2004)


The further moment conditions are as under,

E[(yi,t−s− yi,t−s−1)(ηi + εi,t)] = 0 for s = 1 (2.6)

E[(Xi,t−s−Xi,t−s−1)(ηi + εi,t)] = 0 for s = 1 (2.7)

We use moment conditions given in equations 2.3, 2.4, 2.6 and 2.7 to get GMMsystem estimators. To check the consistency of the estimators derived from GMM wevalidate the assumption that the error terms do not show serial correlation and furthervalidate the instruments. For this purpose we use specification tests suggested by Arellanoand Bond (1991), Arellano and Bover (1995), and Blundell and Bond (1998). The first isthe Sargan test of over-identifying restriction and the second test is to check if the errorterm εi,t is not serially correlated.

Equation (2.8) shows the exact specification for our model.

PCGi,t = β1(LYo)i,t +β2FDi,t +β3FLi,t +β4(FD∗FL)i,t +β5Investmenti,t (2.8)

+β6Trade Opennessi,t +β7Government Sizei,t +β8Inflationi,t

+β9Educationi,t + εi,t +ηi

where, PCG= Per capita growth; (LYo)i,t= initial value of per-capita GDP; FL= Indicatorof financial liberalisation; (FD ∗FL)i,t=Interaction dummy showing financial develop-ment at different level of financial liberalisation. Investment= growth rate of investment;trade openness= import plus export as ratio of GDP; government size= government expen-diture as a ratio of GDP; inflation= log of one plus consumer price index; education= logof secondary years of education; η= country specific fixed effects and; ε = idiosyncraticerror.

We hypothesize β2 > 0 and β4 < 0, which means that individual effects of financialdevelopment is positive but the overall impact becomes less positive or negative withhigher level of financial reforms.

Hadi Method of Multiple Variable Outliers

Since our sample consists of a big heterogeneous group of countries over a longperiod of time between 1973-2005, there is a very strong likelihood of the presence ofoutliers in our sample. Hadi et al. (2009) argues the presence of outliers can affect the


results of the regression analysis that assumes homogeneity of the data and no outliers.Following Easterly et al. (2003), Roodman (2007) we use Hadi (1992) method for iden-tifying multiple outliers to the main variables in our analysis, using 0.01 as the cut ofsignificance level. Variety of analyses in economics literature use this method for detec-tion and exclusion of outliers (see for instance, Rajan and Subramanian (2005))

Hadi (1992) uses a “distance based method” for detection of outliers in a multivariatedata. After measuring the “local distances” it reduces the sample by removing distanceddata iteratively. Hadi (1992) assume X as an nxp data matrix, where n is number ofobservations and p is the number of variables. By ordering and choosing a measure ofoutlyingness, the authors divides data into two subsets: one is ‘basic’ containing p+

1 ‘good’ observations and a ‘non-basic’ containing n− p− 1 observations. Then theauthors computes the relative distance of each point the data set to the centre of the basicsubset. Furthermore, the authors arranges the n observations in ascending order, thendivide the data again into two basic and non-basic subsets containing p−2 and n− p−2observations. They repeat the procedure until they meet an appropriately chosen criterion.The final non-basic set they declare as the outliers.

2.4 Results

This section provides estimation results of the models used in this study. Employingpooled OLS, one-step system GMM and two-step system GMM firstly, we present resultsfrom our baseline regression model with financial development as a main explanatoryvariable. Secondly, we expand the simple baseline regression model by incorporatingthe financial liberalisation index. In the last, we explain full model incorporating ourmeasure of financial development, measures of financial liberalisation, the interactionsbetween financial development and overall index along with components of the financialliberalisation index.

2.4.1 Estimation Results of Baseline Model

As noted earlier, financial development contributes positively to economic growththrough better resource allocations, improving access of businesses to cheap availablefunds and facilitating economic transactions with modern financial infrastructure facil-itating transactions. However, under liberalised financial system, according to our hy-pothesis, the economic growth process distorts due to the dilemma in front of financial


institutions and business organizations between what is socially desirable and what isprofitable to produce (Akerlof and Shiller, 2010).

Before our analysis on the relationship between financial liberalisation and finance-growth relationship, we start our analysis with an examination of the impact of financialdevelopment on long-run economic growth and by incorporating financial liberalisationindex. We run pooled ordinary least squares regression analysis with time dummies, one-step system GMM, and two-step system GMM with GDP/capita growth rate as dependentvariable. The two-step system GMM uses (Windmeijer, 2005) small sample robust stan-dard errors.

In table 2.6 we provide results of the baseline regression model. The first twocolumns present results using OLS, the third and fourth column display results usingone-step system GMM and fifth and sixth columns display results for two-step systemGMM. Moreover, the columns 1, 3, and 5 provide results for the independent effect offinancial development for the bench mark regression model. The columns 2, 4, and 6repeat the analysis for the specification incorporating financial liberalisation.

The results of our benchmark regressions both with and without the financial liberal-isation index confirm the association between long-run economic growth and financial de-velopment consistently across all estimation methods showing coefficient of our measureof financial development as positive and significant across all specifications. The first,third, and fifth columns produce standard results of finance-growth literature showingindividually financial development to be significantly positively related with the growthrate (column (1) FD coeff: 0.552, S.E.: 0.095, column (3) coeff: 1.073, S.E.:0.2777; col-umn (5) coeff: 0.776, S.E.: 0.374) . Coefficient of financial development even improvesmore compare to the results of pooled OLS with same significance when we use one-stepsystem GMM and two-step system GMM.

Results in columns 2, 4, and 6 show that the positive significant effect of financialdevelopment on growth remains intact when we include financial liberalisation index (col-umn (2) coeff: 0.557, S.E.: 0.095; column (4) coeff: 0.878, S.E.: 0.252; column (6) coeff:0.829, S.E.: 0.308). However, we observe that financial liberalisation on its own is notsignificant (column (2) coeff: -0.052, S.E.: 0.752; column (4) coeff: 0.056, S.E.: 2.153;column (6) coeff: 0.192, S.E.: 2.472). Especially for our preferred estimation techniqueof one-step and two-step system GMM it is positive but not significant, which shows thatthe financial liberalisation does not have an independent effect on economic growth rate.

Most of the control variables are significant in the expected manner. The coefficientof the log of initial level of GDP as expected is negative significant that confirms the


Table 2.6: Baseline Regresssion

Dependent Variable: GDP/Capita Growth Rate

OLS one-step SGMM two-step SGMM

(1) (2) (3) (4) (5) (6)

Log in level GDPPC -0.297 -0.302 -0.902 -0.656 -0.824 -0.687

[0.119]** [0.137]** [0.404]** [0.350]* [0.432]* [0.419]

FD 0.552 0.557 1.073 0.878 0.776 0.829

[0.095]*** [0.095]*** [0.277]*** [0.252]*** [0.374]** [0.308]***

Investment 0.225 0.226 0.288 0.270 0.271 0.271

[0.014]*** [0.014]*** [0.054]*** [0.047]*** [0.051]*** [0.045]***

Trade openness 0.002 0.002 -0.013 -0.009 -0.006 -0.004

[0.003] [0.003] [0.012] [0.010] [0.010] [0.012]

Government size -0.044 -0.042 -0.146 -0.153 -0.081 -0.139

[0.023]* [0.023]* [0.102] [0.080]* [0.108] [0.081]*

Education 0.362 0.374 1.154 0.944 1.207 0.918

[0.216]* [0.217]* [0.894] [0.800] [1.134] [1.141]

Inflation(CPI) 0.303 0.306 0.791 0.765 0.857 0.744

[0.116]*** [0.119]** [0.356]** [0.336]** [0.475]* [0.362]**

FL -0.052 0.056 0.192

[0.752] [2.153] [2.472]

Constant 2.653 2.697 6.896 5.119 4.614 4.812

[1.024]*** [1.040]*** [3.553]* [3.098] [3.716] [3.795]

Observations 340 339 340 339 340 339

R2 0.552 0.554

F 33.587 31.058 7.416 8.381 9.024 10.854

Hansen p-value 0.385 0.279 0.385 0.279

AR1 test p-value 0.000 0.001 0.001 0.001

AR2 test p-value 0.473 0.658 0.380 0.663

No of countries 76 76 76 76

No of instruments 54 66 54 66

The table reports regression coefficients and robust standard errors in brackets. * p<0.10, **p<0.05, *** p<0.01. Columns 1-2 report the estimates of the pooled OLS with and withoutthe financial liberalisation index. Columns 3-4 showing the one-step system GMM resultsand columns 5-6 showing the results for two-steps system GMM. All regressions include timedummies.


conditional convergence hypothesis already proven in many growth related studies (Barro,1991, Mankiw et al., 1992), which implies higher level of GDP per capita growth ratefor lower level of initial GDP given other variables. The coefficient of log of inflation ispositive for our benchmark regression; most of them are insignificant though. The positivecoefficient of inflation is in line with many growth studies conducted on such a largesample of countries representing low, middle, and higher income countries (Pollin andZhu, 2006, Li and Zou, 2002). The coefficient of imports plus exports as a ratio of GDPfor the measure of trade openness shows positive coefficients for the fixed effects, while itshows negative sign when we apply system- GMM on the same specification. Rioja andValev (2004) also finds the negative coefficient for the trade openness coefficient undersystem-GMM. The size of the government has statistically significant negative associationwith growth, which shows that size of government results inefficient allocation. Log ofsecondary school education is positive but not significant which is line with literature oneffect of human capital on economic growth23.

2.4.2 Regression with Interaction

According to our main hypothesis the effect of financial development on long-rungrowth is not independent of the level of financial liberalisation. More precisely, wehypothesize that the positive effect of financial development on economic growth dependson the level of financial liberalisation in a country. To formally test our hypothesis, weuse the regression model described in Equation (2.8). Our regression model provideslink between financial liberalisation, financial development and economic growth. Theinteraction term in the our model measures the varying effect of financial development oneconomic growth through the channel of financial liberalisation and its components.

One-step system GMM is preferred in literature due to its ability to correct smallsample bias (Soto, 2009). For the two-step system GMM we remove small sample biaswith Windmeijer corrected standard errors. As noted by Roodman (2009), researchers be-fore applying the Windmeijer’s correction draw inferences from one-step system GMM.We also present first the results for one-step system GMM and then use the two-stepGMM methodology by applying Windmeijer’s correction.

Table 2.7 displays the results for one-step system GMM. The first columns presents

23For instance, Benhabib and Spiegel (1994) & Pritchett (2001) find weak relationship between humancapital and economic growth


Table 2.7: Financial Liberalisation, Financial Development and Economic Growth, One-Step SystemGMM

Dependent Variable: GDP/Capita Growth Rate(1) (2) (3) (4) (5) (6) (7) (8)

FD 2.688 1.839 2.210 2.452 1.494 1.427 2.845 1.702[0.648]***[0.367]***[0.552]***[0.744]***[0.493]***[0.412]***[0.620]*** [0.682]**

FL 0.877[1.940]

FL*FD -2.804[0.809]***

Banking supervision 1.504[0.425]***

Banking supervision*FD -0.752[0.164]***

Entry barriers -0.534[0.315]*

Entry barriers*FD -0.607[0.206]***

Credit controls 0.185[0.446]

Credit controls *FD -0.718[0.272]**

Security markets 0.630[0.589]

Security markets*FD -0.334[0.219]

Privatization -0.382[0.263]

Privatization*FD -0.503[0.181]***

Interest rate liberalisation 0.436[0.400]

Interest rate liberalisation*FD -0.784[0.211]***

Capital controls 0.122[0.538]

Capital controls*FD -0.349[0.281]

Log in level GDPPC -0.568 -0.822 -0.635 -0.499 -0.952 -0.306 -0.772 -0.664[0.389] [0.363]** [0.391] [0.340] [0.449]** [0.334] [0.366]** [0.379]*

Investment 0.249 0.240 0.258 0.262 0.251 0.264 0.290 0.257[0.044]***[0.044]***[0.054]***[0.042]***[0.042]***[0.044]***[0.044]***[0.046]***

Trade openness -0.011 -0.003 -0.008 -0.010 -0.008 0.001 -0.015 -0.012[0.011] [0.011] [0.010] [0.011] [0.010] [0.010] [0.013] [0.011]

Government size -0.121 -0.145 -0.089 -0.134 -0.056 -0.084 -0.092 -0.150[0.079] [0.079]* [0.073] [0.084] [0.066] [0.079] [0.090] [0.087]*

Education 1.219 0.537 1.246 0.740 0.602 0.918 1.508 1.154[0.846] [0.710] [0.824] [0.821] [0.810] [0.778] [0.767]* [0.806]

Inflation(CPI) 0.223 0.179 0.271 0.368 0.512 0.584 -0.013 0.731[0.324] [0.324] [0.336] [0.338] [0.328] [0.315]* [0.277] [0.382]*

Constant 6.148 6.932 7.546 5.427 6.418 2.325 7.375 5.270[3.087]* [2.825]** [3.367]** [3.029]* [3.473]* [2.903] [3.019]** [2.869]*

Observations 339 339 339 339 339 339 339 339F 9.917 11.434 9.544 8.420 11.202 13.070 8.957 8.401Hansen p-value 0.333 0.572 0.552 0.493 0.181 0.505 0.250 0.209AR1 test p-value 0.000 0.000 0.001 0.000 0.000 0.001 0.001 0.000AR2 test p-value 0.667 0.300 0.664 0.651 0.588 0.459 0.598 0.539No of countries 76 76 76 76 76 76 76 76No of instruments 66 66 66 66 66 66 66 66The table reports regression coefficients for one-step system GMM and robust standard errors in brackets. * p<0.10, ** p<0.05, *** p<0.01.

Financial development here is obtained using PCA measures on six measures of financial development: Log of liquidity liability/GDP, log of depositmoney bank assets/ deposit bank assets plus central bank assets, log of private credit/GDP, log of central bank assets/GDP, log of deposit moneybank assets/ GDP, and log bank credit/bank deposit. All regressions results include time dummies. financial liberalisation index is taken from Abiad,Detragiache and Tressel (2008).


the results for financial liberalisation index along its interaction with our measure of finan-cial development. We use full set of other co-variates in all regressions. The coefficient offinancial development has expected positive sign with significance at 1% and it remainsconsistently positive and significant for for overall financial liberalisation index (coeff.:2.688, S.E.: 0.648) and all of its components. The coefficient of financial development incase of securities markets and privatization component decrease in size, but this declineis very marginal.

The coefficient of financial liberalisation is positive but insignificant (for overall FLindex the coeff: 0.877, S.E.:1.940). Similarly, none of the components of financial lib-eralisation except the component of banking supervision is significant individually. Thisshows that individually financial liberalisation or its components have no direct effect ongrowth. The positive individual significant coefficient of banking supervision shows thata higher regulation has positive effects on economic growth, which shows that implemen-tation of financial liberalisation policies with certain level of monitoring entail positiveresults in terms of higher level of long-run economic growth.

The coefficient of interaction term between financial liberalisation and financial de-velopment in column 1 is negative and significant at 1% level of significance. The co-efficient for financial development is 2.688 with S.E 0.648, whereas, the coefficient ofinteraction between overall financial liberalisation index and the value of financial devel-opment is -2.804 with S.E. 0.809.

For the purpose of illustration for overall effect of financial development on GDP/capitagrowth, we use the derivative of GDP/capita growth rate w.r.t. the level of financial de-velopment for the changing levels of financial liberalisation. Equation (2.9) shows thederivative for GDP/capita growth w.r.t to the level of financial development for the differ-ent levels of financial liberalisation.

∂GDP/Capita Growh∂FD

= 2.688−2.804(FL) (2.9)

The negative coefficient for the interaction term between financial liberalisation andfinancial development shows declining impact of financial development on growth as thelevel of financial liberalisation increases. At certain point the net effect of financial de-velopment on growth becomes negative when financial liberalisation reaches to very highlevel. A threshold value for the financial liberalisation can be determined after plugging inthe values for financial liberalisation index that ranges between 0 to 1 in Equation (2.9).


We find that the values higher than 0.95 of the financial liberalisation index make theeffect of financial development on economic growth negative.

The remaining seven columns contain similar results for the seven components24 offinancial liberalisation index one at a time. All interactions terms for components of fi-nancial liberalisation with financial development are negative and except for stock marketreforms all are highly significant, showing a declining impact of financial developmentwith higher levels of financial liberalisation no matter how financial liberalisation is mea-sured.

The coefficient for the interaction between banking supervision and FD is also neg-ative and significant at 1% level of significance. Negative co-efficient of the interactionbetween banking supervision–for which higher level shows higher level of regulation–and financial development shows that higher level banking supervision individually af-fect economic growth positively. However, the negative coefficient for its interaction withfinancial development shows that higher level of banking regulation also decreases theeffect of financial development on economic growth. Nevertheless, as we observe thatthe coefficient of the interaction term -0.752 and S.E.: 0.164 is smaller compare to thecoefficient of financial development: 1.839 and S.E. 0.367, therefore, the overall effectof financial development on economic growth does not fall considerably for certain levelof banking supervision25. This result makes perfect sense as certain level of banking su-pervision is good, but a very high level of the regulations may have adverse effects onthe ability of banking sector to extend credit and to play its role in the economy and,therefore, may affect growth negatively.

Table 2.8 displays the results for the impact of financial liberalisation on finance-growth relationship for our model in Equation (2.8) using to-step system GMM. The firstcolumn shows the results for overall financial liberalisation index and its interaction withfinancial development and other covariates. The results for two-step system GMM aresimilar to the one-step system GMM results explained above. Financial Development ispositive and significant at 1% level of significance for overall financial liberalisation index(coeff: 2.549, S.E.: 0.662) and all of its individual components. The coefficient of finan-cial liberalisation along with all its components except supervision are not statisticallysignificant, which shows that financial liberalisation and most of its components do not

24These seven components include banking supervisory reforms, relaxation of entry barriers, lesseningof credit controls, liberalisation of stock market, privatization, lessening interest rate controls, and capitalcontrols.

25After normalizing the original banking supervision scale between 0-1, the effect of higher level ofbanking supervision turn negative when the value of banking supervision becomes larger than 0.81.


have individual effect on economic growth. However, the interaction terms for financialliberalisation and its components with financial development are negative and significantat conventional level of significance, which again shows that higher level of financialliberalisation or its components cause dampening effect of the financial development oneconomic growth.

Most of the explanatory variables: initial level level of GDP/ capita, investment,trade openness, size of government, investment in human capital, education and infla-tion appear in the results with expected sign and significance. The initial level of GDPper capita is negative and significant in most of the cases at either 5% or 10%, showingthe validity of convergence hypothesis. Investment growth in physical capital is positiveand highly significant even at 1% showing the positive impact of rise in physical capitalgrowth rate on long-run economic growth. Trade openness has mixed sign but here in allcases it appears as insignificant. The investment in human capital is positive but is notsignificant, which is in line with literature showing ambiguous effect of human capitalon economic growth. Coefficient of the size of government is consistently negative butinsignificant in almost all cases with few exceptions where it shows significance at 10%.

2.4.3 Tests for Instruments

To validate the consistency of the estimates under GMM-system, we use tests in-troduced in (Roodman, 2006) command of xtabond2 for exogeneity. Arellano and Bond(1991) propose two approaches to tests for the exogeneity of the instruments used inGMM-system. The first test is related to the issue of serial correlation, and the secondtests named as Hansen test is suggested to test for the over-identification. The test forserial correlation hypothesize no serial correlation in errors (εi,t). The AR(1) p-valuesfor all specifications are less than 0.01, that suggest that there is autocorrelation in thefirst difference, whereas, the second order no serial correlation between errors can not berejected as all values are greater than 0.01.

The Stata output reports (Hansen, 1982) J statistic instead of the Sargan test whenwe use robust option with our GMM-system. Therefore, our results report (Hansen, 1982)J statistic that has null hypothesis that “the instruments as a group are exogenous”. Thereported p-values over all of our specifications has sufficiently higher p-values than 0.01,which confirms the exogeneity of the instruments in our models and shows that the in-struments are not correlated with disturbance process.


Table 2.8: Financial Liberalisation, Financial Development, and Economic Growth, Two-Step Sys-tem GMM


FD 2.549 1.751 2.023 2.165 1.344 1.323 2.712 2.049[0.662]***[0.438]***[0.464]***[0.725]*** [0.529]** [0.526]** [0.793]***[0.628]***

FL 1.454[1.890]

FL*FD -2.592[0.796]***

Banking supervision 1.090[0.523]**


Entry barriers -0.540[0.416]


credit controls 0.311[0.395]

Credit controls*FD -0.615[0.262]**


Securities markets*FD -0.286[0.257]


Privatization*FD -0.492[0.232]**




Capital controls*FD -0.528[0.247]**

Log in level GDPPC -0.553 -0.780 -0.629 -0.654 -0.943 -0.328 -0.689 -0.717[0.435] [0.378]** [0.452] [0.341]* [0.493]* [0.326] [0.481] [0.429]*

Investment 0.234 0.256 0.244 0.254 0.248 0.264 0.279 0.236[0.049]***[0.050]***[0.060]***[0.047]***[0.042]***[0.040]***[0.049]***[0.058]***

Trade/GDP -0.006 -0.002 -0.001 -0.005 -0.006 0.004 -0.011 -0.009[0.010] [0.008] [0.010] [0.012] [0.010] [0.009] [0.015] [0.010]

Government/GDP -0.124 -0.091 -0.036 -0.095 -0.032 -0.043 -0.090 -0.123[0.086] [0.076] [0.077] [0.089] [0.087] [0.097] [0.111] [0.095]

Education 0.833 0.544 1.205 0.905 0.479 0.751 1.237 1.258[1.030] [0.740] [0.861] [0.917] [0.818] [0.826] [1.104] [0.996]

Inflation(CPI) 0.241 0.118 0.349 0.313 0.554 0.590 0.077 0.573[0.408] [0.342] [0.314] [0.465] [0.468] [0.371] [0.280] [0.347]

Constant 5.513 6.647 5.987 5.554 5.449 1.709 6.154 5.775[3.555] [3.103]** [3.222]* [2.993]* [3.905] [2.731] [3.792] [2.749]**

Observations 339 339 339 339 339 339 339 339F 9.286 12.149 11.675 12.048 9.583 13.867 6.379 14.085Hansen p-value 0.333 0.572 0.552 0.493 0.181 0.505 0.250 0.209AR1 test p-value 0.000 0.000 0.000 0.000 0.000 0.001 0.001 0.000AR2 test p-value 0.715 0.490 0.501 0.646 0.624 0.430 0.669 0.407No of countries 76 76 76 76 76 76 76 76No of instruments 66 66 66 66 66 66 66 66The table reports regression results for two-step system GMM and robust standard errors in brackets. * p<0.10, ** p<0.05, *** p<0.01. Financial

Development index here is obtained using PCA measures on six measures of financial development: Log of liquidity liability/GDP, log of depositmoney bank assets/ deposit bank assets plus central bank assets, log of private credit/GDP, log of central bank assets/GDP, log of deposit money bankassets/ GDP, and log bank credit/bank deposit. Financial liberalisation index is taken from Abiad, Detragiache and Tressel (2008).


2.5 Robustness Analysis

To check for robustness, we apply a number of procedures. Firstly, we check and dis-card outliers. Secondly, we check if our results remain the same for alternate definition offinancial development with and without outliers. Finally, we estimate our relationship be-tween financial liberalisation, finance and growth by redefining the financial liberalisationindex. For this purpose we re-calculate the financial liberalisation index that includes allseven of the components used in our original index calculated by Abiad, Detragiache andTressel (2008), but, this new index contrary to Abiad, Detragiache and Tressel (2008),scales banking supervision in reverse order. That is, higher value on 0-3 scale on newindex shows lessening of banking supervision rather than increasing the banking super-vision as reform– as proposed by Abiad, Detragiache and Tressel (2008) in their originalindex.

2.5.1 Exclusion of Outliers

We perform robustness analysis to our potential outliers and once again examinethe relationship between financial development and economic growth for varying levelsof financial liberalisation. For this purpose we follow Hadi’s (1992) method of multiplevariable outliers that is ‘distance based method’ for omitting outliers. Table 2.9 presentsresults using one-step system GMM. The results for two-step step system GMM are pre-sented in the Appendix (see table A.10). We observe that the coefficient of financialdevelopment to be consistently positive and statistically significant for overall definitionof financial liberalisation (coeff: 1.795, S.E.: 0.401) and for its components without anyexceptions. Most of the coefficients are significant at 1% level of significance, that onceagain shows that the our finance-growth relationship survives well after exclusion of theoutliers.

The interaction terms between financial liberalisation and financial development foroverall definition of financial liberalisation (coeff: -2.274, S.E.: 0.626) and its compo-nents is negative significant, which again confirms the dampening effect of financial de-velopment on long-run economic for higher level of financial liberalisation. We observethat after excluding outliers from our analysis, some of the components along with fi-nancial liberalisation individually become negative significant that shows their negativeindividual direct effect on economic growth. The individual components that becomenegative significant include: lessening of credit control and capital account openness.


Table 2.9: Financial Liberalisation, Financial Development, and Economic Growth, One-Step Sys-tem GMM without Oultiers


FD 1.795 1.193 1.521 1.057 1.151 1.046 1.604 1.273[0.401]***[0.316]***[0.388]*** [0.632]* [0.517]** [0.353]***[0.422]***[0.347]***

FL -3.573[1.633]**

FL*FD -2.274[0.626]***





Credit controls -0.718[0.267]***

Credit controls*FD -0.279[0.224]


Securities markets*FD -0.390[0.224]*



Interest rate liberalisation -0.364[0.260]


capital account openness -0.875[0.391]**


Log in level GDPPC -0.102 -0.505 -0.244 -0.216 -0.599 -0.193 -0.291 -0.434[0.370] [0.319] [0.346] [0.311] [0.386] [0.318] [0.358] [0.334]

Investment 0.280 0.269 0.254 0.280 0.268 0.284 0.275 0.292[0.034]***[0.042]***[0.038]***[0.038]***[0.041]***[0.041]***[0.036]***[0.039]***

Trade openness 0.015 0.011 0.012 0.015 0.005 0.012 0.011 0.009[0.008]* [0.010] [0.008] [0.010] [0.010] [0.008] [0.007] [0.007]

Government size -0.012 -0.082 -0.052 -0.095 -0.006 -0.036 -0.027 0.003[0.063] [0.065] [0.061] [0.072] [0.067] [0.070] [0.061] [0.064]

Inflation(CPI) 0.264 0.214 0.207 0.621 0.190 0.317 0.149 0.331[0.257] [0.250] [0.222] [0.295]** [0.251] [0.246] [0.245] [0.261]

Education 2.247 0.676 1.478 1.593 1.613 1.461 2.080 2.570[0.821]*** [0.832] [0.906] [0.866]* [0.986] [0.970] [0.777]***[0.708]***

Constant 1.100 3.116 1.986 0.907 2.855 0.232 1.834 2.467[2.318] [2.487] [2.212] [2.490] [2.777] [2.260] [2.327] [2.141]

Observations 328 328 328 328 328 328 328 328F 17.995 17.216 18.507 13.696 14.110 13.513 21.434 14.297Hansen p-value 0.758 0.482 0.911 0.891 0.580 0.855 0.808 0.774AR1 test p-value 0.001 0.000 0.000 0.001 0.000 0.001 0.002 0.001AR2 test p-value 0.662 0.990 0.890 0.596 0.874 0.829 0.436 0.655No of countries 73 73 73 73 73 73 73 73No of instruments 66 66 66 66 66 66 66 66The table reports regression results for one-step system GMM after excluding outliers using Hadi (1992) method and robust standard errors are

in brackets. * p<0.10, ** p<0.05, *** p<0.01. Financial Development index here is obtained using PCA measures on six measures of financialdevelopment: Log of liquidity liability/GDP, log of deposit money bank assets/ deposit bank assets plus central bank assets, log of private credit/GDP,log of central bank assets/GDP, log of deposit money bank assets/ GDP, and log bank credit/bank deposit. Financial liberalisation index is taken fromAbiad, Detragiache and Tressel (2008).


However, individually banking supervision is still positive and significant that shows theimportance of banking supervision on economic growth.

2.5.2 Alternate Measures of Financial Development

Although principal component analysis is gaining popularity in finance-growth liter-ature for construction of representative indexes, it can be criticised as a complex measurethat may have little economic interpretation. The biggest criticism this technique hasreceived due to its arbitrariness in the choice of number of components (see Vyas andKumaranayake (2006)). In the following we present our results using alternate definitionsof financial development.

Private credit/GDP ratio. For an alternate definition of financial development26 weuse private credit/GDP ratio, which is one the most commonly used measure of financialdevelopment. Table 2.10 shows one-step system GMM results for the effect of financialdevelopment on growth through financial liberalisation using PVT/Y as an indicator offinancial development. For even further robustness check, we run two-step and one-stepsystem GMM without outliers (see table A.11 & table A.12 in the Appendix).

As in our main analysis, table 2.10 shows the similar signs and significance for finan-cial development, financial liberalisation and their interactions, with only one exception.The coefficient of financial development for individual component of credit control ispositive but not significant. However, rest of the coefficients are positive and significantshowing positive relationship between financial development and economic growth foralmost all definitions of financial liberalisation. The coefficient of financial liberalisationis positive but not significant. Most of other components of financial liberalisation for ourspecification are positive except the coefficient of the component of privatization. They,however, differ in significance. For instance, the component of entry barriers, credit con-trols, privatization and capital account openness are not significant even at 10% level,whereas, banking supervision, securities market reforms, interest rate liberalisation arepositive and significant at 1%, 5%, and 10% levels respectively.

The interactions terms for all variables are consistently negative and significant ex-cept for credit controls. Moreover, as noted before, the size of the coefficients for the in-teractions is small. This implies that the negative impact of financial liberalisation startsbecoming effective on the finance-Growth relationship only at higher levels. Most of

26In our main analysis we used PCA based measure of financial development calculated from six differentmeasures of financial development.


Table 2.10: Financial Liberalisation, Financial Development, and Economic Growth, One-Step Sys-tem GMM with PVT/Y


FD 0.093 0.046 0.064 0.054 0.067 0.033 0.078 0.066[0.028]***[0.017]*** [0.027]** [0.033] [0.029]** [0.019]* [0.031]** [0.029]**

FL 2.085[2.121]

FL*FD -0.102[0.031]***

Banking Supervision 2.045[0.562]***


Entry Barriers 0.162[0.434]

Entry barriers*FD -0.020[0.009]**



Security Markets 1.073[0.492]**

Securities markets*FD -0.022[0.011]**


Privatization*FD -0.012[0.008]

Interest rate liberalisation 0.828[0.457]*

Interest rate liberalisation*FD -0.024[0.011]**

capital account openness 0.192[0.655]

Capital controls*FD -0.021[0.011]*

Log in level GDPPC 0.180 -0.391 -0.086 0.022 -0.066 0.206 -0.021 0.180[0.362] [0.424] [0.378] [0.409] [0.477] [0.394] [0.325] [0.385]

Investment 0.298 0.271 0.271 0.305 0.287 0.294 0.307 0.304[0.045]***[0.044]***[0.046]***[0.046]***[0.042]***[0.048]***[0.046]***[0.046]***

Trade openness 0.002 0.003 -0.000 0.000 0.003 0.008 0.001 0.002[0.006] [0.006] [0.006] [0.005] [0.005] [0.005] [0.006] [0.006]

Government size -0.136 -0.145 -0.099 -0.168 -0.120 -0.084 -0.098 -0.133[0.094] [0.106] [0.074] [0.087]* [0.093] [0.086] [0.109] [0.104]

Inflation 0.689 0.672 0.750 0.831 0.736 0.906 0.711 1.021[0.386]* [0.356]* [0.336]** [0.414]** [0.381]* [0.355]** [0.376]* [0.424]**

Education 0.170 -0.202 0.413 -0.034 -0.043 0.057 -0.144 0.357[0.709] [0.740] [0.693] [0.767] [0.677] [0.731] [0.734] [0.728]

Constant -3.495 -0.308 -1.007 -3.018 -2.720 -4.197 -3.142 -4.287[2.306] [2.581] [2.569] [2.869] [3.139] [2.722] [2.344] [2.336]*

Observations 361 361 361 361 361 361 361 361F 11.378 10.856 12.211 8.457 10.956 10.340 8.283 9.119Hansen p-value 0.423 0.593 0.574 0.665 0.417 0.667 0.345 0.392AR1 test p-value 0.000 0.000 0.001 0.000 0.000 0.001 0.000 0.000AR2 test p-value 0.623 0.338 0.728 0.703 0.665 0.645 0.408 0.570No of countries 77 77 77 77 77 77 77 77No of instruments 66 66 66 66 66 66 66 66

The table reports regression results using one-step system GMM and robust standard errors are in brackets. * p<0.10,** p<0.05, *** p<0.01. Financial Development index here is private credit/GDP ratio. Financial liberalisation index istaken from Abiad, Detragiache and Tressel (2008).


the controls show expected sign at reasonable level of significance as explained in mainanalysis and baseline regression models.

∂GDP/Capita Growh∂FD

= 0.093− .0102(FL) (2.10)

The Equation (2.10) uses estimates from the one-step system GMM using privatecredit/ GDP ratio provided in table 2.10. By plugging different values for financial liber-alisation ranging from 0-1, we find 91.18% (approx) as a threshold for financial liberali-sation after which according to our estimates the effect of financial development on GDPper capita growth rate become negative.

Liquidity Liability Ratio: We use liquid liability ratio (LLR) as another alternatemeasure of financial development to check robustness against our definition of financialdevelopment26. We find that once again our results are robust to the definition of financialdevelopment. Table 2.11 shows the results for LLR as an alternate measure of financialdevelopment, using one-step system GMM technique. Once again for further robustnesswe also present results for LLR using two-step system GMM and to determine if theoutliers are not deriving the results we also run one-step system GMM without outliers(for the results see table A.13 & table A.14 in the Appendix).

We note that the coefficient of financial development is consistently positive andsignificant except for measure of credit control and capital account openness. Financialliberalisation individually and other components show insignificant coefficients, exceptfor banking supervision and interest rate controls, which are positive significant at 1% and10% respectively. The interaction terms also as previously shown using PVT/Y is negativeand significant at reasonable level of significance except for credit control and capitalaccount openness. Other explanatory variables remain almost the same as in baselineregression.

2.5.3 Reconstruction of the Liberalisation Index

As our main index of financial reform uses banking supervision in reverse order,showing higher value of supervision as reform, we reverse this component to constructan index from purely liberalisation perspective. For our new index higher values for allcomponents including banking supervision show higher level of openness27.

27Although according to Abiad, Detragiache and Tressel (2008), the overall index they define explainslevel of openness/liberalisation and reversing one component will not change the results in any material


Table 2.11: Financial Liberalisation, Financial Development, and Economic Growth, One-Step Sys-tem GMM with LLR


FD 0.070 0.045 0.060 0.035 0.053 0.053 0.062 0.051[0.024]*** [0.017]** [0.026]** [0.029] [0.024]** [0.018]*** [0.025]** [0.031]

FL 1.797[2.311]

FL*FD -0.081[0.028]***


Banking supervision*FD -0.019[0.007]**

Entry barriers 0.354[0.510]




Security Markets 0.909[0.625]


Privatization 0.592[0.556]






LLYo 0.271 -0.280 -0.059 0.028 0.117 0.257 0.088 0.191[0.372] [0.435] [0.381] [0.417] [0.480] [0.425] [0.374] [0.382]

Investment 0.303 0.268 0.280 0.312 0.294 0.294 0.320 0.303[0.046]***[0.047]***[0.043]*** [0.048]*** [0.043]***[0.046]***[0.048]***[0.046]***

Trade openness 0.003 0.002 -0.001 -0.004 0.003 0.013 0.001 0.002[0.005] [0.006] [0.006] [0.006] [0.004] [0.005]** [0.006] [0.006]

Government size -0.161 -0.139 -0.104 -0.201 -0.112 -0.090 -0.107 -0.146[0.093]* [0.110] [0.073] [0.089]** [0.094] [0.080] [0.118] [0.101]

Inflation 0.668 0.555 0.638 0.848 0.705 0.566 0.625 0.945[0.419] [0.400] [0.341]* [0.469]* [0.443] [0.312]* [0.403] [0.463]**

Education 0.560 0.105 0.581 0.260 0.073 0.297 -0.063 0.665[0.770] [0.676] [0.700] [0.807] [0.713] [0.752] [0.782] [0.736]

Constant -3.838 -0.682 -1.435 -2.470 -3.933 -4.803 -3.692 -4.140[2.263]* [2.476] [2.547] [2.994] [3.086] [2.784]* [2.513] [2.209]*


The table reports regression results using one-step system GMM and robust standard errors are in brackets. * p<0.10,** p<0.05, *** p<0.01. Financial Development here is Liquidity Liability Ratio. Financial liberalisation index is takenfrom Abiad, Detragiache and Tressel (2008).


Table 2.12: Estimates using Financial Liberalisation Index

(1) (2)One-step System GMM Two-step System GMM

FD 2.903 3.146[0.918]*** [0.870]***

FL 0.324 -0.545[2.143] [1.752]

FL*FD -3.290 -3.569[1.190]*** [1.132]***

Log in level GDPPC -0.416 -0.391[0.481] [0.377]

Investment 0.239 0.257[0.043]*** [0.044]***

Trade openness -0.002 -0.008[0.011] [0.010]

Government size -0.098 -0.113[0.096] [0.080]

Inflation (CPI) 0.368 0.399[0.463] [0.339]

Education 0.949 1.282[1.079] [0.827]

Constant 3.777 4.472[4.141] [3.138]

Observations 339 339F 7.856 8.706Hansen p-value 0.327 0.327AR1 test p-value 0.000 0.001AR2 test p-value 0.490 0.449Number of countries 76 76Number of instruments 66 66

Financial Liberalisation index is the normalised index incorporating our new supervision compo-nent. The reverse banking supervision is calculated by subtracting the actual value of the bankingsupervision indicator from its maximum value. The new banking supervision component ranges from0-3. The value 0 shows the least highest amount of banking supervision and 3 shows the least bankingsupervision. Standard errors in brackets. * p<0.10, ** p<0.05, *** p<0.01

Table 2.12 shows the results for the financial liberalisation index using reversedbanking supervision component without outliers28 using one-step and two-step systemGMM. The column 1 presents results using one-step system GMM and column 2 presentsresults for two-step system GMM.

As expected the coefficient of financial development is highly significant for bothone-step and two-step system GMM showing positive relationship between financial de-velopment and long-run economic growth. The interaction term between the new financialliberalisation index and our measure of financial development consistently show negative

sense, the new index is calculated to confirm the claim and justify the original index as truly liberalisationindex as we claim in our study.

28We apply Hadi (1992) procedure to detect and remove outliers.


significant coefficient, which shows, as previously, that higher level of liberalisation re-duces the positive effect of financial development on economic growth. The financialliberalisation index on its own is not significant and most of the controls appear withexpected signs at reasonable level of significance.

2.6 Conclusion

In this chapter we investigate the conditional effects of financial development oneconomic growth, using financial sector liberalization as a conditioning variable. Morespecifically, we study the implications of financial liberalization interacting with financialdevelopment in impacting on economic growth. Our aim is to combine financial devel-opment, financial liberalization and growth through two testable hypotheses: first, therelationship between financial development and economic growth is conditional upon thelevel of liberalization in the financial sector; and second, an excessive level of liberaliza-tion is associated with a weak or negative effect of financial development on economicgrowth.

We employ a comprehensive measure of financial development by applying PCAon the frequently used measures of financial development across the time, while forrobustness we use two most prominent among them: private credit to GDP ratio andLiquid-Libaiblity ratio. Further, we use a multiplicative interaction model to capture theconditional effects of financial development on growth which is estimated by employingone-step and two-step system GMM estimators to take account of country specific char-acteristics, as well as dynamics and endogeneity. We take care of influential outliers byapplying the Hadi method of multiple variable outliers.

Our regression results show that the marginal effects of financial development oneconomic growth is positive and significant, while the marginal effect of financial liberal-ization is generally insignificant. Further, the relationship between financial developmentand growth is conditional upon the level of financial liberalisation; that is, it decreasesas the level of liberalization increases and even becomes negative at very high levels ofliberalization. Thus, the negative interaction between financial development and finan-cial liberalization suggests that at a very high level of liberalization adding more financialdevelopment may not be a growth promoting policy.

The results are consistent across all specifications with different definitions of fi-nancial development and for majority of the components of financial liberalisation index.


Our results are also robust to the elimination of outliers and for our new index of financialliberalisation. The findings in this chapter adds to the growing empirical work analysingthe causes of the breakdown of the finance-growth relationship. It suggests that relax-ing the rules and regulations have amplified the impacts of inherent weaknesses in thefinancial system on the economies. The countries where financial liberalisation crosses acertain level the effect of financial development vanishes and can also become negative iffinancial liberalisation increases unabatedly.

On the policy front, our study sugests the followings: first, the governments shouldnot go for excessive liberalization of the financial sector by removing the regulationstoo much in order to promote financial sector development with an expectation to fostergrowth; second, deregulation in the financial sector does not have automatic positive ef-fect, instead government should always be watchful so that excessive liberalization doesnot take place.

Chapter 3

What Makes Financial liberalisationWork: The Impact of Sequencing onFinancial Crises

3.1 Introduction

Substantive liberalisation of financial markets started in 1970s and gathered momen-tum during the 1980s and 1990s. The predominant motive for financial sector reform wasto increase the role of markets in the determination of interest rates and in the allocationof credit (Caprio et al., 1996). Under these reforms governments took steps to privatisestate owned financial institutions, correct fiscal imbalances, lower trade barriers and elim-inate capital controls (Edwards, 2009). The initial expectations of these reform were thatthey would enhance economic growth through their positive impact on financial develop-ment without compromising the macroeconomic stability (Easterly et al., 1997, Arun andTurner, 2002).

After four decades, however, there are still ambiguities present in the theoretical andempirical literature related to the exact impact of financial reforms on financial sector andoverall economic stability. There are many reasons for these ambiguities: the first mightbe due to the way the financial reforms are introduced1. Secondly the effects of these poli-cies might be different in the short and long-run (Bandiera et al., 2000). Thirdly, instabil-ity may arise due to low level of government credibility (Calvo, 1987, 1989). Fourthly, as

1Edwards (1990) argues that overall speed and sequencing of economic reforms is the least understoodaspect of overall economic reforms.

64

Chapter 3. What Makes Financial Liberalisation Work 65

Demirgüç-Kunt and Detragiache (1998b) argue that the liberalisation may affect stabilityas it increases chances of high risk taking behaviour on the part of banks and other finan-cial institutions. Finally, Caprio and Summers (1993) & Hellmann et al. (1994) show thatanother reason for instability might be the higher competition resulting from the removalof ceilings on deposit interest rates and entry barriers. According to them higher com-petition reduces monopoly profits that lead to the choice of risky portfolios to maintainprofitability.

Edwards (2009) suggests that financial reform sequencing is the core mechanismfor successful reform policies. The right sequencing has potential to bring a controlledform of liberalisation, which results in stable economic outcomes. A wide literature, forinstance, argues that removing capital controls too early can cause real trouble for thehealth of an economy (Edwards, 2009, Massad, 1998). According to this literature, pro-gression to more liberal capital accounts leads to a higher level of capital inflows. Insmall financially liberalised economies these higher level of capital inflows cause assetprice and exchange rate inflation. Higher capital availability also increases domestic de-mand that raises inflation rate and effect the current account deficit adversely (Massad,1998). Moreover, the capital flows in countries as a result of capital account liberalisationare of temporary nature, which flow in search of higher return. These type of flows cancreate further problems if they cease abruptly.

The main objective of this chapter is to estimate the impact of the sequence of finan-cial sector reforms on the probability of financial crises. Using panel data for 28 countriesfrom 1973 through 2005 this study investigates whether a particular sequence of financialsector reform has a bearing on the probability of financial crisis. This study is uniquefrom previous studies on sequencing of financial liberalisation and crisis in several ways.Firstly, we introduce distance–deviations of the level of financial reform from a bench-mark/reference level of reform2 – as a proxy for sequencing of financial sector reforms.In the literature, we find data for financial sector liberalisation/reform, but there is nomeasure for the sequence of financial reform available. The literature generally studiesthe impact sequencing by giving anecdotal evidences and qualitative analysis. In thischapter, we endeavour to provide a distance measure that can represent the measure forsequence. We use deviations of reform measures of countries from our benchmark coun-tries that follow a particular sequence. Secondly, the scope of this study is larger thanprevious studies in terms of the number of sectors analysed to address the question of

2This method was introduced by Zeleny (1974). Now many studies use this method to construct varietyof indexes (see for instance, Nathan et al. (2008) and Sarma (2008)).


appropriate sequencing of financial sector reforms. Most of the available literature takesinto account the choice across the trade liberalisation and financial liberalisation (Ed-wards, 2009) or real sector liberalisation and financial sector liberalisation (Arestis andDemetriades, 1997). In this study we seek to identify an optimal sequence of financialreform between domestic market, stock markets, and capital accounts. Finally, using ourproxy of liberalisation sequence we analyse the impact of different sequencing strategiesof financial sector reforms on financial crises.

To analyse the impact of reform sequencing on financial crises we compare the ex-perience of countries which have followed different patterns of financial reforms. In thispursuit we face two primary challenges. The first issue is of taking into account theprogress and reversals in the financial reforms. Secondly, since the available literaturelacks a concrete variable that measures the sequencing of the reforms, another issue is ofdesigning a proxy variable for measuring sequencing of financial reforms.

In this chapter we define financial sector as a combination of three different sectors:the domestic financial sector, the capital account and the stock market3. Moreover, byfinancial sector reforms we mean removal of repressive interest rates structure, allow-ing foreign currency deposits, allowing foreigners to hold domestic equity, and allowingbanking and other financial institutions to borrow abroad. We record progress and rever-sals by giving ranks to the liberalisation policies between 1 and 3. Where, 1 shows fullrepression, 2 shows partial liberalisation and 3 shows full liberalisation4. By giving rank-ing between 1 and 3 on yearly basis, we can take into account the progress and reversalsof financial reforms every year.

To design a proxy for sequencing of financial reforms we build a distance metric.The metric for distance is defined as the square of Euclidean distance. We measure Eu-clidean distance as a sum of the squares of the differences between rankings of financialreforms of our countries in sample and our benchmark country over different years. Themeasure of distance provides a comparison between sequences that our sample and bench-mark countries have followed to liberalise their financial sector. This methodology thatinvolves measure of a distance from ideal dates back to Zeleny (1974). Now this distancebased approach has been used in literature to construct different types of indexes. For

3Kaminsky and Schmukler (2008) define deregulation of the domestic banking industry, removal ofcontrols on international capital flows, and the liberalisation of the domestic stock market as the mainaspects of financial liberalisation. These aspects in our view comprehensively capture the domestic andinternational liberalisation of financial sector.

4These ranks are recoded in reverse order from the original ranks given by Kaminsky and Schmukler(2008) in their data set on financial liberalisation. They record 1 as fully liberalised, 2 as partially repressedand 3 as fully repressed.


instance, Nathan et al. (2008) used the distance-based measure from ideal to constructHDI index. According to them this distance-based measure from ideal provides a signalfor future course of action. Sarma (2008) also uses the distance based measure from idealto construct and index of financial inclusion. Therefore, a measure of sequencing basedon distance from ideal can be useful in understanding the impact of a deviations from aparticular sequencing–that our reference country follows– on likelihood of emergence offinancial crises.

By using a novel approach to measure the sequencing of financial reforms and cover-ing both domestic and international dimensions of the policies of financial liberalisation,this study draws some general conclusions. Using standard probit model we observe thatthe sequence which a particular country follows has a bearing on financial crises. Betweenthe domestic sector, the stock market and the capital account, a sequence that emerges asthe most resilient to crisis is the sequence that places the capital account liberalisationlast. However, according to our results a country can choose between stock market anddomestic market liberalisation as the first step towards the liberalisation of the financialsector.

The rest of the chapter is structured as follows: Section 3.2 reviews the literature oncauses of financial crises, objectives of optimal financial reforms, and linking them withtheir impact on financial crises. Section 3.3 describes the different sequencing patternsthat different countries have followed and criteria for benchmarks. Section 3.4 defines thesample, main dependent and independent variable and controls. Section 3.5 explains themethodology used in this chapter. Section 3.6 provides econometric results based on theprobit model and also provide results for different robustness checks. Section 3.7 providesconcluding remarks.

3.2 Literature Review

This section reviews the theoretical and empirical literature on the causes of finan-cial crisis, objectives of financial reform, along with linking the optimal sequencing offinancial reforms with its impact on financial crises.

3.2.1 What Causes Systemic Fragility and Financial Crises?

The role and scope of the financial sector has changed remarkably during the lastfew decades. The modern financial sector is strongly connected with other sectors of


the economy through its functions of borrowing and lending. Credit from banks is moreeasily available for investments than ever before, connecting virtually every economicagent in the economy to the banks and other financial institutions.

This interwoven nature of the financial sector to the other sectors makes it systemic–failure of the financial system causes failure of the other sectors in the economy. Thereare many reasons for the financial sector to be systemically important. The first reasonis the inherent interdependence between the financial firms (McCleskey, 2010, p.3). Ac-cording to McCleskey (2010) financial firms exchange cash between themselves for shortterm borrowing and lending in the form of repurchase agreements and other short terminstruments. For the purpose of collateral, these transactions are backed by the long-termassets held by these institutions. Therefore, the credibility of these transactions betweenfinancial institutions depends upon the valuation of the assets being used to back up thesetransactions. This type of interdependence gives rise to two types of contagion channels:“counter-party risk” and “information transmission” (Chakrabarty and Zhang, 2012). Thecounter-party risk and information transmission play vital role in spreading the effect ofthe fall of the big institutions to the other firms (Chakrabarty and Zhang, 2012). Accord-ing to Giesecke (2004), information transmission mechanism causes financial distress tospread to the whole financial system when individual firms update their information aboutdefault of other firms over time.

Therefore, a small loss of value to the assets being used to back the loans can cre-ate havoc in the markets affecting all the financial institutions connected with each otherthrough these types of borrowing and lending. The collapse of Lehman Brothers and BearStearns in 2007-08, leading to the worst recession in the U.S. after the great depressionof 1930s, shows the intensity of the inter linkages between financial institutions and theworsening impacts this interconnectedness put on the whole economy (McCleskey, 2010).Following the bankruptcy filed by Lehman Brothers, more than one hundred firms re-vealed their financial linkages and exposures to Lehman (Chakrabarty and Zhang, 2012),which shows how large contagious effects can arise when big financial firms in the well-connected financial systems fail.

Similarly, commercial banking that covers the largest portion of financial activitieslink the financial sector to real economy. Distrust about banks raises alarms that causelarge scale withdrawals. Heavy withdrawals from the banking system affects lending ca-pability of banks that brings chaos to the whole economic system. Rajan (2006) attributesthe spread of risks to the ease of communication and advancement in information tech-nology along with deregulation and institutional change that have brought the financial


sector so near to the other sectors.

Due to this interconnectedness when the financial sector fails, it leads to the failureof other sector connected directly or indirectly through their functioning of borrowingand lending. Caprio (2003) estimated the cost of banking crisis in Japan as 20 percent ofGDP. Barth et al. (2006) estimate a fiscal cost of more than $ 1 trillion of the crises in thebanking sector for developing countries during 1980s and 1990s. Similarly the bankingcrisis in Chile and Argentina of the early 1980s exceeded 40 percent of GDP. The extentof the scale of crises in financial sector brings disruptions to the whole economies. Oncethe financial sector is in trouble, it is unlikely to help other sectors of the economy .

3.2.2 What should Optimal Financial Reforms Achieve?

The overall purpose of financial reforms is to facilitate the functioning of finan-cial institutions and bring markets into play for efficient allocation of financial assets.The policies of financial liberalisation are introduced to enhance economic performance.These reforms were intended to optimise the role of financial intermediaries in mobi-lization and efficient allocation of savings, reducing cost by improving competitivenessamong financial institutions, and providing bigger range of financial products to tailoredinvestment needs (Fischer, 1993).

The evidence related to the success or failure of financial reforms in many aspectsis mixed. The overall growth rates in many countries have risen considerably (Ranciereet al., 2006, Johnston and Sundararajan, 1999), financial systems have become more ad-vanced, banking and other financial services are available at the doorstep of even the mostremote areas, investor can diversify risks considerably, and many other goods the financialreforms have brought to the world. But at the same time the financial world has becomemore risky and instabilities in the system have also risen (see Stiglitz (2000b)).

The primary objective of financial liberalisation is increased growth through in-creased investment and higher opportunities for mutual trade between local and interna-tional markets. Liberalisation of the financial markets provides opportunities to both thelocal and foreign investors to transact and gain the benefits of trade based on comparativeadvantage. The impact of financial liberalisation on economic growth is mixed. Manystudies using cross-country relationship emphasise on the long term benefits of financialliberalisation through a positive link between real interest rates and economic growth orefficiency of investments (see for instance, Gelb (1989)). However, Arestis et al. (2002)emphasise that due to high fragility of the financial system, the studies showing long term


benefits of financial liberalisation using cross-country relationship have become flawed.

The second objective of financial reforms policies is to maintain macroeconomicstability while targeting economic growth. Arestis and Glickman (2002) analyse the in-stability threats to the growth and employment emanating from the financial sector un-der financial liberalisation. Analysing the Minsky’s hypothesis of “financial instability”,Arestis and Glickman (2002) show that financial liberalisation increase the threats to thegrowth and employment. A big strand of literature associate the rise in probability ofcrises to the liberalisation of the financial sector (see for instance, Hellmann et al. (2000),Arestis and Glickman (2002)). Financial liberalisation cause “Speculation led growth” inLDCs that result in risky investments, unstable financial structure that leads to the higherchances of crises (Grabel, 1995). However, at the same time, it is also important to takeinto account that financial liberalisation increases the volume of the transactions and in-creases the exposure of the financial sector, which might cause short term instabilities inthe system. These short-term instabilities, however, disappear in the long run (see Noy(2004), Bandiera et al. (2000))

The third objective of financial reforms is to enhance the allocative efficiency offinancial resources. While on the one hand Cho (1988), studying the impact of the policiesof financial liberalisation in Korea, finds a significant positive impact of the policies offinancial liberalisation on efficiency of allocation of credit. Similarly, Han Kim and Singal(2000) shows that stock market liberalisation increases the market efficiency that affectspositively to the productivity of capital. Abiad, Oomes and Ueda (2008) also concludethat financial liberalisation improves efficiency by creating equal access to credit andreducing the variability of expected return across firms. On the other hand for sub-SaharanAfrica, Lensink (1996) show that the introduction of financial liberalisation of the formalsector without taking into account the information of the informal sector led to decline insavings and hence quality of investment.

3.2.3 Sequencing of Financial Reforms and Crises

Johnston and Sundararajan (1999) emphasise that the financial sector reforms shouldbe introduced in such a manner that they improve the efficiency of the financial sectorwhilst keeping financial stability intact. Initial experience of reforms failed miserablywhen Southern Cone countries comprising Argentina, Chile and Uruguay plunged intoeconomic difficulties following liberalisation of their financial sectors. The dismal ex-perience of these countries with financial reforms triggered a very strong debate about


the effectiveness and management of financial reform process. The literature that fol-lowed these series of events on the one hand is divided on the issue between “big-bang”versus gradual implementation of reforms (see Bhagwati (1982), Choksi and Papageor-giou (1986), Calvo et al. (1993), Wei (1997)), and on the other hand between differentsequencings of financial reforms of different sectors of the financial system (see for in-stance, (Edwards, 2009)).

The proponents of the big-bang approach assume that it is nearly impossible to sug-gest a sequence for the financial reforms, therefore, big-bang or a shock approach is theonly option available for reforms in the financial sector (see for example Choksi and Pa-pageorgiou (1986)). According to Wei (1997) the big bang approach is preferable if thereis a strong support for the implementation of reform, as due to political support it canbring benefit faster than the gradualist approach. Whereas, some criticise the big-bangapproach as it causes a great amount of dislocation of GDP and employment and mightpotentially be dangerous for the economy because of the conflicting nature of the re-forms. Wei (1997) also support the gradualist approach as against of big bang approachwhen there is ex-ante uncertainty present about the success or failure of reform process.

Agénor and Montiel (2008, p.663) emphasise the timing, speed and sequencing ofreforms against the desirable policy of introducing all reforms together because of theassociated adjustments costs and political and administrative constraints. A vast literaturediscusses conditions required as a prerequisite for liberalisation and especially relaxingcapital controls5. The major debate in sequencing literature revolves around liberalisationof domestic capital market and relaxation of capital controls, order between trade andcapital account, and further also this debate involves the ordering between other structuralreforms and financial sector reforms (Edwards, 1990, Agénor and Montiel, 2008).

Eichengreen (2001) notes that “Capital account liberalization, it is fair to say, re-mains one of the most controversial and least understood policies of our day”. Theoreti-cally, liberalisation of capital accounts can bring benefits for the developing countries interms of higher level of cheaper funds available for investments (Johnston, 1998). How-ever, the downside of these flows is that they flood local market with short-run capital.These short-term flows results in local currency appreciation, which causes real exchangerate to rise. The appreciation of real exchange rate makes exports of a local country lesscompetitive. In the presence of less developed domestic financial sector Johnston (1998)highlight the risk of misuse of these funds. Moreover, there is likelihood that investors

5see Dooley (1995) for detailed literature for review.


will stop this short term flow of capital even at the slightest hint of domestic economic in-stability. If these capital flows cease abruptly, recipient countries force into all economicand financial problems6. Conley and Maloney (1995) using two period model explainthat boom-bust cycle in the period two is associated with the opening of capital accountthat results in rise in consumption leading to current account deficits and sharp rise inborrowing by private agents.

The debate about the optimal sequencing of financial liberalisation has always takena centre stage in literature related the implementation of reforms. The available litera-ture provides evidence of success and failure of many countries that followed differentsequencing and speed of financial reforms. However, as Galbis (1994) argues that noneof those experiences can be termed as an optimal solution. Nevertheless, we find consen-sus both in theoretical and empirical literature that capital account liberalisation shouldcome last on the list of financial reforms (Johnston, 1998, Edwards, 2009). Stiglitz (2002)argues that the IMF’s policies that forced emerging countries to follow a wrong patternof opening of the financial sector led to the crises in those countries. Stiglitz stronglybelieves that IMF′s policies of forcing the countries to go for full capital account liberal-isation was flawed. According to Stiglitz (2002) “capital account liberalisation was thesingle most important factor leading to the crisis” in East Asia.

Johnston and Sundararajan (1999) argue for an integrated approach to capital ac-count liberalisation and domestic financial reforms. They argue that capital account liber-alisation should be sequenced in a manner that reinforces domestic financial liberalisationand allows for institutional capacity building to manage the additional risks. If domesticsector’s liberalisation is not complete, liberalisation of the capital account sector mightlead to the misallocation of resources (Agénor and Montiel, 2008).

The foregoing literature review suggests a clear link between the sequence of finan-cial reforms and the likelihood of crises in many countries. However, we find dearth ofsystematic analysis about the link between sequence of financial reforms and financialcrises. Especially there is no study available that uses a quantitative measure to anal-yse the impact of sequencing of financial reforms on crisis. The present study fills thisgap by introducing a quantitative proxy measure for the sequencing of financial reforms.Moreover, this study also examines the possibility for an optimal sequencing of financialreforms that may reduce the likelihood of financial crises.

6see Rodrik and Velasco (1999)


3.3 Sequencing of Reforms: Choice of Benchmark

The literature about financial reforms, their sequencing and crises suggests that thereseems to be a consensus among economists and policy makers that the sequence of do-mestic sector liberalisation first and capital account liberalisation last is expected to leadto higher growth, better macroeconomic stability and less vulnerability to financial cri-sis. In this section, we identify the countries that followed this sequence, use them asbenchmark/reference countries, one at a time, and examine whether deviation away fromthe benchmark country’s liberalisation scores leads to increased vulnerability to financialcrisis.

For the choice of a benchmark sequencing we select a country as our reference pointon the basis of the sequence the country has followed and its better performance comparedto other countries in terms of our outcome variable of crisis. Primarily, we choose a singlesequence that our first reference country followed as our benchmark. Then we performour analysis by generalizing the benchmark to all countries. Through this process wecan analyse how deviations from different countries– following different sequencing offinancial reforms– affect the probability of crisis. For stable macroeconomic conditions,theoretically, a preferred sequence should follow a pattern which does not expose theeconomy to the capital flows and especially capital inflows from other countries when itsfinancial sector is in its nascent stage (Agénor and Montiel, 2008). Empirical evidencealso suggests that the countries like Chile that opened their capital account only graduallywith more emphasis on prudential regulations have been able to attain stability and haveavoided crises consistently.

Our sample consists of countries that have followed different sequencings for open-ing of their financial sectors. If we group countries according to the first sector theyopened fully, we can categorise them into three groups. The first group that liberalised itsdomestic markets first, the second group that liberalised its capital account first and thethird group that liberalised the stock market first.

Moving forward through this initial criteria table 3.1 shows further groupings basedon sequencing represented by the dates of liberalisation between domestic sector, stockmarket and capital account. Broadly speaking we can make six further groupings on thebasis of the similarity in the sequences7 of our sample countries. The first group com-prising Brazil, Chile, Colombia, Ireland, and Philippines liberalised domestic sector first,

7For sizable groups we cannot find exactly the same sequencing of reform, therefore, for the comparisonpurpose we choose sequencing that are closely similar.


followed by stock market and then capital market. The second group comprising Ar-gentina, Indonesia, Korea, Norway, and Taiwan followed the sequence of domestic sectorfirst, capital market second and stock market last. The third group comprises Finland,Japan, Peru, and Venezuela that opened capital account first, then stock market and inthe end domestic sector. The fourth group comprises one country in our sample that isMexico that opened capital account first, then domestic sector and then stock markets.The fifth group comprises Canada, Denmark, France, Italy, Portugal, Sweden, and HongKong– they opened their stock market then domestic sector and then capital account.Sixth group comprises Malaysia, Spain, Thailand, United Kingdom, and United Statesthat opened their stock market first, then their capital account and finally their domesticmarket.

Table 3.1: The Sequence of Full liberalisation in Sample Countries

Country Domestic Market Capital Account Stock Markets

Brazil Jan-76 Jan-00 Jun-91Chile May-75 Sept-98 Jan-92Colombia Sep-80 Sep-98 Jan-91Ireland Feb-86 Jan-92 Jan-92Philippines Dec-82 N/A Jan-94Argentina Jan-77 Dec-78 Jan-89Indonesia Jan-83 Jan-88 Aug-89Korea Jan-95 Jan-96 May-98Norway Jan-79 Jan-88 Jan-89Taiwan Jul-89 Jan-97 Apr-98Finland Jan-90 Jun-89 Jan-90Japan Dec-91 Jul-80 Jan-85Peru Jan-91 Jan-91 Jan-92Venezuela Aug-81 Jan-73 Jan-77Mexico Apr-89 Jan-73 Jan-91Canada Jan-73 Mar-75 Jan-73Denmark Jan-81 Oct-88 Jan-73France Jan-85 Jan-90 Jan-73Germany Jan-73 Mar-81 Jan-73Hong Kong Jul-01 Jan-73 Jan-73Italy Jan-81 Jan-92 Jan-73Portugal Mar-90 Aug-92 Jan-73Sweden Jan-85 Jan-89 Jan-80Malaysia Apr-04 Sep-94 Jan-84Spain Jan-81 Jan-80 Jan-73Thailand Jun-92 Jan-92 Jan-90United Kingdom Jan-81 Oct-79 Jan-73United States Jan-82 Jul-73 Jan-73Source: Authors calculations

FirstSecodThird

All countries, with a few notable exceptions, have maintained full liberalisation onceafter opening their different sectors. As the progresses and reversals in the reforms ofdifferent sectors may weaken the meaning of our distance measure, we look at the numberof reversals in the sample of countries for all three sectors. Table B.1 in the Appendix


provides the list of number of reversals for the sample, which asserts that there are notmany reversals in reforms. This assures us that our distance measure would not be biaseddue to reversals.

The experience of the countries in different groups is mixed in terms of the crisesthat they have faced during the last four decades. The first three columns of the table 3.2provides frequency of the occurrence of banking, currency and twin crises8 in our samplecountries for the whole four decade, while the last three columns provide the same forthe period after 1984. Total number of crisis episodes9 during our sample period is 192for banking crisis, 182 for currency crises and 58 episodes of twin crises. Whereas, outof 192 episodes of currency crises 140 episodes of banking crises occurred after 1984 in24 countries out of 28 countries from our sample. The countries that avoided consistentlybanking crises after 1984 included Chile, Germany, Ireland, and Portugal.

Table B.2 in the Appendix provides data on average levels of financial reforms indexfor three sectors for 1973-2005. We divide our sample ranging from 1973 to 2005 in threesub-periods10. The first period ranges from 1973-83, the second ranges from 1984-94 andthe last period ranges from 1995-2005. The financial reforms indexes range from 1 to 3,with higher value showing higher level of liberalisation in the respective sectors for theperiod11. The data in the table show gradual progression for all countries from less liberalto more liberal in three time periods. The values in the table present the speed of reform indifferent sectors, higher value expressing higher speed, which shows if behaviour of thecountries towards liberalisation is cautious or reckless. As mentioned earlier a cautiousbehaviour can be categorised as a behaviour under which countries give more importanceto the domestic markets or stock markets liberalisation than to the capital account in theirearly stages of opening up of their financial sector so that they could save themselves fromun-stabilizing capital inflows or outflows.

8Following Reinhart and Rogoff (2009) we define banking crisis as bank runs, currency crisis as 15%or more annual depreciation of local currency against dollar or other major currencies, and twin crisis aswhen a country faces both banking and currency crises simultaneously. The detailed definitions for crisesare provided in section 3.4.

9Number of episodes define here shows number of years countries faced crisis.10We define these sub-periods as initial, middle and final stages of liberalisation over our sample period.

The initial period shows the level of liberalisation in the start of the liberalisation process in our samplecountries, whereas, the middle stage takes into account the corrective actions in form or reversals andprogression of liberalisation following the initial experiences with the liberalisation and the final stagerepresents the final policy stance on liberalisation.

11We use Kaminsky and Schmukler’s (2008) indexes for financial liberalisation in reverse order to takeinto account the extent of liberalisation for our sample countries for the domestic market, stock market andcapital account liberalisation. Kaminsky and Schmukler (2008) uses the scale of 1 to show full liberalisa-tion, 2 for partial liberalisation and 3 for repression.


Table 3.2: Frequency of Crises

1973-2005 1984-2005Country Banking Currency Twin Banking Currency Twin

Crisis Crisis Crises Crisis Crisis Crises1 2 3 4 5 6

Argentina 10 20 6 7 9 3Brazil 6 24 4 6 14 4Canada 3 0 0 1 0 0Chile 5 13 5 0 3 0Columbia 8 19 8 5 13 5Denmark 6 0 0 6 0 0Finland 4 1 1 4 1 1France 2 2 0 2 1 0Germany 3 4 0 0 3 0Hong Kong 8 4 1 8 3 1Indonesia 8 5 3 8 3 3Ireland 0 5 0 0 3 0Italy 6 3 1 6 2 1Japan 10 1 0 10 0 0Korea 11 4 1 10 1 1Malaysia 9 1 1 9 1 1Mexico 9 11 4 7 7 3Norway 7 2 0 7 1 0Peru 9 18 7 7 9 5Philipines 12 5 4 8 3 2Portugal 0 7 0 0 1 0Spain 9 4 2 1 2 0Sweden 4 4 1 4 2 1Taiwan 4 1 0 2 1 0Thailand 14 3 3 9 2 2United Kingdom 6 6 2 2 1 0United States 8 3 0 7 2 0Venezuela 11 12 4 4 11 3Total 192 182 58 140 99 36

The table presents frequency of crises in our sample countries for two different timeperiods. Columns 1, 2, & 3 define number of banking, currency and twin crises for theperiod of 1973-2005 (full sample period) and columns 4, 5, & 6 show the number of crisesfor the period between 1984-2005.

The countries that fit to the criteria are few. Chile again fits to the criterion wellin terms of its average level of liberalisation in three periods during 1973-2005. Theaverage level of financial reforms in Chile during 1973-83 period was 2.55 for domesticsector, 1.36 for capital account and 1.00 for stock market, which shows cautiousnesstowards opening the capital account in the start of liberalisation process while givingmore emphasis to the opening of domestic sector.

To summarise, among 28 countries, only Chile fits into all three criteria we analysedabove: (i) Chile followed the sequence that is supported by the majority of economists andpolicy makers; (ii) Chile is one of the very few countries in the sample, which recoveredfrom its financial crisis robustly and since 1984 successfully avoided any banking crisis;


(iii) Chile also shows cautiousness in opening up of its capital market when we look atthe speed of reform scenario. On this basis we chose Chile’s ordering of reforms as anindicative benchmark sequencing. Many studies attribute consistent performance of Chileto the order in which it opened its financial sector (see for example, Massad (1998)). Chileis the only country out of that group that recovered from the worst crisis and maintaineda stable economy during all crises and even during the recent financial crisis of 2007-08(see Hornbeck (2009)).

A large number of studies have taken Chile as a reference country because of itsunique sequencing of financial reforms (McKinnon, 1982, Echeverria et al., 2004, Mas-sad, 1998, Hornbeck, 2009). It opened its current account before relaxing the restrictionson capital movement. It went for rapid elimination of trade restrictions while freeing cap-ital movement only gradually (Velasco, 1991, Echeverria et al., 2004). McKinnon (1982)describes Chile as“ a norm or standard of reference”. He argues that despite some earlymistakes during 1973-1975 after military take over on the part of Chile, the liberalisationorder laid out by Chile is successful.

Massad (1998) argues that the macroeconomic success of Chile during the 1990swas majorly due to its strategy of opening capital account gradually along with othermacroeconomic policies. Chile’s step by step policy of opening capital accounts providesa good example of neither shutting doors to international capital nor allowing foreigncapital to flood local markets and hamper macroeconomic performance.

Hornbeck’s (2009) report presented to Congress uses the reforms of Chile as a guide-line for the USA for future stance on regulation and financial reforms. The report con-cludes that due to continuous updating of the regulatory system the Chilean financial sys-tem has been able to maintain competitiveness with prudence. The report identifies thatthe Chilean banking system, due to wise policy changes, has emerged as the most stablebanking system among emerging economies in 2009. According to the report the Chileancapital adequacy ratio is well above 12% similar to the United States with a banking sectorwith an annual average return on equity of 15% over the past.

Historically, the reforms introduced by Chilean economy were comprehensive thatinvolved all markets. In the mid 1970s Chile’s economy transformed completely fromstate controlled to an opened economy. It introduced complete deregulation, overhauledthe financial system, introduced free pricing system, slashed government expenditures,and privatised many state owned organizations (Collins, 1995, Velasco, 1991). How-ever, initial experience of Chile with financial reforms worsened when Chilean economyplunged into the worst crisis in early 1980s. Following the crisis, Chile transformed and


improved its regulatory policy by overhauling its General Banking Act. They simplifiedand integrated their financial system to the regulatory policies.

3.4 Data

In this chapter we use a sample consisting of a set of 28 countries over the periodfrom 1973-2005. We choose this sample mainly on the basis of available data for the ex-tent of liberalisation for domestic markets, stock markets, and capital accounts. Table B.3in the Appendix provides list of countries in our sample. The sample of countries has rep-resentations from both emerging and developed countries. The representative countriesare from the groups of the G-7 countries12 , the Asian countries13, the European group14,and the Latin American countries15. Table B.4 in the Appendix provide definition, con-struction and sources of the data set.

3.4.1 The Dependent Variable: Financial Crises

We define systemic crisis as “the likelihood of a sudden, usually unexpected, col-lapse of confidence in a significant portion of the banking or financial system with poten-tially large real economic effects” (Bartholomew and Whalen, 1995, pg.7). We use theoccurrence of financial crisis as currency, banking or twin crises dummy as a dependentvariable. For this purpose, we use Reinhart and Rogoff’s (2009) definition of currencyand banking crises16. They define currency crisis as 15% or more annual depreciation oflocal currency vs. US dollar or some relevant anchor currency like UK pound, the Frenchfranc, the German DM or Euro. Their definition of banking crisis is based on the eventof bank runs that lead to the closure, merging or takeover by the public sector of one ormore financial institutions.

We also use twin crises (banking and currency crisis together) dummy for our anal-ysis following Kaminsky and Reinhart (1999) and Bordo et al. (2001). The choice of

12G-7 countries include Canada, France, Germany, Italy, Japan, the United Kingdom, and the UnitedStates.

13The Asian region includes Hong Kong, Indonesia, Malaysia, the Philippines, (South) Korea, Taiwan,and Thailand.

14The European group comprises Denmark, Finland, Ireland, Norway, Portugal, Spain and Sweden.15Latin American countries include Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela.16Reinhart and Rogoff’s (2009) data set is one of the richest data sources for reference of dates of different

types of crises.


currency, banking, and twin crises is appropriate as these crises have occurred most fre-quently in the financially liberalised countries (Hutchinson and Glick, 2000). Most citedexamples are Thailand, Indonesia, Korea, Chile, Finland, Mexico, Norway that facedthese two types of crises after introducing financial liberalisation reforms (Kaminsky andReinhart, 1999, Hutchinson and Glick, 2000).

3.4.2 The Main Explanatory Variable of Interest: Distance

The main variable of interest in this study is our distance measure. This distancevariable measures the deviation of level of financial reforms of the sample countries fromour benchmark country’s level of financial liberalisation17. As mentioned before, the dis-tance based approach was introduced by Zeleny (1974) and now used by other academicsfor the construction of different indexes. Nathan et al. (2008) used this distance-based ap-proach to construct the HDI. They supported this method as the distance-based approachfrom ideal provided them index that fulfilled the criteria of symmetry, uniformity andsignaling. We construct this variable using Spearman distance as a square of Euclideandistance levels of financial reforms between benchmark country and other countries insample.

Distance = ∑(FLi,t−FLr,t)2

= (DFSi,t−DFSr,t)2 +(KAi,t−KAr,t)

2 +(SMi,t−SMr,t)2

where,FLi,t=Level of liberalisation reform of country i at time t on the scale of 1-3, where, 1shows full repression, 2 shows partial liberalisation and 3 shows full liberalisationFLr,t=Level of liberalisation reforms in benchmark country r at time tDFSi,t=Level of domestic financial sector liberalisation of country i at time t

DFSr,t=Level of domestic financial sector liberalisation of benchmark country r at time t

KAi,t=Level of capital account liberalisation of country i at time t

KAr,t=Level of capital account liberalisation of benchmark country r at time t

SMi,t=Level of stock market liberalisation of country i at time t

SMr,t=Level of stock market liberalisation of benchmark country r at time t

17The criteria for selection of benchmark or reference countries are defined in section 3.3.


We use data from Kaminsky and Schmukler (2008) for the timing and extent of thereforms. Kaminsky and Schmukler’s (2008) data set cover all chronological events relatedto the regulation of the capital account, the domestic banking sector, and the stock market.The regulations of capital account covers policies and reforms on off-shore borrowing bydomestic financial institutions, offshore borrowing by non-financial corporations, multi-ple exchange rate markets, and controls on capital outflows. These aspects cover both theregulations introduced or relaxed on capital inflows or outflows. Offshore borrowing bydomestic and non financial sector cover the capital inflows and the last two aspects coverthe reforms on the capital outflows.

The reforms related to the liberalisation of the domestic financial system coverchronology of events related to the regulations on deposit interest rates, lending interestrates, allocation of credit, and foreign currency deposits. For the third component of dataset stock market liberalisation Kaminsky and Schmukler’s (2008) data set covers regula-tions on the acquisition of shares in the domestic stock market by foreigners, repatriationof capital, and repatriation of interest and dividends. The Kaminsky and Schmukler’s(2008) liberalisation index ranges from 1 to 3. On the given 1 to 3 scale 1 means countryis fully liberalised, 2 means the country is partially liberalised and 3 means not liber-alised (repressed). For the ease of analysis we recode the financial index of Kaminskyand Schmukler (2008) in reverse order.

Table B.5 in the Appendix, adapted from Kaminsky and Schmukler (2008), iden-tifying three regimes “fully liberalised”, “partially liberalised”, and “repressed” explainthe criteria for the level of full, partial and no liberalisation of the aspects of domes-tic financial reforms, capital account reforms and stock market reforms. Kaminsky andSchmukler’s (2008) data set provides comprehensive account of the extent of financialliberalisation in sample countries over the period of 1973 to 2005. The full, partial, andno liberalisation of capital account takes into account the restrictions on extent of bor-rowing by banks and corporations abroad along with reserve requirements and existenceof multiple exchange rates. Similarly, the criteria defined for domestic financial sectorexplain restrictions on borrowing and lending interest rates along with credit controls andallowing deposit in foreign currency. Finally, the stock market criteria take into accountthe scale of domestic equity that foreign investors can hold, permission of repatriation ofcapital and dividends, and the percentage of companies’ outstanding equity that foreigninvestors can hold. These defined criteria take good account of changes that over the yearscountries have brought in their financial system targeting financial liberalisation.


3.4.3 Control Variables

We choose control variables following the literature on determinants of banking cri-sis, currency crisis and twin crises 18. The list of variables, their definition and sources arein the Appendix (refer table B.4). Our controls include variables related to the macroe-conomic conditions, the cost of borrowing, measure of currency overvaluation, level offinancial development, and position of net foreign assets as ratio of GDP. In the follow-ing we explain definitions and rationale of our covariates which we use for our modelspecifications.

• We use real interest rate to measure how cost of borrowing affects the balance sheetof financial institutions. The real interest rate is defined as the difference betweennominal interest rate and contemporaneous rate of inflation. This measure allowsinvestigation about how the change in nominal interest rate and price level follow-ing the liberalisation of the financial market affect the saving and lending behaviour.We expect positive sign for the coefficient of real interest rate following Demirgüç-Kunt and Detragiache (1998a) for the likelihood of financial crises, especially, forthe emergence of problems in the banking sector. Demirgüç-Kunt and Detragiache(1998a) show that higher level of real interest rates are associated with higher sys-temic problems in banking sector, as the rise in real interest rate increases the costof borrowing for the banks, which reduces profitability of the banks, and increaseslikelihood of bank runs (Demirgüç-Kunt and Detragiache, 1998a).

• We also use a measure of currency overvaluation as a control variable. FollowingCartapanis et al. (2002) and Ranciere et al. (2006) we calculate currency overvalu-ation as deviation of REER from a Hodrick-Prescott (HP) trend. The deviations forREER are calculated using HP filter. We expect positive value for the coefficient forthe real exchange rate overvaluation19. Kaminsky et al. (1998) show real exchangerate overvaluation as a highly significant indicator for currency crises.

• Private credit to GDP ratio to measures financial depth. The financial crises areusually followed by rise in credit levels in an economy (Laeven and Valencia, 2008).We include this control to see the impact of private credit on crises and especiallyon banking crisis.

18For instance, see Demirgüç-Kunt and Detragiache (1998a), Kaminsky (2006)19Ranciere et al. (2006) argue that overvalued currency has positive effect on crises.


• Net foreign assets to GDP ratio (NFA/GDP) measures availability of net foreignassets as a ratio of GDP. Consideration of position of net liabilities of countries asa trigger for crisis is in debate since the emergence of eurozone debt crisis20. Fol-lowing Lane and Milesi-Ferretti (2007) we expect negative sign for the coefficientof NFA/GDP.

• Government size proxied as government final consumption as percentage of GDP.The higher amount of government expenditure according to IMF increases the in-cidence of economic crisis (Blecker, 1999). The idea of reducing government ex-penditures to deal with crises got mixed support from academics. Stiglitz (2000a)favoured the policy of reducing government expenditure by IMF in Latin Americaafter the crises in 1980s. However he opposed the same remedy for East Asiancountries as they were already running large surpluses. He argued that reducing thesize of the economy would further aggravates the economic crisis by shrinking thesize of economy. We include government expenditures to see whether there is anyeffect of the government expenditure in the economy on likelihood of crises.

• GDP per capita growth rate as a measure of economic growth provides capturesadverse macroeconomic shocks (Demirgüç-Kunt and Detragiache, 1998b). Barro(2001) attributes fall in economic growth as a major cause of crisis in east Asiancountries in 1997-98. In line with Barro (2001) we expect negative impact of higherlevel of economic growth on crises. Hnatkovska and Loayza (2004) studying thecross country relationship also conclude as negative correlation between volatilityof macroeconomic variables and economic growth.

• We also use M2 to Total Reserves as control21. M2 to foreign reserves ratio’s gap inthe literature is used as an indicator for the financial vulnerability (see for instanceKaminsky (2006), Calvo and Mendoza (1996), Qin and Luo (2014)). FollowingGlick and Hutchison (2005) we expect positive significant coefficient for currencycrisis. Calvo and Mendoza (1996) also found high and unstable M2 to foreignreserve ratio responsible for the Mexican and Argentinean crises.

20Luis et al. (2013) show that Greece, Ireland and Portugal whose NFA/GDP ratios were between -70%for Ireland and -95% in Portugal at the onset of crisis.

21For detailed theoretical and empirical discussion about M2/ reserve see Galindo and Maloney (2002)


3.5 Methodology

3.5.1 Main Methodology

The most popular econometric technique to estimate the probability of crises is theprobit(or logit). Under this type of specification crisis is defined as a dependent variableassuming a value of one or zero based on whether the crisis does or does not occur. Specif-ically, the probit(or logit) model provides probability of the variables that are discrete innature. To determine the impact of deviation of level of liberalisation of individual coun-tries from the respective values of the bench mark country we estimate probability of ourcrises variable employing probit model as our main methodology. We also use fixed ef-fect logit model suggested by Chamberlain (1979) along with logisitic regression analysisand ivprobit analyses as robustness check. We perform this exercise to test the hypothesisthat whether the distance variable, calculated as the sum of square deviations of level offinancial reform in country i at time t from the level of reform in benchmark country r

at time t, impacts the probability of different crises significantly after controlling othervariables affecting the probability of crisis.

In the following we show derivation for our methodology. Starting with bivariate re-lationship between crisis and financial liberalisation, our model takes the following form.

Crisisi,t = αi +βFLi,t + εi,t (3.1)

Crisisr,t = αr +βFLr,t + εr,t (3.2)

Here, Equation (3.1) shows relationship between our dichotomous variable of cri-sis with the level of financial liberalistion reform for our sample countries i at time t.Whereas, Equation (3.2) shows the relationship between crisis and financial liberalisa-tion reforms for our reference country r at time t. By subtracting Equation (3.2) fromEquation (3.1), we get,

(Crisisi,t−Crisisr,t) = (αi−αr)+β (FLi,t−FLr,t)+(εi,t− εr,t)

We assume Crisisr,t takes value zero for all periods for our ideal reference country r,therefore we get,

Crisisi,t = α +β (FLi,t−FLr,t)+µi,t (3.3)


Equation (3.3) shows the final form of the specification. In our specification weare interested in the impact of differentials between the financial reforms in our samplecountries and reference country. By incorporating the other control variables, our modeltakes the following form:

Crisisi,t = α +β1(Distance)i,t +β2Currency Overvaluationi,t +β3Private Credit/GDPi,t

+β4NFA/GDPi,t +β5GDP/Capitai,t +β6Government sizei,t +β7RIRi,t

+β8M2/Total Reservesi,t +µi,t(3.4)

Let i = 1,2, ....,n denote countries and t = 1,2, ....,T denotes observations for theith country. Let Crisisi,t be the dependent variable showing banking crisis, currency crisisor twin crises dummy. The value 1 for the banking crisis shows that countries experiencethe banking crisis and 0 if the countries do not face the crisis. Similarly value 1 for thecurrency crisis shows that the countries face currency crisis and 0 when countries do notface the currency crisis. The twin crises dummy variable takes value one when the countryexperiences both of the crises in the same year and zero otherwise.

In our specification the probability of crisis at a particular time period is a functionof distance variable. Since we use the probit functional form, our coefficients do notexplain change in the dependent variable for one unit change in the independent variables.The crisis variable on the left side of specification measures a unit change in cumulativenormal probability for a unit change in x variables. We, therefore, report marginal effectsfor the probit model estimation for the results of our main specifications.

3.5.2 Alternative Specification

Along with our main specification using distance variable as our main explanatoryvariable, we also employ an alternative specification for our analysis. For the alternativespecification we assume that crisis in a country i at time t is a function of financial liber-alisation and the sequence it follows. We use sum of indexes showing level of openness,dummy for full liberalisation and dummy for partial liberalisation to measure financialliberalisation. To take into account sequencing of financial liberalisation, we introducedomestic market first (DMF) dummy along with stock markets first (SMF) dummy. DMFtakes value ‘1’ when a country belongs to the group that liberalised its domestic marketfrist and ‘0’ otherwise. Similarly SMF takes value ‘1’ when country belongs to the group


that liberalised its stock market frist and ‘0’ other wise.

Equation (3.5) in the following shows the alternate specification.

Crisisi,t = α +β1(FL)i,t +β2(FLi,t ∗DMFi)+β3(FLi,t ∗SMFi)+β4Xi,t +νi,t (3.5)

where,FL=Cumulative liberalisation reform index taking sum of different ranking of financialliberalisation reform in domestic market, stock market, and capital accounts in countriesin a particular year/Dummy for full or partial liberalisation taking value 1 if country isfully or partially liberalised and 0 other wiseDMF=Dummy showing 1 for the countries that liberalised their domestic markets firstand 0 otherwiseSMF=Dummy showing 1 for the countries that liberalised their stock markets first and 0otherwise

This specification shows that the probability of crisis is associated with the level ofliberalisation and its sequencing of the opening of its financial sectors. The interactionterms show the effect of change in probability of crisis with the changed sequence ofliberalisation. For this specification we divide our whole sample into three groups. Thefirst group contains the countries that followed domestic market first strategy, the sec-ond group comprising the countries that liberalised stock markets first and the last groupcontains the group of countries that liberalised their capital account first22. In our specifi-cation we use dummies for the categories of domestic sector first liberalisation group andstock market first liberalisation group. For a given base category of capital account firstliberalisation we expect negative sign for the interaction terms between financial liberali-sation variable and domestic sector first dummy and interaction of financial liberalisationwith stock market first dummy. The negative sign shows following a particular strategyof opening domestic sector or stock market first reduces the probability of crisis with thehigher level of liberalisation.

22Ideally, we should introduce further groupings into our analysis based on the permutation betweendomestic market, stock market and capital account. Due to small sample, higher number of groups reducethe size of individual samples considerably that might affect the estimation results.Therefore, for currentanalysis we divide our sample in 3 major groups.


3.6 Results

3.6.1 Main Analysis

We start our analysis by looking into the impact of financial sector liberalisation onour crisis variables before going into the formal analysis of order/sequencing of financialliberalisation on the likelihood of crisis.

Financial Liberalisation and Crisis

Table 3.3 shows marginal effects for the relationship between financial liberalisationand likelihood of banking, currency and twin crises using probit model. Our dependentvariable is the dummy for banking, currency and twin crises. For a particular type of crisisthe dummy variable take value ‘1’ if a country in a particular year experiences the crisisand ‘0’ otherwise23. For liberalisation we use dummy variable that takes value ‘1’ if therespective country is liberalised for a period of 6 or more months during a year, otherwiseit takes value ‘0’.

In table 3.3 columns 1 and 2 show results for banking crisis, columns 3 and 4 forcurrency crisis and columns 5 and 6 for twin crises. Furthermore, in columns 1, 3, and5 we do not control for time and country fixed effects, whereas, in columns 2, 4, and 6we incorporate both country and time fixed effects. All specification include full set ofcontrol variables. The specification with the full set of country and time fixed effects andcontrols (columns 2, 4, and 6) is our preferred specification.

In table 3.3 for our preferred specifications (columns 2, 4, and 6) we observe thatthe probabilities of banking crisis, currency crisis and twin crises fall with the liberal-isation of financial sector. These results are in line with the results of Kaminsky andSchmukler (2003) who argue that in the long-run markets attain stability following theliberalisation policies. Noy (2004) also conclude that the positive relationship betweenliberalisation of financial markets and banking crisis is at most a medium run threat. Thecoefficient of liberalisation for banking crisis and twin crises show 12.7% and 9.9% fallin the probability of banking and twin crises respectively and are significant at 10% levelof significance. The coefficient for the currency crisis for our preferred specification in

23We use criteria for full liberalisation on the basis of Kaminsky and Schmukler’s (2008) criteria for fullliberalisation. They take value ‘1’ for full liberalisation when domestic markets, stock markets, and capitalaccounts are fully liberalised in a particular country and ‘0’ otherwise. The criteria of liberalisation forindividual sectors are explained in the Appendix.


Table 3.3: Marginal Effects using Probit for Liberalisation Dummy

Banking Crisis Currency Crisis Twin Crises(1) (2) (3) (4) (5) (6)

Liberalisation 0.010 -0.127* -0.088** -0.051 -0.029 -0.099*0.038 0.077 0.036 0.054 0.019 0.059

Currency overvaluation -0.000 0.002** 0.001* 0.001* 0.000 0.0010.001 0.001 0.001 0.001 0.000 0.001

Private Credit/GDP 0.002*** 0.004*** -0.003*** -0.001 -0.000 -0.0000.000 0.001 0.001 0.001 0.000 0.001

NFA/GDP -0.102*** -0.130* -0.058 -0.129*** -0.025 -0.0200.033 0.067 0.040 0.043 0.017 0.035

GDP/Capita -0.028*** -0.031*** -0.022*** -0.017*** -0.009*** -0.016***0.005 0.006 0.004 0.004 0.002 0.004

Government size -0.011*** 0.047*** -0.013*** -0.010 -0.004** 0.0090.003 0.011 0.003 0.008 0.002 0.006

Real interest rate 0.000* 0.000* 0.000* 0.000 0.000** 0.000**0.000 0.000 0.000 0.000 0.000 0.000

M2/Total reserves -0.005** -0.003 0.002*** 0.005*** -0.002* 0.0010.002 0.002 0.001 0.002 0.001 0.002

Pseudo R-square 0.115 0.280 0.214 0.399 0.208 0.315χ2 81.738 486.604 101.728 914.040 90.406 412.589Observations 704 574 704 628 704 328Year Dummy No Yes No Yes No YesCountry Dummy No Yes No Yes No Yes

The table reports regression coefficients and robust standard error. * p<0.10, ** p<0.05, *** p<0.01.Columns 1-2 report the marginal estimates of the standard probit model for banking crisis as a dependentvariable, which take value 1 if country faces banking crisis and 0 otherwise. Columns 3-4 show marginalestimates for currency crisis that takes value 1 when a particular country faces currency crisis in a particularyear and 0 otherwise. Similarly columns 5-6 present results for twin crises that take value 1 when countryfaces both banking and currency together in a particular year and 0 otherwise. Liberalisation is a dummyvariable that takes value 1 if the country in a particular year either fully or partially liberalised for six ormore than six months and zero otherwise.

column 4 shows that the probability of crisis falls by 5.1% but is not significant. Thestated negative relationship between financial liberalisation and likelihood of crises showthat financial liberalisation itself is not a threat to the stability of the financial sector or theoverall economy.

Most of the control variables in our preferred specification in columns 2, 4, and 6are significant with expected signs. For our preferred specifications for banking crisis, asexpected the measure of currency overvaluation shows positive sign and it is significantat 5%. For the currency crisis it is again positive and significant at 10%. The sign for thetwin crisis is also positive but it is statistically insignificant. The positive and significantsign of currency overvaluation for banking and currency crises show that crises are clearlylinked with the levels of currency overvaluation.

The coefficient for private credit/ GDP as a measure of financial depth is positive and


significant, which as expected shows that the higher value of private credit as a percent-age of GDP increases the likelihood of banking crisis (as main measure of crisis). Ourresults for private credit/GDP are in line with Laeven and Valencia (2008) who show thepositive effect of accumulation of private credit on emergence of crises. However, we getinsignificant coefficient of private credit/GDP for currency crisis and twin crises.

As mentioned earlier the accumulation of net foreign assets as a ratio of GDP im-prove resilience against the emergence of crises. We get negative sign for NFA/GDP forbanking, currency and twin crises. However, it is significant for banking crisis only at10% (see column 2 of table 3.3). Our measure of growth also shows expected sign withhigh level of significance. The coefficients for economic growth variable are negative andhighly statistically significant for all specifications, showing a well established negativerelationship between growth and crises.

The role of government expenditures in emergence of crises depends on the countryspecific factors24 . The coefficient for the government expenditure as percentage of GDPas a measure of size of the government shows mixed sign. For the banking crisis it changesboth significance and sign when we run our model with and without country and time fixedeffects. For our preferred specification for banking crisis it, interestingly, shows positivesignificant coefficient that shows that higher level of government expenditures contributepositively to the banking crisis. The coefficients of government expenditures for currencyand twin crises, however, are insignificant.

The coefficients of real interest rate is positive and significant for all types of crises.Higher level of interest rate increase costs for the banking sector that causes low prof-itability and may cause severe problems for the financial sector and overall economy(Demirgüç-Kunt and Detragiache, 1998a).

M2 to total reserve ratio shows positive significant coefficient with positive sign forcurrency crisis. Whereas, the coefficient for banking crisis in negative but insignificant.For twin crisis the coefficient is positive but again insignificant. Calvo (1996) show themixed results for this ratio with examples of Austria where M2/ total reserves was quitelarge but was stable as compare to Mexico and Argentina where it caused high instabilitiesin the economy.

24Stiglitz (2000a) supported the reduced government expenditure recipe of IMF for Latin America butopposed it for East Asian economies. He based his argument on the levels of deficit/surpluses in the respec-tive economies for success of this policy.


Optimal Sequencing: Benchmark Chile

Following the analysis between financial liberalisation and financial crises, we fur-ther our analysis to investigate if there is negative association between financial liberali-sation and crises, what lead to the crises in liberalised economies. By employing probitmodel, table 3.4 displays the marginal effects for the baseline specification for our mainbenchmark country Chile25. In our specification we use distance as main explanatoryvariable. As shown in subsection 3.4.2 the variable distance provides a measure of devia-tion between ranking of financial reforms in our reference country and other countries insample. Since we take Chile as our initial benchmark country for this specification, wecalculate distance between rankings of our sample countries i at time t from the rankingof financial reforms in Chile (r) at time t.

Columns 1, 3 and 5 in table 3.4 show co-efficient of our main specifications withoutcountry and year fixed effects, whereas, in the column 2, 4, and 6 we add country and yearfixed effects. The specification with the full set of country and year fixed effects (column2, 4, and 6) is our preferred specification. We further mainly focus on the emergence ofbanking crisis out of all the three crises.

The estimates in column 2 show positive and significant coefficient for the deviationsbetween the level of financial reforms in Chile and other countries in our sample. Thedistance away from Chile also shows consistently positive sign also for currency and twincrises, however, the distance variable is not significant for currency crisis. The coefficientof distance for banking crisis is significant even at 1% that shows significant rise in theprobability of banking crises in sample countries if they differ in their ordering from theordering that Chile has followed. The column 2 shows approximately 4% higher chancesof banking crisis for sample countries if their sequencing differs from Chile. The resultsconfirm that the order of liberalisation of financial sector by Chile has unique importancefor stability of the financial sector in general and banking sector in particular. As shownfor our main analysis, most of the other explanatory variables for this specifications showexpected sign at conventional level of significance.

Searching for Optimal Sequencing: Benchmarking all Countries, One-by-One

Since this study does not confine itself to the analysis of a particular country thathas followed a particular ordering of financial liberalisation, we in our further analysis

25As we have shown that Chile due to its unique sequencing of financial liberalisation and its performanceduring many regional and global crises is the most suitable country as an initial benchmark.


Table 3.4: Probit Model for Chile using Sum of Square Deviations


Distance (Chile) 0.018*** 0.036*** 0.002 0.003 0.002 0.022***0.006 0.011 0.006 0.008 0.003 0.006

Currency overvaluation -0.000 0.001 0.001** 0.001 0.000 0.0000.001 0.001 0.001 0.001 0.000 0.001

Private Credit/GDP 0.003*** 0.004*** -0.003*** -0.001 -0.000 0.0000.000 0.001 0.001 0.001 0.000 0.001

NFA/GDP -0.113*** -0.128* -0.054 -0.130*** -0.026 -0.0200.033 0.067 0.040 0.043 0.017 0.032

GDP/Capita -0.028*** -0.032*** -0.022*** -0.017*** -0.009*** -0.015***0.005 0.006 0.004 0.004 0.002 0.003

Government size -0.011*** 0.041*** -0.015*** -0.011 -0.004** 0.0060.003 0.010 0.003 0.007 0.002 0.005

Real interest rate 0.000** 0.000** 0.000 0.000 0.000* 0.000**0.000 0.000 0.000 0.000 0.000 0.000

M2/ Total reserve -0.005** -0.002 0.002*** 0.005*** -0.002* 0.0030.002 0.003 0.001 0.002 0.001 0.002

Pseudo R-square 0.126 0.291 0.205 0.397 0.203 0.337χ2 90.333 686.736 102.471 918.345 90.743 432.346Observations 704 574 704 628 704 328Year Dummy No Yes No Yes No YesCountry Dummy No Yes No Yes No Yes

The table reports regression coefficients along with robust standard errors. * p<0.10, ** p<0.05, ***p<0.01. Columns 1-2 report the marginal estimates of the standard probit model for banking crisis as adependent variable, which take value 1 if country faces banking crisis and 0 otherwise. Columns 3-4 showmarginal estimates for currency crisis that takes value 1 when a particular country faces currency crisis ina particular year and 0 otherwise. Similarly columns 5-6 present results for twin crises that take value 1when country faces both banking and currency together in a particular year and 0 otherwise. Our mainexplanatory variable distance Chile measures sum of square of Euclidean difference between the level ofliberalisation in country i at time t and Chile (r) at time t.

generalise the choice of benchmark country to all countries in our sample. Table 3.5 pro-vides marginal estimates for our distance variable for each of the country in our sampleas a benchmark using probit model. To determine if a particular sequence that a countryfollows has special bearings on the likelihood of crises, we group our results on the ba-sis of the ordering of financial reforms in respective benchmark countries. We combineresults on the basis of the order of liberalisation between domestic markets, stock marketand capital accounts.

In table 3.5 columns 1, 3, and 5 show the estimates for the regressions withoutcountry and year dummies, whereas, columns 2, 4 and 6 show estimates incorporatingtime and country dummies. All specifications use full set of controls. We start our analysiswith the first group that liberalised domestic sector first, then stock markets and in the end


capital accounts26. This group includes: our initial benchmark country Chile along withBrazil, Colombia, Ireland, and Philippines. We observe that the coefficients for distancefor this group are consistently positive and significant. The consistently positive andsignificant results for this group shows that an order that places liberalisation of domesticmarket first, stock market second and capital accounts make countries resilient to crisesand if a country follows different ordering, the probability of crisis for that country risesconsiderably.

The estimates show mixed sign and significance for the second group in table 3.5,which comprises countries that liberalised their domestic sector first, then capital accountand finally stock markets. The estimates for distance variable from Korea and Taiwanshow the positive significant estimates, whereas, other countries’ estimates show insignif-icant results. The third group follows sequence in order of capital account, stock marketsand domestic market. For the banking crisis except Japan and Venezuela other countriesshow no significant estimates. Deviation from Mexico–that opened its capital accountfirst– does not increase probability of banking crisis for the other countries in sample.The fifth group comprising Canada, Denmark, France, Germany, Italy, Portugal, Sweden,and Hong Kong broadly followed order of first stock markets, second domestic marketsand in the end capital account. For banking crisis the deviation coefficient for some ofthe countries in this group show the positive significant result. The last group compris-ing Malaysia, Spain, Thailand, UK and USA have opened their stock market first, capitalaccount second and domestic market in the end. In this group, for our preferred specifica-tion for banking crisis only deviation from Spain shows the positive significant estimates,whereas for others countries the results show insignificant results.

The analysis of all countries’ result show that deviation from those countries whichdelayed the opening of their capital account increases likelihood of banking crisis signif-icantly. The consistent positive significant estimates for the distance from the countriesthat did not expose their countries capital flows show that the ordering of financial reformshas significant impact on stability of the banking sector and overall financial sector.

26Theoretically speaking as mentioned before this group should show higher resilience to the emergenceof crises, as it allows opening up of the domestic market and stock market before capital account liberali-sation, which provides ample opportunity to domestic financial sector to develop before its exposure to theforeign flows.


Table 3.5: Probit Marginal Effects for All Countries With and Without Country Dum-mies


Sequence Domestic Market Stock Markets Capital AccountDistance (Brazil) 0.019*** 0.014* 0.003 0.005 0.004 0.010**

0.007 0.008 0.007 0.009 0.003 0.004Distance (Colombia) 0.020*** 0.027*** -0.002 -0.010 0.001 0.012**

0.006 0.009 0.006 0.010 0.003 0.005Distance (Ireland) 0.031*** 0.031*** 0.019*** 0.020** 0.009*** 0.020***

0.008 0.011 0.007 0.008 0.003 0.006Distance (Philippines) 0.030*** 0.037*** 0.017*** 0.016* 0.005 0.026***

0.007 0.012 0.006 0.009 0.003 0.008Sequence Domestic Market Stock Markets Capital AccountDistance (Argentina) -0.005 -0.005 -0.003 -0.002 -0.004 -0.001

0.006 0.007 0.005 0.005 0.002 0.005Distance (Indonesia) 0.015** 0.003 0.009 0.008 0.004 0.015*

0.008 0.010 0.007 0.008 0.003 0.008Distance (Korea) 0.021*** 0.023*** 0.004 0.005 0.002 0.010*

0.005 0.008 0.005 0.007 0.002 0.005Distance (Norway) 0.005 -0.001 0.008 0.011* 0.004* 0.013***

0.006 0.008 0.005 0.006 0.002 0.005Distance (Taiwan) 0.020*** 0.024*** 0.009* 0.012 0.005** 0.019***

0.006 0.009 0.005 0.008 0.002 0.006Sequence Domestic Market Stock Markets Capital AccountDistance (Finland) 0.015** 0.009 0.011* 0.013* 0.005** 0.011**

0.007 0.009 0.006 0.007 0.002 0.005Distance (Japan) 0.017*** 0.031*** 0.013** 0.012* 0.005** 0.012**

0.006 0.011 0.005 0.007 0.002 0.006Distance (Peru) 0.013** 0.010 -0.003 -0.002 0.000 0.007

0.005 0.008 0.005 0.007 0.002 0.005Distance (Venezuela) -0.006 0.014* 0.004 0.005 0.001 0.008

0.004 0.008 0.004 0.006 0.002 0.005Sequence Domestic Market Stock Markets Capital AccountDistance (Mexico) 0.028 0.038 -0.010 -0.009 -0.008 0.064

0.020 0.027 0.020 0.032 0.026 0.043Sequence Domestic Market Stock Markets Capital AccountDistance (Canada) 0.003 0.012 0.009*** 0.009 0.004** 0.013**

0.005 0.008 0.004 0.006 0.002 0.005Distance (Denmark) 0.016** 0.018* 0.013** 0.011 0.005** 0.012*

0.006 0.011 0.005 0.007 0.003 0.006Distance (France) 0.017*** 0.020** 0.016*** 0.016** 0.007*** 0.014***

0.006 0.009 0.005 0.006 0.002 0.005Distance (Germany) 0.007 0.012 0.010** 0.010 0.005** 0.013**

0.005 0.009 0.004 0.007 0.002 0.005Distance (Italy) 0.017*** 0.031*** 0.013** 0.012* 0.005** 0.012**

0.006 0.011 0.005 0.007 0.002 0.006Distance (Portugal) 0.010 0.006 0.005 0.007 0.002 0.009*

0.006 0.009 0.005 0.007 0.002 0.005Distance (Sweden) 0.025*** 0.021** 0.017*** 0.019*** 0.009*** 0.019***

0.007 0.010 0.006 0.007 0.003 0.006Distance (Hong Kong) 0.015*** 0.010 0.016*** 0.018** 0.006** 0.011*

0.006 0.011 0.005 0.008 0.002 0.007Sequence Domestic Market Stock Markets Capital AccountDistance (Malaysia) 0.007 -0.020 0.002 0.026** 0.003 0.007

0.007 0.014 0.006 0.011 0.003 0.009Distance (Spain) 0.017*** 0.022** 0.017*** 0.016** 0.007*** 0.016***

0.006 0.009 0.004 0.007 0.002 0.006Distance (Thailand) 0.014** 0.007 0.000 0.001 0.000 0.003

0.006 0.008 0.005 0.006 0.002 0.005Distance (UK) 0.013** 0.012 0.017*** 0.015** 0.006*** 0.013**

0.006 0.009 0.004 0.006 0.002 0.005Distance (USA) 0.008 0.009 0.014*** 0.013* 0.005*** 0.009**

0.005 0.007 0.004 0.007 0.002 0.004

The table reports regression coefficients and robust standard errors for only main explanatory variabledistance. * p<0.10, ** p<0.05, *** p<0.01. Columns 1-2 report the marginal estimates of the standardprobit model for banking crisis as a dependent variable, which take value 1 if country faces bankingcrisis and 0 otherwise. Columns 3-4 show marginal estimates for currency crisis that takes value 1 whena particular country faces currency crisis in a particular year and 0 otherwise. Similarly columns 5-6present results for twin crises that take value 1 when country faces both banking and currency togetherin a particular year and 0 otherwise. Our main explanatory variable distance measures sum of squareof Euclidean difference between the level of liberalisation in country i at time t and respective referencecountry (r) at time t.


3.6.2 Robustness

We perform initial robustness analysis using alternative specification for our model,and grouping of the countries according to their strategy of full liberalisation of differentsectors.

Alternative Specification: Table 3.6 shows marginal probit estimates for alternativespecification shown in Equation (3.5). Column 1 shows estimates for the level of financialliberalisation. Columns 2 and 3 show the similar estimates for the dummy variables of fullliberalisation and partial liberalisation27. The interaction terms in these specifications areintroduced to analyse how the effect of liberalisation policies change for the countries thatadopt different ordering of opening of their capital account, stock markets, and domesticmarkets.

In this specification we introduce dummies showing the ordering of domestic marketfirst (DMF) and stock market first (SMF), keeping opening of capital account first (CAF)ordering as a base category. The dummy for DMF takes value ‘1’ if a country belongsto the group that liberalised its domestic market first and ‘0’ otherwise. Similarly SMFdummy takes value ‘1’ if country belongs to the group that liberalised stock market firstand ‘0’ otherwise.

The estimates of the interaction terms between liberalisation and DMF and SMF incolumn 1 show countries that liberalise domestic market first or stock market first face lesslikelihood of banking crisis as compare to the countries that liberalise their capital accountfirst. According to our estimates higher level of liberalisation with DMF policy reducesthe probability of crisis by 4.9%, whereas the fall with SMF is 10.8% compare to the CAFstrategy–both coefficients are significant at 5% and 1% respectively. The interaction termsbetween DMF, SMF and full liberalisation dummy in columns 2 show the probability ofbanking crisis fall by 16.8% and 17.3% respectively when countries belongs follow DMFand SMF policy to liberalise their financial sector. Our estimates with partial liberalisationdummy also show fall in the probability of banking crisis by 22.1% when countries followDMF and fall of probability by 35.3% when countries belong to group that preferredSMF policy of liberalisation. Most of the explanatory variables show expected signs atconventional level of significance.

Grouping of countries on the basis of their policies of reforms: We performfurther robustness analysis by dividing countries into groups according to the ordering

27For the dummies of full liberalisation and partial liberalisation we use criteria defined by Kaminskyand Schmukler (2008). Their criteria for full liberalisation and partial liberalisation dummies depend on theextent of liberalisation in domestic markets, stock markets, and of capital accounts.


Table 3.6: Results for Alternative Specification

Dependent Variable: Banking Crisis(1) (2) (3)

Liberalisation Index 0.0260.022

Full liberalisation (d) 0.0930.089

Partial liberalisation (d) 0.0910.088

Liberalisation*DMF -0.049**0.025

Liberalisation*SMF -0.108***0.030

FL*DMF (d) -0.168***0.056

FL*SMF(d) -0.173**0.073

PL*DMF (d) -0.221***0.064

PL*SMF (d) -0.353***0.092

Currency overvaluation 0.003** 0.003** 0.004***0.001 0.001 0.001

Private Credit/GDP 0.005*** 0.004*** 0.004***0.001 0.001 0.001

NFA/GDP -0.130** -0.101 -0.129**0.065 0.065 0.065

GDP/Capita -0.032*** -0.032*** -0.032***0.006 0.006 0.006

Government size 0.044*** 0.047*** 0.052***0.011 0.011 0.011

Real interest rate 0.000 0.000 0.0000.000 0.000 0.000

M2/Total reserves -0.004 -0.003 -0.0040.002 0.002 0.002

Psuedo R-square 0.294 0.285 0.293χ2 838.594 514.929 695.553Observations 574 574 574Year dummies Yes Yes YesCountry dummies Yes Yes Yes

The table reports marginal regression coefficients and robust standard error of probitmodel. * p<0.10, ** p<0.05, *** p<0.01. The dependent variable (banking crisis) isequal to one if a country faces banking crisis in a particular year and zero otherwise. Col-umn 1 reports the estimates incorporating liberalisation index calculated as the sum of levelof liberalisation for domestic market, stock market and capital accounts on on the scale 1-3,where 1 shows least liberal and 3 shows the most liberal. Column 2 reports the results forthe full liberalisation (FL) that takes value one if country is fully liberalised and zero other-wise. Similarly, column 3 reports results for partial liberalisation (PL) that takes value oneif a particular country is partially liberalised and zero otherwise. DMF is Domestic MarketFirst variable that takes value one when a country belongs to the group that has opened itsdomestic sector first and zero otherwise. SMF is a variable that takes value when countrybelongs to the group that liberalise its stock market first.


they have followed to open the financial sector. For this purpose, we divide our sampleinto three groups: the first group comprising countries that started liberalisation reformsby opening domestic market first, the second group shows the countries that liberalisedstock market first, and the third group with liberalisation of capital account first strategy.

Table 3.7: Marginal Probit Effects for Group of Countries

Dependent Variable: Banking CrisisDomestic Market Stock Market Capital Accounts

First First First(1) (2) (3)

Distance (Chile) 0.045 0.065*** 0.102**0.029 0.017 0.043

Currency overvaluation 0.005* 0.002 0.010*0.003 0.002 0.006

Private Credit/GDP 0.014*** 0.002 0.022***0.005 0.001 0.006

NFA/GDP -0.306** 0.343*** -0.0870.140 0.111 0.119

GDP/Capita -0.020* -0.083*** -0.0180.011 0.018 0.013

Government size 0.056*** 0.083*** 0.097*0.019 0.023 0.052

Real interest rate 0.000 -0.012* -0.001**0.000 0.006 0.000

M2/Total reserves -0.020 0.010** -0.0010.014 0.004 0.008

Pseudo R-square 0.322 0.549 0.538χ2 682.465 568.741 1255.777Observations 186 201 115Year dummies Yes Yes YesCountry dummies Yes Yes Yes

The table reports regression coefficients and robust standard error. * p<0.10, ** p<0.05, *** p<0.01.The dependent variable (banking crisis) is equal to one if a country faces banking crisis in a particularyear and zero otherwise. Our main explanatory variable distance (Chile) measures sum of square ofEuclidean difference between the level of liberalisation in country i at time t and Chile (r) at timet. Column 1 reports results for the countries that belong to the group that liberalised their domesticmarket first. Column 2 shows the results for the countries that belong to the group that liberalised theirstock market first and the last column 3 shows results for the group of countries that liberalised theircapital account first.

Table 3.7 shows marginal probabilities of banking crisis for the three groups of coun-tries using probit model. All specifications include full set of control variables. Weobserve clear evidence that the coefficient of distance away from Chile is significantlydifferent for each group. The probability of banking crises in case of group comprisingcountries that liberalised domestic sector first is 4.5% which is significant just above the10% level of significance. Whereas, the coefficient for the group that liberalised stock


market first is 6.5% which is significant at 1%. The coefficient of distance for the groupcomprising countries that liberalised capital account first is 10.2% which is the highestamong the three groups and significant at 5%. The results show that the probability ofcrises increases gradually if a country follows a different ordering from our benchmarkcountry Chile and belongs to the group that has opened stock market first or capital ac-count first.

For further robustness check we select a group of countries that has followed the-oretically preferred strategy of domestic financial liberalisation first following the stockmarket and capital account liberalisation. For this purpose we select the group of coun-tries that has followed a sequence similar to Chile. That is, broadly speaking the openingof financial sector in order of domestic market, stock market and in the end capital ac-count. This group includes Brazil, Colombia, Ireland, and Philippines along with Chile.As we have seen in table 3.5 that the deviations away from this group create positive sig-nificant effect on the probability of crises on other countries, in the following robustnessanalysis we check if the results are driven by the definition of our main interest variable,functional form or possible reverse causality. Because of our main focus on banking cri-sis, we choose only occurrence of a banking crisis as our main dependent variable. In thefollowing we present our results.

Alternative metric for distance: Columns 1 and 2 in table 3.8 provide estimatesfrom the set of specification for our select group of countries using a different metricfor distance. For this purpose we define the distance variable as the sum of absolutedeviations. The sum of absolute deviations give equal weights as compare to sum ofsquare deviations that give more emphasis on extremes.

To check if weighing the extremes affect the result we use sum of absolute deviationsof our sample countries from our benchmark countries as an alternative form of distancemetric. Column 1 shows estimates for distance using all covariates but no countries oryears dummies. Column 2 provides estimates using both countries and years dummiesalong with all covariates. The distance coefficients for all countries are positive signifi-cant. However, distance from Brazil loses its significance when we include countries andyears dummies. The positive significant results show that the farther is a country awayfrom the order of our group of benchmark countries, the higher are the chances of bankingcrisis.

Fixed Effect Logit Model and Logistic Model: To check if the results are drivenby the functional form we have chosen for our main analysis, we perform analysis using


Table 3.8: Robustness with Different Specifications and Definition of Our Independent Vari-able

Dep Variable: Banking Crisis

Abs Dev FE Logit Odds Ratio IV-Probit

(1) (2) (3) (4) (5) (6) (7) (8)

Distance (Brazil) 0.035*** 0.020 0.148*** 0.095 1.117*** 1.109 0.088*** 0.088*

0.011 0.015 0.049 0.067 0.046 0.077 0.030 0.053

Distance (Chile) 0.033*** 0.049*** 0.204*** 0.166*** 1.112*** 1.246*** 0.082*** 0.200***

0.011 0.016 0.071 0.048 0.043 0.087 0.026 0.066

Distance (Colombia) 0.035*** 0.048*** 0.201*** 0.164** 1.125*** 1.195*** 0.108*** 0.231***

0.011 0.016 0.046 0.070 0.039 0.079 0.025 0.063

Distance (Ireland) 0.048*** 0.038** 0.208*** 0.173** 1.192*** 1.206** 0.129*** 0.135**

0.012 0.019 0.053 0.072 0.054 0.093 0.034 0.060

Distance (Philippines) 0.042*** 0.050** 0.262*** 0.199*** 1.186*** 1.236** 0.145*** 0.227***

0.013 0.022 0.054 0.076 0.050 0.102 0.030 0.064

Year dummy No Yes No Yes No Yes No Yes

Country dummy No Yes No No No Yes No Yes

Other Covriates Yes Yes Yes Yes Yes Yes Yes Yes

The table reports regression coefficients and robust standard errors in brackets. * p<0.10, ** p<0.05, ***p<0.01. The dependent variable (banking crisis) is equal to one if a country faces banking crisis in a particularyear and zero otherwise. Columns 1-2 report the estimates of the standard probit model incorporating distancevariable using sum of absolute deviation. Columns 3-4 showing results for FE logit and columns 5-6 showingthe results for logistic regression for our distance variable calculated as sum of square of deviations. Columns7-8 refer to the Instrumental Variable (IV) estimates to control the endogeneity of distance variable, all IVregressions are estimated using Newey’s two-step estimators (IVPROBIT). Though results are not reported wehave performed Wald χ2 test for over-identifying restrictions with the null hypothesis of validity of instrument.All our regression pass the wald test for exogeneity.

fixed effect logit model and logistic model. Columns 3, 4, 5, and 6 of table 3.8 show es-timates for fixed effect logit model and logistic model at the place of porbit estimates forthe selected group of countries. The results we obtain are very similar to our benchmarkregression results showing positive and significant coefficients for distance away from ourbenchmark countries in this group. With the exception of Brazil for which the distancevariable loses significance again when year dummies are introduced in fixed effect logitmodel and country and year fixed effects are introduced in logistic model, all other coeffi-cient are consistently positive and significant. The results yet again show that the orderingthat places the capital account in the end and allows liberalisation of domestic market andstock market before that is potentially a strategy that increases resilience to the problemsin the financial sector.


The issue of endogeneity: A potential problem of endogeneity may affect our previ-ous results as there is always likelihood that countries change their policies of liberalisa-tion following the emergence of crises. Although as table B.1 in the Appendix shows veryfew occurrences of the events of reversals in financial reforms in our sample country, westill check for the problem of endogeneity. To deal with the possible simultaneity prob-lem between the measure of financial reform and crisis, we provide results using two-stepprobit model with endogenous regressors in columns 7 and 8 of table 3.8.

We use lag value of our main explanatory variable as instrument following Buchet al. (2013, p.15). The lagged value of the explanatory is used widely in literature to dealwith the problem of simultaneity between two variables (for instance, see Clemens et al.(2012), Green et al. (2005)). The results are robust and similar to the simple probit modeland other specifications we have used. The distance variable reported here show positiveand significant coefficient for all countries in the group that have followed sequencing ofdomestic market first, stock market second and capital account liberalisation last.

We use Wald test of exogeneity with the null hypothesis of validity of instruments.The results for the Wald test are not reported in table 3.8. According to our results wecannot reject our null hypothesis of validity of instruments, which shows that our measureof sequencing of reforms is not endogenous.

3.7 Conclusion

The occurrence of frequent financial crises leading to large scale economic catastro-phes in different parts of has always attracted a great amount of attention from academiaand circles of policy makers to understand connections between uncertainties of globalfinancial system and overall economic performance. Different strands of literature in fi-nance deal with this investigation differently and reach to different conclusions. Alongwith many other observations, a wide perspective on this issue is that the real problem liesin the manner countries manage their financial sectors in terms of the policies of reforms.This chapter by addressing the issue whether different sequencing of internal financialreforms affect the probability of financial crises searches for an optimal sequence that re-duces the probability of crises using probit model for the sample of 28 countries over theperiod of 1973-2005.

We focused on the episodes related to the liberalisation of domestic financial mar-kets, stock markets and capital account. Financial crises are measured as the onset of


banking crisis, currency crisis and twin crises. We were particularly interested in analysinghow different orderings between domestic market, stock market and capital account-defined here as sequencing of reforms- affect the probability of banking crisis.

By introducing a novel quantitative measure of sequencing and analysing interac-tions between liberalisation of domestic market, stock market and capital account weconclude that the sequence that a country follows to open its financial sector has very im-portant bearings on the stability of the economy. McKinnon (1973), Edwards (1984), andStiglitz (2002) also reached to the similar conclusions showing that a particular sequencethat a country follows to liberalise its financial sector has very important bearings on itsmacroeconomic performance.

Following the consensus in literature that a gradual approach that gives ample oppor-tunity to develop internal financial system before removing controls on capital flows makecountries resilient to financial crises, we developed a distance measure from the coun-tries that followed more gradual approach and liberalised capital account at a later stage.Due to the large literature favouring Chile’s financial reforms as prudent and gradual, westarted our analysis by selecting Chile as an indicative benchmark country showing opti-mal reform sequencing. Later we generalised the benchmark to observe the pattern in ourfull sample by taking all countries as our benchmark. According to our results there isevidence that suggest that a capital account first strategy increases the likelihood of crises.The group that came out the most resilient to the crises and especially to the banking crisisfrom the whole sample is the group of countries that liberalised their capital account last.The domestic market and stock market both, however, come out as a preferred strategyover capital account as first strategy to liberalise the financial sector.

We find that the likelihood of crises increases when countries follow more abruptapproach that allows free capital movements in the start of liberalisation process. Weconclude that likelihood of crises is associated with the order of liberalisation. However,we do not find any single prescription for optimal ordering/sequencing of financial sectorreforms.

An important lesson of this work is the need of cautious attitude towards liberalisa-tion of the financial sector, especially, for capital accounts liberalisation. Our results arein conformity of theories that suggest that if a country by any means or reason decides togo for a gradual approach to reform, it should keep capital account liberalisation in end ofits reform agenda. The results we obtain are robust to different definition of distancing,specification, econometric modelling and disaggregation of sample in groups.

Chapter 4

Thesis Conclusion

This thesis deals with the effect of financial sector liberalization on finance-growthrelationship and the likely impact of its sequencing on the vulnerability of an economy tofinancial crisis. In what follows, we explain the contributions we made in this literatureand, subsequently, suggest possible areas of further research.

Until recently, the existing literature on finance and growth has been quite clear aboutthe positive significant role of financial development on economic growth (Beck et al.,2000). However, recent financial crisis since 2007 has shaken this understanding as thecountries that are financially more developed were hit the hardest by the crisis. Followingthis, the recent literature has shown that countries at the heart of the financial crisis mayhave financial systems that are “too large" and these exist not necessarily because ofgood policies and institutions, rather because of poor regulatory systems or excessivefinancial liberalization (see Arcand et al. (2012)). This suggests that excessive financialdevelopment is an amplifying factor behind the financial crisis and negative growth (seeArcand et al. (2012), Ductor and Grechyna (2011)).

Recognising the above, the literature related to the issues of the impact of financialliberalisation on the finance-growth relationship, and the effect of sequencing of financialreforms on the sustainability of is growing. However, it provides expositions using basiceconometric analysis and anecdotal evidences (Demetriades and Rousseau, 2011, Arestis,2006). Similarly, the literature related to the issues of sequencing is mainly qualitative inthat it analyses the impact of sequencing on different macroeconomic variables withoutusing any quantitative measure of sequencing (Edwards, 2009, 1990).

To deal with the highlighted shortcomings in the literature this thesis deals withthe issues of the impact of financial liberalisation on finance-growth relationship using a

100

Chapter 4. Thesis Conclusion 101

more systematic analysis. Looking back at the previous literature, the broader conclusionsrelated to the interaction between these variables can be classified as follows. The firststrand of literature that studies the relationship between financial development and growthhas wide consensus that financial development affects growth positively (see for instance,Rajan and Zingales (1996), Beck et al. (2000)). The second strand investigating the rela-tionship between financial reform and growth shows mixed results. For instance, Levine(2001) believes the answer to the question as to whether international financial liberalisa-tion improves growth is “yes”. Whereas, Krugman et al. (1993) shows that internationalfinancial liberalisation does not spur growth. The third strand deals with the issue of fi-nancial reforms and financial crisis. This literature argues the involvement of higher levelof reforms to more systemic instability in the financial system that results in a negativeimpact on economic growth (see for instance Caprio et al. (2005)) . Our contribution inthis thesis is twofold. First, we analyse systematically the role of financial liberalisationin finance-growth relationship. Using data for 88 countries over the period of 1973 to2005, we find that excessive liberalization in the financial sector makes it less effective togeneral economic growth. Second, we investigate optimal sequence of reforms that maymake the countries less vulnerable to financial crisis. In doing this, we seek to providea measure of sequence of financial reform by constructing a distance measure from thecountries that followed more gradual approach and liberalised their capital account at alater phase. Our analysis shows that the experience of the countries that delayed or fol-lowed very gradual approach to liberalise their capital accounts have been less vulnarableto financial that those countries that allowed for shock approach or liberalised their capitalaccount before bringing reforms in other sectors. To the best of our knowledge, this is thefirst study that suggests a quanifiable measure, though an indirect one, for sequence offinancial reform in the literature.

The broader conclusions drawn in this thesis are in line with many other studies1.We find very strong support for the role of finance in economic growth that is dependenton policies of liberalisation. In our systematic analysis we reach the conclusion that thereis a limit of liberalization beyond which financial development looses its effectiveness ingenerating economic growth. Excessive or poorly implemented and managed liberalisa-tion policies amplify the risks of reckless behaviour of the financial institutions regarding

1For instance, in line with Rajan and Zingales (1996), Beck et al. (2000) we find positive relationshipbetween financial development and economic growth. We also find that the relationship is weakeningbetween financial development and economic growth, as concluded by Rousseau and Wachtel (2011). LikeEdwards (2009) and others we also come to the conclusion that the sequencing in which financial reformsare introduced has very large impact on overall macroeconomic performance and, especially, on increasingvulnerability of the financial system and likewise the whole economy


the allocation of funds and future decision making. This affects the economy adverselyand diminishes the positive relationship between finance and economic growth throughits negative impact on the stability of the financial system.

The other conclusion in this thesis relates to the issue of the sequencing of financialreforms. We find that a wrongly ordered sequence of reforms increases vulnerabilitiesof the financial system and overall economy to financial crisis significantly. We broadlyconclude that maintaining controls on capital inflows and outflows, when the domes-tic financial system is in its nascent stage, make countries more resilient to the inherentinstabilities of the financial system. Therefore, correct sequencing has bearing on an in-dividual country’s resilience to financial crises. Liberalisation of the capital account tooearly in the stage of overall financial reforms, especially when domestic and stock mar-kets are repressed, increases the likelihood of crises. In the following we provide detailedconclusions of our chapters and provide policy guidelines.

4.1 Main Findings and Policy Issues

A long held belief among the economists is that the financial sector development hasa positive effect on economic development process. The literature over the last 20 yearshas culminated in a strong belief that a well functioning financial sector is the essencefor the functioning of modern market economies. However, following the recent crisis of2007-08 this existing body of literature about the positive relationship between financeand growth seems inadequate and fragmented. There is dearth of the literature inquir-ing why the crisis hit countries went into recession despite having high level of financialdevelopment. To deal with the shortcomings of the literature the second chapter in thisthesis deals with the question about the causes for this insignificant (Aghion et al., 2004)or at worst negative relationship (Arcand et al., 2012) between finance and growth. Morespecifically, we provide a systematic explanation for why this finance-growth relation-ship broke recently during the 2007-08 crisis. To this end, we emphasise the importanceof liberalisation with adequate regulations to achieve the macroeconomic objective ofeconomic growth without compromising the stability of the countries. The chapter thenidentifies the threshold beyond which the impact of financial liberalisation on the finance-growth relationship turns insignificant or negative .

This chapter provides a unique assessment of the effects of overall financial liber-alisation policies and its components on the finance-growth relationship for a panel of88 countries over 33 years. Specifically, it uses a comprehensive measure of financial


development obtained through principle component analysis (PCA) and financial liberali-sation index comprising seven different components to analyse the true impact of financialliberalisation on the finance-growth relationship. We use a modern dynamic panel datatechnique to take into account the problems of parametric heterogeneity, endogeneity, out-liers, and omitted variables in growth equations. It questions the mindset that higher levelof financial liberalisation improves financial development in terms of more credit avail-ability and higher opportunities for risk diversification that in turn lead to more growth.In other words whether a larger financial sector resulting from financial liberalisation isalways good for overall economic growth. We do not find a positive effect of higherfinancial liberalisation on finance-growth relationship. In particular, we find that thereis strong positive relationship between lower level of financial liberalistion and finance-growth relationship, however, we observe that there is a threshold (which we estimate as0.85-1 on the scale of 0-1 for the level of liberalistion) beyond which financial liberali-sation starts affecting the finance-growth relationship negatively. We observed that ourresults are robust to using different definitions of financial development and omission ofoutliers.

Our empirical findings show that there is a robust negative association betweenhigher levels of financial liberalisation and the finance-growth relationship. We observethat an insignificant or negative relationship between finance and growth is associatedrobustly with increasing level of financial liberalisation. This result is in line with pre-vious research showing the ineffectiveness of financial liberalisation in inducing growthand investment (Rodrik et al., 1998, Edwards, 2001, Broner and Ventura, 2010). Thus, weconclude that the malaise the world has been mired in during the periods of financial crisesis due to the unregulated and excessive liberalisation of financial markets. This detrimen-tal behaviour is in turn related to the inordinate behaviour of financial institutions that asa result of financial liberalisation involve in excessive risk taking and reckless decisionsthat value short-term profits at the expense of long term stability.

The results of this chapter have much relevance for the current debate regardingfinancial reforms. Given the non-linear effect of finance on development, a poorly reg-ulated financial system can further increase systemic vulnerabilities and can make coun-tries more prone to crises. Our explanation provides a guideline to policy makers that theyshould focus on reforms that give due share to the regulations rather than simply going forfinancial libralisation. Thus, our empirical findings show that a financial sector that growsunnecessarily bigger increases vulnerabilities rather than increasing overall benefits to thesociety.


In Chapter three our contribution to the literature is related to the sequencing offinancial liberalisation reforms and financial crises. Although an extraordinary rise in thevolume of international financial transactions during the last century is the achievement ofliberal capital policies, the world has also paid large costs for this liberty in terms of hugefinancial crises. Despite the fact that the reasons for these intermittent series of crises aremany, there is a strong role of capital account liberalisation (see for instance, Eichengreenet al. (1998)). Ostry et al. (2012) argue that implementation of capital controls savescountries from instabilities during hard times. Following the financial crisis of 2008-09,the International Monetary Fund– after long advocacy of a more open capital account–has also changed its stance on low levels of capital controls as a central feature of itspolicy agenda. Due to the adverse impacts of large and volatile capital flows on financialstability and economic growth IMF now recommends the management of capital flows aspart of an overall approach to safeguarding liberal economies from shocks (Arora et al.,2013).

The empirical evidence also suggests that countries, for example, Chile that em-ployed effective prudential regulations have been able to avoid crises consistently, es-pecially in their banking sector. This chapter takes up the issue of the sequencing offinancial reforms and links its impact to the emergence of financial crises. In this chapterwe hypothesize that the success or failure of financial reforms, in terms of impacting thelikelihood of financial crises, depends on how these reforms are ordered. To the best ofour knowledge there is no systematic empirical study available that links the sequencingof financial liberalisation to the financial crisis.

This chapter provides an especial assessment of the effect of sequencing of financialsector reforms on financial crises for a panel of 28 countries over the period 1973-2005.Specifically, this chapter uses a novel proxy quantitative measure of sequencing that al-lows to identify a possible sequence that makes countries less vulnerable to crises, whichwe do not find in the previous literature studying the sequencing of financial reforms.Our proxy for sequencing takes into account the interaction between liberalisation of thedomestic market, stock market and capital accounts.

Our empirical results show that the probability of financial crises increases whencountries lack effective prudential regulation for their domestic financial sector with in-adequate capital controls on short-term capital flows. The major reason for that is thateffective regulations provide safeguard against short term “toxic” assets that are respon-sible for the crisis in many countries (Rodrik and Velasco, 1999). We start our analysis


by taking Chile as initial benchmark country because of its policy of opening capital ac-count gradually after allowing financial sector to develop. The coefficient for our distancevariable using Chile shows that the larger is the difference between policies of Chile andother countries, the higher are the chances of financial crises, especially banking crisis.

Our empirical results do not change when we generalised our benchmark and takeevery country as benchmark one by one. The results suggest that the distance away fromthe countries that libealised their capital account last increases the probability of crisis.Further our analysis for an alternative specification using dummies variables for the poli-cies of domestic market first or stock market first revealed that higher financial liberalisa-tion with domestic market first or stock market strategy reduces the probability of crisis.Our results are robust to the changes of specification, functional form, and our proxy ofdistance.

However, we find no single optimal order of financial liberalisation that provides apanacea to the problem of rising instabilities in the economies. According to our findings,a country should not invite superfluous capital inflows until its internal structure is notstrong enough to be able to allocate these flows efficiently and its economy is not ableto weather unexpected reversals of these flows. According to our results the order thatcomes out to be more resilient to the crises is the order in which capital account liberal-isation comes last after liberalisation of domestic markets and stock markets. However,as a starting point the countries face an equal choice between domestic market and stockmarket liberalisation.

The results of this chapter are relevant for questions related to the policy stancethat countries need to take to avoid future crises such as the crisis of 2007-08. Giventhe adverse role of unharnessed capital flows in the recent series of crises, there is aneed to have a uniform policy of capital controls at least for short term flows. In the pastcountries such as Malaysia received so much criticism when they imposed capital controlsfollowing the East Asian financial crisis. However, sufficient capital controls can providecentral banks more controls of their monetary policy through which they can stimulategrowth and increase rates of recovery from crises.

4.2 Areas for Further Research

The present research can be extended in a number of directions. First, while thestudies in this thesis, spanning over the period of 1973-2005, improve our understanding


about the impact of financial liberalisation on finance-growth relationship and financialcrises, we can further enhance this understanding by taking into account the changes inthe regulatory framework that have taken place following the financial crisis of 2007-08.The main difficulty for such an enhancement lies on the limitation of data availability.The most updated datasets available on financial liberalisation cover the period between1973 and 2005. Our future research agenda include updating of the data set to date for thechanges in the policies of financial reforms related to the domestic market, stock marketand capital accounts liberalisation.

Second, in this thesis we use linear scales for the measurement of financial liberal-isation. As linear scales are equally spaced in our study, this automatically assumes thatmoving from one scale to the other always signifies same amount of reform/improvementof the sectors. But these reforms and their effect may be nonlinear on growth and volatil-ity (for instance, see Kose et al. (2003), Arteta et al. (2001)). A possible extension forthis study might involve using a non-linear scale for the ordinal rankings assigned for themeasurement of financial liberalisation. Through the non-linear scaling we can capturethe varying effects these policies create on macroeconomic fundamentals and financialcrises.

Third, our study only dealt with broad sequence of reforms between the sectors,without necessarily looking into precise timing of the reforms. Any two countries mayfollow the same sequence of reforms but they may differ in precise timing of the reformsof respective sectors. By taking this timing issue, we may get even better understandingof comparison between countries.

And finally, only analysis of quantity of reforms might not be sufficient for the anal-ysis of the impact of reforms on growth and volatility, we also need to take into accountthe quality of reforms. As Johnston and Sundararajan (1999) argue that banking crisisresulting from poorly managed reforms can affect growth adversely, a further analysis isrequired where both quality and quantity aspects are taken together into account.

Appendices

107

Appendix A

Table A.1: Literature on Relationship between Financial Liberalisation, Financial Devel-opment and Growth

Study Main variable(s) Methodology Sample size Main Finding(s)

Financial Development

Demetriades andHussein (1996)

Bank deposit/GDP,private credit/ GDP

Co-integrationand ECM

16 countries Bi-direction causality

King and Levine(1993a)

liquid liabilities/GDP, Pvtcredit/total credit, Pvtcredit/GDP.

Cross country 80 countries over1960-89

FD affects contemporaneous andfuture growth rates

Rioja and Valev(2007)

Pvt credit, Commercial vs.Central bank ,and liquidliabilities to GDP

GMM 74 countries over1961-1995

Finance affects growth throughcapital accumulation

Beck et al. (2000) pvt credit (equals thevalue of credits byfinancial intermediaries tothe pvt sector by GDP)

Dynamic panel 63 countries over1960-1995

Financial intermediaries GDPgrowth through effects on TFP

Rajan and Zingales(1996)

Ratio of domestic creditplus stock marketcapitalization to GDP andAccounting standards

Test for anindustry’s needfor externalfinance

US as benchmarkin 44 countriesover 1980-1990

FD has a substantial influence oneconomic growth

Carlin and Mayer(2003)

Pvt credit/ GDP Cross section andIV

OECD countriesover 1970-1995

Strong positive impact of thestructure of financial systems onthe growth and investment ofindustries in different countries

De Gregorio andGuidotti (1995)

Domestic credit to privatesector to GDP

OLS 1960-85 Negative impact of financialintermediation and economicgrowth for Latin Americancountries

Arestis andDemetriades (1997)

Log of ratio M2 tonominal GDP forGermany, ratio ofdomestic bank credit toGDP for USA

OLS 1979(1)-91(4) The effect of financialdevelopment on growth vary acrosscountries

continued on the next page

108

109

Table A.1Literature on Relationship between Finance and Growth – continued


Cojocaru et al.(2011)

domestic credit to pvtcredit, Liquidliabilities(M3), log ofdomestic credit providedby the banking sector

System GMMand DifferenceGMM

CEE and CIScountries over1990-2008

Positive effect of financialintermediation on economic growth

Federici and Caprioli(2009)

Average of bank basedand market basedindicators

VAR framework Quarterly data setfor the period1990-2000 for 39countries

Open capital markets leads tomore grwth but also cause crisis

Levine (1999) Private Credit to GDPratio

Cross sectionandGMM

71 countries over1960-1995

The legal and regulatory reformsboost banking sector developmentand accelerate economic growth

Neusser and Kugler(1998)

GDP of financialinstitutions, insurancecompanies, and pensionfunds

Multivariate timeseries

13 OECDcountries overthe period ofthree decades

Financial sector GDP isco-integrated OECD countriesmanufacturing TFP

Rousseau andWachtel (2011)

Credit allocated to pvtsector

Barro’s modeland SystemGMM


The effect of financial developmenton economic growth dampens withrecurring crises

Shan et al. (2003) Bank credit to privatesector/GDP

Granger causalityand VAR.

9 OECDcountries

There is little support that financeleads development

Levine and Zervos(1998)

Size of equity market,turnover to GDP

OLS and IV 47 countries over1976-1993

Stock market and banking sectordevelopment affect growthpositively

Arcand et al. (2012) Pvt credit/GDP, sharestraded/total capitalization

Cross sectionalregression &System GMM

1960-2010 Too much finance affects outputgrowth negatively

Rousseau andWachtel (1998)

Assets of commercialbanks, saving institutions,insurance companies, andpension funds

VECM 5 countries over1870-1929

Measures of financial intensityhave long run relationship with realper capita levels of output and themonetary base

Favarra (2003) Pvt Credit/GDP, liquidliabilities/GDP

GMM and PMG 87 countries over1960-1998

Financial development does nothave a first-order effect oneconomic growth

Financial Sector Reforms

Tressel andDetragiache (2008)

Pvt credit/GDP, FinReform Index, legal origin

Dynamic ARDL 1973-2006 The positive effect of financialdevelopment is linked withdeveloped political institutions

Klein and Olivei(2008)

Liquid Liabilities/GDP,Fin intermediary claims topvt sector/GDP, Binaryvar showing capitalcontrols

OLS 1976-1995 &1986-1995

Capital account liberalisation andfin intermediation affect growthpositively

Bekaert et al. (2005) Private credit/GDP, Equitymarket turnover, Binaryvar for equity mktliberalisation

GMM Four samplescontaining 95,75, 50 and 28

Equity market liberalisation leadsto economic growth

Bekaert et al. (2011) Quinn′s capital accountopenness indicator,official liberalisationindicator, Pvt credit/GDP

Methodology byKing and Levine(1993a)


Financial openness has greatereffects on TFP as compare toinvestment


110



Ayhan Kose et al.(2009)

De-jure and De-Factomeasures

Dynamic Panel 67 countries over1966-2005

De-jure measures compare toDe-facto measures show clearpositive effect on TFP

Ang (2011) Pvt credit/GDP, FL byAbiad et al.

Co-integration 44 countries over1973-2005

Financial reforms reducesinnovation

Ang and McKibbin(2007)

PCA of liquid liabilities,assets, and pvt credt/GDP,index of repression

Co-integration 1960-2001 Financial depth and economicdevelopment are positively related,but causality runs from economicdevelopment to FD.

Roubini and Sala-iMartin (1992)

Index of interest ratedistortions

Barro‘ model 98 countries over1960-1985

Financial repression has negativeeffect on economic growth

Fry (1997) Real interest rate ondeposits

Iterative 3SLS 16 developingcountries over1970-88

Financial Reforms works only witheffective fiscal reform

Levine (2011) Official power index,private monitoring index

Cross country 1960-2005 Effective governance of financialregulatory institutions influencesgrowth

Kaminsky andSchmukler (2008)

Liberalisation index,off-shore borrowing bynon fin inst

NBER approach 1960-2005 Impact of liberalisation depends onthe level of development

Demetriades andRousseau (2011)

FD as M3-M1, FL byAbiad et al.(2008)

Cross section 84 countries over1975-2004

The effect of financial developmentafter financial reforms has declinedin recent decades

Levine (1998) Pvt credit/GDP, legalindicators

GMM and LM 1976-1993 The legal rights explain differencesin the bank sector developmentacross countries

Loayza and Ranciere(2006)

Pvt credit/GDP Panel ECM andARDL


Relationship is positive in the longrun but negative in the short run

La Porta et al. (2002) Pvt credit/GDP, Govtownership index

OLS and IV 92 countries Government ownership reducesgrowth of per capita income

Andrianova et al.(2011)

Govt ownership of banksCaprio et al.(2008)

OLS and IV 1998-2005 Government ownership isassociated with higher economicgrowth

Demirgüç-Kunt andDetragiache (1998b)

Real short term interestrate, bank cash andreserves/bank asset

Multivariate logit 53 countries over1980-95

Banking crises are more likely tooccur in countries with liberalisedfinancial structure.

Cavallo and Cavallo(2010)

Current Account Reversal,Sudden Stops in capitalflows, Hyperinflation

System GMM 78 countries over1970-2004

Stronger political institutes canmitigate or even reverse the longrun negative impact of bankingcrises.

Barth et al. (2004) Bank activity regulatoryvariable

OLS and Logit 107 countries Restrictions hinders bankdevelopment and stability

Barth et al. (2001) Pvt credit/GDP, index ofbank regulation

Probit Over 60 countries More restrictions cause bankingsector to suffer more from bankingcrisiss

Rioja and Valev(2004)

commercial bankassets/commercialbank+central bank assets,Dummy showing Highand low regions

Dynamic GMM 74 countries over1960-1995

The effect on growth is notuniformly positive


111



Buera et al. (2009) Pvt Credit+Pvt bondcapitalization+stockmarket capitalization/GDP

Calibration 1993-2003 Financial frictions distortallocation of capital andentrepreneurial talent acrossproduction units, adverselyaffecting measured productivity

Weller (2001) Credit, IndustrialProduction

Univariate andmultivariateanalysis

27 emergingeconomies over1973-1998

The likelihood of currency crisesand banking crises increases afterFL

Cerra and Saxena(2008)

Currency, banking andtwin crisis

Christina Romerand David Romer(1989)Methodology

190 countriesover 1960-2001

The likelihood of currency crisesand banking crises increases afterFL

Laeven et al. (2010) Asset prices, credit growth DescriptiveAnalysis

USA Financial crises have exposed theflaws in pre-crisis policies,therefore, there is need to rethinkabout the architecture ofmacroeconomic policy

Taylor (2009) Spread between threemonths LIBOR and threemonths overnight indexswap

Descriptive All countriesover 2007-2009

Government actions andinterventions prolong and worsenthe financial crisis

King and Levine(1993b)

Credit issued to privateenterprizes/credit issuedto central and localgovernments+creditissued to public andprivate enterprizes

3SLS 5 countries Financial systems spur growth dueto their role in innovation

112

Table A.2: List of Sample Countries

1 Albania 45 Kazakhstan2 Algeria 46 Kenya3 Argentina 47 Korea4 Australia 48 Kyrgyz Republic5 Austria 49 Latvia6 Azerbaijan 50 Lithuania7 Bangladesh 51 Madagascar8 Belarus 52 Malaysia9 Belgium 53 Mexico10 Bolivia 54 Morocco11 Brazil 55 Mozambique12 Britain 56 Nepal13 Bulgaria 57 Netherlands14 Cameroon 58 New Zealand15 Canada 59 Nicaragua16 Chile 60 Nigeria17 China 61 Norway18 Colombia 62 Pakistan19 Costa Rica 63 Paraguay20 Czech Republic 64 Peru21 Denmark 65 Philippines22 Dominican Republic 66 Poland23 Ecuador 67 Portugal24 Egypt 68 Romania25 El Salvador 69 Russia26 Estonia 70 Senegal27 Ethiopia 71 Singapore28 Finland 72 South Africa29 France 73 Spain30 Georgia 74 Sri Lanka31 Germany 75 Sweden32 Ghana 76 Switzerland33 Greece 77 Tanzania34 Guatemala 78 Thailand35 Hong Kong 79 Tunisia36 Hungary 80 Turkey37 India 81 Uganda38 Indonesia 82 Ukraine39 Ireland 83 United States40 Israel 84 Uruguay41 Italy 85 Uzbekistan42 Jamaica 86 Venezuela43 Japan 87 Vietnam44 Jordan 88 Zimbabwe

113

Table A.3: Principal Component Analysis for Financial Development

Year PCA1 PCA2 PCA3 PCA4 PCA5 PCA61973 0.5396 0.2706 0.1261 0.0485 0.0121 0.0031974 0.5583 0.2249 0.169 0.0338 0.0117 0.00231975 0.5532 0.2166 0.1773 0.0382 0.0111 0.00361976 0.575 0.216 0.1601 0.0343 0.0109 0.00361977 0.5601 0.2215 0.1664 0.034 0.014 0.0041978 0.5782 0.21 0.1617 0.0314 0.013 0.00571979 0.563 0.2335 0.1566 0.0303 0.0111 0.00561980 0.571 0.2416 0.1333 0.0331 0.0156 0.00541981 0.5837 0.2438 0.1253 0.0294 0.0153 0.00261982 0.5965 0.2247 0.1301 0.0317 0.0148 0.00211983 0.5983 0.2037 0.1424 0.0378 0.0156 0.00221984 0.5673 0.2655 0.1296 0.0298 0.0066 0.00111985 0.5542 0.2533 0.1571 0.0285 0.0058 0.00111986 0.564 0.2632 0.1386 0.0287 0.0045 0.00111987 0.5579 0.2599 0.144 0.0332 0.0037 0.00141988 0.5495 0.2768 0.1367 0.0314 0.004 0.00171989 0.5438 0.2854 0.1397 0.0253 0.0036 0.00221990 0.5493 0.2868 0.1357 0.0209 0.0048 0.00251991 0.5369 0.3148 0.1166 0.0252 0.0041 0.00231992 0.5502 0.3127 0.1047 0.0243 0.0054 0.00271993 0.5624 0.2736 0.1191 0.0356 0.0069 0.00231994 0.5916 0.2566 0.1083 0.0352 0.0057 0.00271995 0.5941 0.2515 0.1054 0.0403 0.0058 0.00291996 0.5996 0.222 0.1347 0.0338 0.0058 0.0041997 0.6088 0.2303 0.1125 0.0391 0.0047 0.00441998 0.5989 0.224 0.1302 0.0376 0.0054 0.00391999 0.5797 0.2243 0.1517 0.0345 0.0055 0.00432000 0.6041 0.197 0.1438 0.0456 0.0064 0.00322001 0.577 0.2164 0.1348 0.0617 0.0073 0.00282002 0.5622 0.2305 0.1367 0.0588 0.0086 0.00332003 0.5616 0.2311 0.136 0.0577 0.0101 0.00362004 0.5755 0.2345 0.126 0.0505 0.0097 0.00372005 0.585 0.2239 0.1278 0.0497 0.0102 0.0035Source: Author’s calculation

114

Table A.4: Principal Component Analysis for Financial Devel-opment Vector 1

Year LL/Y DBATBCB PVT/Y CBA/Y DBA/Y BC/BD1973 0.4688 0.3937 0.5385 -0.094 0.5233 0.22921974 0.481 0.4094 0.5289 -0.146 0.514 0.18921975 0.4853 0.4125 0.5234 -0.1931 0.5122 0.14431976 0.4592 0.409 0.517 -0.2059 0.5017 0.24591977 0.4551 0.4082 0.5221 -0.211 0.5088 0.22421978 0.4344 0.4228 0.5157 -0.2254 0.4985 0.25941979 0.429 0.4078 0.5224 -0.2233 0.5053 0.26761980 0.3795 0.435 0.5157 -0.2557 0.4949 0.30081981 0.3652 0.4424 0.5115 -0.2611 0.4845 0.32631982 0.3816 0.4418 0.5083 -0.2513 0.4877 0.31591983 0.3911 0.4462 0.5087 -0.2506 0.4954 0.28461984 0.504 0.2427 0.5384 0.2139 0.5336 0.25851985 0.5132 0.2356 0.5444 0.189 0.541 0.23721986 0.5075 0.2466 0.5409 0.1481 0.5338 0.28661987 0.5165 0.2364 0.5426 0.1607 0.5375 0.26091988 0.512 0.2362 0.5462 0.1209 0.5393 0.27971989 0.5185 0.2079 0.5495 0.159 0.5456 0.25111990 0.5108 0.2195 0.5457 0.1705 0.5418 0.26551991 0.5083 0.2737 0.5508 0.1042 0.5456 0.23411992 0.4734 0.3688 0.5384 -0.0431 0.5247 0.26991993 0.4584 0.3863 0.5263 -0.1278 0.5028 0.30751994 0.4418 0.4128 0.5156 -0.1478 0.4956 0.31811995 0.4481 0.3977 0.5124 -0.1466 0.4897 0.34221996 0.4698 0.4222 0.5081 -0.1681 0.4966 0.26071997 0.459 0.4264 0.5041 -0.1973 0.4874 0.27731998 0.465 0.4326 0.5109 -0.1963 0.4962 0.22541999 0.4829 0.4266 0.5177 -0.1393 0.5089 0.19612000 0.4637 0.4026 0.5072 -0.2087 0.4959 0.27612001 0.4629 0.3707 0.5175 -0.202 0.5076 0.28642002 0.4601 0.3622 0.5265 -0.1725 0.5133 0.29442003 0.4554 0.3714 0.5252 -0.182 0.5114 0.29022004 0.4401 0.3764 0.5148 -0.2391 0.4989 0.30582005 0.4367 0.3798 0.5097 -0.2714 0.497 0.2908

Source: Author’s calculationLL/Y=Log of Liquid Liabilities to GDP RatioBA/BCBA =log of Deposits money bank assets to deposit money bankassets and central bankPVT/Y=log of Bank private credit to GDP (%)CBA/Y=log of Central bank assets to GDP (%)DBA/Y=log of Deposits money bank assets to GDP (%)BC/BD=log of bank credit to bank deposit

115

Table A.5: Cross-Correlation Table for Components of FD

Variables LL/Y BA/BCBA PVT/Y CBA/Y DBA/Y BC/BDLL/Y 1.0000

BA/BCBA 0.3747 1.0000(0.0000)

PVT/Y 0.8722 0.5473 1.0000(0.0000) (0.0000)

CBA/Y 0.2181 -0.5610 -0.0704 1.0000(0.0000) (0.0000) (0.1679)

DBA/Y 0.9390 0.4929 0.9433 0.1079 1.0000(0.0000) (0.0000) (0.0000) (0.0341)

BC/BD 0.1163 0.2660 0.5281 -0.2475 0.3042 1.0000(0.0236) (0.0000) (0.0000) (0.0000) (0.0000)

1 LL/Y=Log of Liquid Liabilities to GDP Ratio2 BA/BCBA =Log of Deposits money bank assets to deposit money bank assets andcentral bank3 PVT/Y=Log of Bank private credit to GDP ratio4 CBA/Y=Log of Central bank assets to GDP ratio5 DBA/Y=Log of Deposits money bank assets to GDP ratio6 BC/BD=Log of bank credit to bank deposit ratio7 p-values in bracket

116

Tabl

eA

.6:C

ross

-Cor

rela

tion

Tabl

e

Var

iabl

esD

.Cre

dit

Cr.

Cei

lings

Cr.

Con

trol

sI.R

cont

rols

Ent

ryB

arri

ers

Supe

rvis

ion

Priv

atiz

atio

nIn

t.Cap

ital

Sec.

Mar

kets

D.C

redi

t1.

000

Cr.

Cei

lings

0.40

81.

000

(0.0

000)

Cr.

Con

trol

s0.

972

0.68

61.

000

(0.0

00)

(0.0

00)

I.R

cont

rols

0.61

90.

576

0.65

81.

000

(0.0

00)

(0.0

00)

(0.0

00)

Ent

ryB

arri

ers

0.53

80.

403

0.57

70.

560

1.00

0(0

.000

)(0

.000

)(0

.000

)(0

.000

)Su

perv

isio

n0.

581

0.47

10.

611

0.58

40.

571

1.00

0(0

.000

)(0

.000

)(0

.000

)(0

.000

)(0

.000

)Pr

ivat

izat

ion

0.48

70.

297

0.49

60.

442

0.42

60.

485

1.00

0(0

.000

)(0

.000

)(0

.000

)(0

.000

)(0

.000

)(0

.000

)In

t.C

apita

l0.

556

0.46

90.

595

0.59

80.

513

0.57

30.

521

1.00

0(0

.000

)(0

.000

)(0

.000

)(0

.000

)(0

.000

)(0

.000

)(0

.000

)Se

c.M

arke

ts0.

601

0.50

20.

639

0.62

40.

553

0.64

10.

496

0.67

61.

000

(0.0

00)

(0.0

00)

(0.0

00)

(0.0

00)

(0.0

00)

(0.0

00)

(0.0

00)

(0.0

00)

D.C

redi

t=D

irec

ted

Cre

dit

Cr.

Cei

lings

=C

redi

tCei

lings

I.R.C

ontr

ols=

Inte

rest

Rat

eC

ontr

ols

117

Tabl

eA

.7:D

ata

App

endi

x

Var

iabl

esD

efini

tion

Sour

ceG

DP

perc

apita

grow

thra

teG

DP

grow

thpe

rcap

itaat

cons

tant

pric

ess

inlo

calc

urre

ncy

Wor

ldB

ank

Fina

ncia

lRef

orm

Inde

xC

redi

tCon

trol

sC

redi

tres

tric

tion

toce

rtai

nse

ctor

san

dhi

ghre

serv

ere

quir

emen

tsfr

omce

ntra

lban

kA

biad

etal

.(200

8)In

tere

stR

ate

Con

trol

sW

heth

erin

tere

stra

tes

are

“adm

inis

trat

ivel

yse

tinc

ludi

ngw

heth

erth

ego

vern

men

tdir

ectly

cont

rols

Abi

adet

al.(2

008)

inte

rest

rate

s,or

whe

ther

floor

s,ce

iling

s,or

inte

rest

rate

band

sex

ist."

Ent

ryB

arri

ers

Whe

ther

ther

eis

rest

rict

ions

onen

try

offo

reig

nba

nks

,sco

peof

bank

san

dge

ogra

phic

area

sA

biad

etal

.(200

8)St

ate

Ow

ners

hip

inth

eB

anki

ngSe

ctor

Whe

ther

the

stat

eow

nsba

nks.

The

data

base

isco

ded

as“t

hres

hold

of50

%,2

5%an

d10

%to

delin

eate

Abi

adet

al.(2

008)

the

grad

esbe

twee

nfu

llre

pres

sion

and

full

liber

alis

atio

n"C

apita

lAcc

ount

Res

tric

tions

Whe

ther

rest

rict

ions

onin

tern

atio

nalfi

nanc

ialt

rans

actio

nsar

eim

pose

dA

biad

etal

.(200

8)Pr

uden

tialR

egul

atio

nsan

dSu

perv

isio

nW

heth

er“c

ount

ryad

opts

risk

-bas

edca

pita

lade

quac

yra

tios

base

don

the

Bas

elIc

apita

lacc

ord"

Abi

adet

al.(2

008)

Sec

Mar

ketP

olic

ies

Whe

ther

gove

rnm

ente

ncou

rage

sor

rest

rict

sde

velo

pmen

tofs

ecur

ities

mar

kets

Abi

adet

al.(2

008)

Liq

uid

Lia

bilit

ies/

GD

P(%

)R

atio

ofliq

uid

liabi

litie

sto

GD

PB

eck

etal

.(20

09)

Dep

bank

asse

ts/d

epba

nk+c

entr

alba

nkas

sets

Rat

ioof

depo

sitm

oney

bank

clai

ms

ondo

mes

ticno

nfina

ncia

lrea

lsec

tor(

asde

fined

abov

e)B

eck

etal

.(20

09)

toth

esu

mof

depo

sitm

oney

bank

and

Cen

tral

Ban

kcl

aim

son

dom

estic

nonfi

nanc

ial

real

sect

or(a

sde

fined

abov

e)re

alse

ctor

(as

defin

edab

ove)

Ban

kPv

tCre

dit/G

DP

Priv

ate

cred

itby

depo

sitm

oney

bank

sto

GD

PB

eck

etal

.(20

09)

Cen

tban

kas

set/G

DP

Cla

ims

ondo

mes

ticre

alno

nfina

ncia

lsec

torb

yth

eC

entr

alB

ank

asa

shar

eof

GD

PB

eck

etal

.(20

09)

Dep

osit

Mon

eyba

nkas

se/G

DP

Cla

ims

ondo

mes

ticre

alno

nfina

ncia

lsec

torb

yde

posi

tmon

eyba

nks

asa

shar

eof

GD

PB

eck

etal

.(20

09)

Ban

kC

redi

t/Ban

kD

epPr

ivat

ecr

edit

byde

posi

tmon

eyba

nks

asa

shar

eof

dem

and,

time

and

savi

ngde

posi

tsin

depo

sitm

oney

bank

sBec

ket

al.(

2009

)G

over

nmen

tsiz

eG

ener

algo

vern

men

tfina

lcon

sum

ptio

nex

pend

iture

(%of

GD

P)W

orld

Ban

kTr

ade

open

ness

Trad

eis

the

sum

ofex

port

san

dim

port

sof

good

san

dse

rvic

esm

easu

red

asa

shar

eof

gros

sdo

mes

ticpr

oduc

tW

orld

Ban

kIn

vest

men

tGro

wth

Gro

ssca

pita

lfor

mat

ion

atco

nsta

ntpr

ice

inlo

calc

urre

ncy

(ann

ual%

grow

th)

Wor

ldB

ank

Infla

tion

Log

of1+

Con

sum

erpr

ice

inde

x(2

005

=10

0)W

orld

Ban

kE

duca

tion

Log

ofav

erag

eye

arof

seco

ndar

ysc

hool

ing

Bar

ro(1

996)

118

Table A.8: Summary Statistics

Variable Mean Std.Dev.

Min. Max. N

GDP per capita growth rate 1.864 4.769 -45.32 25.113 2644Financial Reform Index 0.496 0.304 0 1 2572Financial Development

Liquid Liabilities/ GDP (%) 48.534 34.513 0.025 255.936 2259Dep bank assets/dep bank+central bank assets 81.956 18.358 9.122 100 2397Bank Pvt Credit/GDP 40.316 34.926 0.005 200.988 2331Cent bank asset/GDP 7.428 8.964 0 77.092 2239Deposit Money bank asset/ GDP 49.933 40.11 0.001 251.553 2337Bank Credit/Bank Dep 102.612 53.623 5.962 615.721 2478

Control VariablesGovernment size 14.98 5.819 1.375 43.479 2573Trade openness 66.689 48.757 6.320 430.563 2595Investment Growth 4.896 19.079 -376.2 223.084 2354Inflation 47.253 34.904 0 115.871 2426

119

Table A.9: Cross-Correlation Table

Variables GDPPCG LLYo FL FD Inv TO G InflationEducationGDPPCG 1.0000

LLYo 0.0742 1.0000(0.1082)

FL 0.2074 0.5640 1.0000(0.0000) (0.0000)

FD 0.2666 0.6378 0.3740 1.0000(0.0000) (0.0000)(0.0000)

Inv 0.5462 -0.1577 0.0152 -0.1220 1.0000(0.0000) (0.0011)(0.7554)(0.0192)

TO 0.1513 0.2388 0.3528 0.1959 0.0107 1.0000(0.0009) (0.0000)(0.0000)(0.0001)(0.8242)

G -0.0962 0.4509 0.3073 0.2743 -0.1849 0.1123 1.0000(0.0362) (0.0000)(0.0000)(0.0000)(0.0001)(0.0151)

Inflation 0.3163 0.2597 0.6637 0.4062 0.1133 0.2585 -0.0339 1.0000(0.0000) (0.0000)(0.0000)(0.0000)(0.0239)(0.0000)(0.5034)

Education 0.0252 0.7116 0.6584 0.3814 -0.1184 0.2333 -0.2605 0.3774 1.0000(0.5894) (0.0000)(0.0000)(0.0000)(0.0157)(0.0000)(0.0000)(0.0000)

GDPPCG=GDP per capita growth rateLLYo =log of initial level of GDPFD=Financial DevelopmentInv=InvestmentTO=Trade opennessG=Size of Governmentp-values in bracket

120

Table A.10: Financial Liberalisation, Financial Development, and Economic Growth, Two-Step Sys-tem GMM without Outliers


FD 1.669 1.174 1.468 0.674 1.050 1.076 1.756 1.685[0.504]***[0.314]*** [0.570]** [0.469] [0.679] [0.339]***[0.468]***[0.612]***

FL -3.581[1.455]**

FL*FD -2.074[0.741]***

Banking supervision 0.627[0.477]




Credit controls -0.714[0.246]***


Securities markets 0.442[0.501]




Interest rate liberalisation -0.128[0.393]


Capital controls -0.776[0.319]**



Investment 0.275 0.276 0.250 0.250 0.277 0.282 0.290 0.292[0.041]***[0.043]***[0.031]***[0.037]***[0.049]***[0.037]***[0.040]***[0.048]***

Trade openness 0.012 0.008 0.004 0.011 0.002 0.005 0.004 0.005[0.011] [0.015] [0.013] [0.014] [0.014] [0.011] [0.012] [0.010]

Government size -0.007 -0.052 -0.046 -0.062 0.014 -0.021 0.028 0.032[0.077] [0.089] [0.066] [0.083] [0.095] [0.103] [0.084] [0.085]

Inflation(CPI) 0.270 0.305 0.177 0.496 0.213 0.118 -0.160 0.192[0.326] [0.271] [0.246] [0.334] [0.332] [0.362] [0.393] [0.315]

Education 2.257 0.707 1.964 2.231 1.589 2.255 2.304 2.835[0.905]** [1.100] [1.235] [0.858]** [1.384] [0.943]** [0.797]***[0.787]***

Constant 1.536 3.522 4.895 2.351 2.761 3.139 3.656 4.262[2.832] [2.601] [3.272] [2.610] [2.813] [2.647] [3.272] [2.668]

Observations 328 328 328 328 328 328 328 328F 16.510 15.650 17.533 14.536 11.650 19.115 17.818 12.503Hansen p-value 0.758 0.482 0.911 0.891 0.580 0.855 0.808 0.774AR1 test p-value 0.001 0.000 0.000 0.004 0.000 0.000 0.001 0.000AR2 test p-value 0.641 0.993 0.726 0.548 0.878 0.722 0.489 0.533No of countries 73 73 73 73 73 73 73 73No of instruments 66 66 66 66 66 66 66 66The table reports regression results for two-step system GMM after excluding outliers using Hadi (1992) method and robust standard errors are in

brackets. * p<0.10, ** p<0.05, *** p<0.01. FD here is obtained using PCA measures on six measures of FD: Log of liquidity liability/GDP, logof deposit money bank assets/ deposit bank assets plus central bank assets, log of private credit/GDP, log of central bank assets/GDP, log of depositmoney bank assets/ GDP, and log bank credit/bank deposit. FL index is taken from Abiad, Detragiache and Tressel (2008).

121

Table A.11: Financial Liberalisation, Financial Development, and Economic Growth, Two-Step Sys-tem GMM with PVT/Y


FD 0.082 0.035 0.057 0.038 0.051 0.020 0.073 0.050[0.032]** [0.017]** [0.037] [0.033] [0.038] [0.017] [0.032]** [0.033]

FL 1.862[2.001]

FL*FD -0.088[0.038]**




Entry barriers*FD -0.019[0.012]



Securities markets 0.961[0.480]**



Privatization*FD -0.008[0.006]





Log in level GDPPC 0.241 -0.274 0.141 0.257 0.000 0.272 0.056 0.293[0.368] [0.394] [0.426] [0.582] [0.376] [0.371] [0.332] [0.439]

Investment 0.286 0.273 0.253 0.298 0.262 0.277 0.290 0.296[0.054]***[0.042]***[0.048]***[0.042]***[0.045]***[0.042]***[0.046]***[0.050]***

Trade openness 0.002 0.004 0.001 0.003 0.004 0.007 0.002 0.002[0.005] [0.006] [0.006] [0.006] [0.004] [0.005] [0.005] [0.005]

Government size -0.151 -0.123 -0.108 -0.150 -0.128 -0.090 -0.113 -0.144[0.086]* [0.102] [0.064]* [0.094] [0.081] [0.091] [0.085] [0.091]

Inflation(CPI) 0.773 0.579 0.677 0.713 0.862 0.958 0.815 1.077[0.400]* [0.430] [0.336]** [0.410]* [0.392]** [0.344]*** [0.426]* [0.413]**

Education 0.045 0.065 0.371 -0.045 -0.144 0.455 -0.120 0.226[0.906] [0.526] [0.843] [0.846] [0.708] [0.898] [0.501] [1.127]

Constant -3.864 -0.559 -2.352 -3.853 -3.019 -4.427 -3.748 -4.649[2.720] [2.682] [2.965] [3.847] [2.736] [2.470]* [2.400] [3.048]


The table reports regression results using two-step system GMM and robust standard errors are in brackets. * p<0.10,** p<0.05, *** p<0.01. FD here is private credit/GDP ratio. FL index is taken from Abiad, Detragiache and Tressel(2008).

122

Table A.12: Financial Liberalisation, Financial Development, and Economic Growth, One-Step Sys-tem GMM with PVT/Y without Outliers


FD 0.066 0.024 0.051 0.044 0.058 0.031 0.057 0.048[0.019]*** [0.013]* [0.022]** [0.031] [0.030]* [0.020] [0.020]*** [0.025]*

FL 1.378[1.990]

FL*FD -0.080[0.023]***





Credit controls -0.247[0.493]


Securities markets 1.117[0.499]**






Capital account openness -0.358[0.486]


Log in level GDPPC -0.082 -0.421 -0.292 -0.248 -0.563 0.049 -0.253 -0.255[0.395] [0.354] [0.371] [0.393] [0.400] [0.384] [0.371] [0.385]

Investment 0.264 0.262 0.240 0.269 0.247 0.268 0.251 0.265[0.039]***[0.044]***[0.041]***[0.041]***[0.041]***[0.040]***[0.039]***[0.039]***

Trade openness 0.019 0.010 0.015 0.015 0.006 0.019 0.016 0.016[0.009]** [0.009] [0.009] [0.010] [0.010] [0.009]** [0.007]** [0.009]*

Government size -0.041 -0.076 -0.043 -0.110 -0.071 -0.037 -0.077 -0.087[0.064] [0.080] [0.068] [0.066]* [0.066] [0.077] [0.070] [0.070]

Inflation (CPI) 0.547 0.539 0.377 0.775 0.409 0.521 0.569 0.687[0.296]* [0.285]* [0.265] [0.347]** [0.259] [0.334] [0.296]* [0.271]**

Education 0.431 0.213 0.389 0.522 0.424 0.301 0.400 0.585[0.148]*** [0.183] [0.183]** [0.212]** [0.223]* [0.213] [0.199]** [0.196]***

Constant -3.448 -0.678 -2.267 -3.170 -1.343 -4.443 -2.727 -3.587[2.166] [1.979] [2.166] [2.229] [2.239] [2.175]** [2.092] [2.213]


The table reports regression results for two-step system GMM after excluding outliers using Hadi (1992) method androbust standard errors are in brackets. * p<0.10, ** p<0.05, *** p<0.01. FD here is private credit/GDP ratio. FL indexis taken from Abiad, Detragiache and Tressel (2008).

123

Table A.13: Financial Liberalisation, Financial Development, and Economic Growth, Two-Step Sys-tem GMM with LLR


FD 0.067 0.040 0.060 0.034 0.050 0.051 0.059 0.050[0.023]*** [0.017]** [0.027]** [0.030] [0.021]** [0.022]** [0.024]** [0.032]

FL 1.405[2.131]

FL*FD -0.080[0.025]***


Banking supervision*FD -0.018[0.007]**





Securities markets 0.924[0.570]




Interest rate liberalisation 0.873[0.430]**




Log in level GDPPC 0.301 -0.313 -0.176 0.118 0.093 0.182 0.194 0.162[0.455] [0.458] [0.460] [0.483] [0.584] [0.554] [0.451] [0.496]

Investment 0.279 0.260 0.273 0.306 0.281 0.280 0.312 0.290[0.049]***[0.049]***[0.036]***[0.048]***[0.042]***[0.052]***[0.044]***[0.046]***

Trade openness 0.004 0.004 0.001 -0.001 0.006 0.012 0.002 0.001[0.006] [0.006] [0.006] [0.006] [0.005] [0.006]** [0.005] [0.007]

Government size -0.161 -0.125 -0.098 -0.189 -0.117 -0.086 -0.123 -0.161[0.071]** [0.103] [0.068] [0.096]* [0.089] [0.088] [0.128] [0.093]*

Inflation(CPI) 0.646 0.375 0.507 0.675 0.556 0.423 0.635 0.880[0.389] [0.366] [0.306] [0.503] [0.459] [0.335] [0.427] [0.476]*

Education 0.603 0.381 0.761 -0.005 0.375 0.761 0.004 0.630[0.853] [0.766] [1.047] [0.997] [0.860] [1.040] [0.810] [1.150]

Constant -3.593 0.189 -0.694 -2.758 -3.286 -3.880 -4.190 -3.782[2.800] [2.584] [2.978] [3.359] [3.798] [3.169] [2.691] [2.907]


The table reports regression results using two-step system GMM and robust standard errors are in brackets. * p<0.10,** p<0.05, *** p<0.01. FD here is liquid liability ratio. FL index is taken from Abiad, Detragiache and Tressel (2008).

124

Table A.14: Financial Liberalisation, Financial Development, and Economic Growth, One-Step Sys-tem GMM with LLR without outliers


FD 0.042 0.023 0.042 0.010 0.045 0.028 0.051 0.045[0.017]** [0.012]* [0.019]** [0.023] [0.019]** [0.013]** [0.016]*** [0.018]**

FL 1.273[1.917]

FL*FD -0.055[0.024]**

Banking supervision 0.575[0.604]

Banking supervision*FD -0.009[0.007]



credit controls -0.627[0.567]


Security markets 0.849[0.505]*






Capital controls -0.063[0.455]



Investment 0.266 0.249 0.237 0.260 0.248 0.266 0.258 0.268[0.039]***[0.042]***[0.035]***[0.040]***[0.040]***[0.041]***[0.036]***[0.040]***

Trade openness 0.017 0.011 0.015 0.008 0.008 0.016 0.018 0.013[0.010]* [0.012] [0.010] [0.011] [0.011] [0.009]* [0.009]* [0.009]

Government size 0.003 -0.026 0.004 -0.045 -0.013 0.043 0.004 0.016[0.060] [0.077] [0.067] [0.072] [0.069] [0.073] [0.069] [0.063]

Inflation(CPI) 0.429 0.401 0.288 0.700 0.274 0.262 0.315 0.406[0.334] [0.303] [0.318] [0.374]* [0.302] [0.331] [0.330] [0.312]

Education 2.122 1.491 1.992 2.021 1.888 1.833 1.617 2.240[0.866]** [0.922] [0.902]** [0.927]** [0.982]* [0.974]* [0.908]* [0.878]**

Constant -0.917 0.903 -0.522 0.659 -0.535 -1.520 -1.042 -1.224[2.744] [2.867] [2.902] [3.051] [3.228] [2.661] [2.673] [2.675]


The table reports regression results for one-step system GMM after excluding outliers using Hadi (1992) method androbust standard errors are in brackets. * p<0.10, ** p<0.05, *** p<0.01. FD here is liquid liability ratio. FL index istaken from Abiad, Detragiache and Tressel (2008).

Appendix B

Table B.1: Number of Reversals after Full liberalisation

Country DFS KA SM FL

Argentina 2 3 1 2Brazil 1 0 0 1Canada 0 0 0 0Chile 1 0 0 1Colombia 1 0 0 0Denmark 0 0 0 0Finland 0 0 0 0France 0 0 0 0Germany 0 0 0 0Hong Kong 0 0 0 0Indonesia 0 1 0 0Ireland 0 0 0 0Italy 0 0 0 0Japan 0 0 0 0Korea 0 0 0 0Malaysia 0 1 1 0Mexico 0 1 0 0Norway 1 0 0 0Peru 0 0 0 0Philippines 0 N/A 0 0Portugal 0 0 1 0Spain 0 1 0 0Sweden 0 0 0 0Taiwan 0 0 0 0Thailand 1 1 0 0United Kingdom 0 0 0 0United States 0 0 0 0Venezuela 2 3 3 3Source: Authors calculations

125

126

Table B.2: Average Index of Financial Reforms

DFS Capital Account Stock Markets

Country 1973-83 1984-94 1995-05 1973-83 1984-94 1995-05 1973-83 1984-94 1995-05

Argentina 2 2.273 2.819 1.819 1.455 2.819 1.455 2.091 2.546

Brazil 1.546 2.182 3 1 2 2.546 2 2.364 3

Canada 3 3 3 2.819 3 3 3 3 3

Chile 2.546 2.91 3 1.364 1 2.637 1 2 3

Colombia 2.091 2.91 3 1 1 2.637 1 1.728 3

Denmark 2 3 3 1 3 3 3 3 3

Finland 1 2.273 3 1 2 3 2 2.455 3

France 1 2.819 3 1 3 3 3 3 3

Germany 3 3 3 2.273 3 3 3 3 3

Hong Kong 1 1 2.364 3 3 3 3 3 3

Indonesia 1.637 3 3 1.546 1 1 1 2 3

Ireland 1 2.728 3 1.455 2 3 2 2.273 3

Italy 1.728 3 3 1 3 3 3 3 3

Japan 1.455 2.273 3 1.728 1 3 1 2.819 3

Korea 1 1.637 1 1 1 2.91 1 1.364 2.728

Malaysia 1.455 1.546 2.182 1.455 2.091 1.819 2.091 3 2.637

Mexico 1.819 2.091 3 2.728 1 3 1 1.91 3

Norway 1.182 2.455 3 1.182 1 3 1 2.546 3

Peru 1.819 1.728 3 2 1 3 1 1.546 3

Philippines 1.273 3 3 1.637 1 2 1 1.91 3

Portugal 1 2.455 3 1 1.546 3 1.546 2.637 3

Spain 2.182 3 3 2.182 3 3 3 3 3

Sweden 1.546 2.91 3 1 2.364 3 2.364 3 3

Taiwan 1 2.364 3 1 1 2.819 1 1.728 2.728

Thailand 1 1.819 2.364 1.273 1 2.728 1 2.091 3

United Kingdom 1.546 3 3 2.273 3 3 3 3 3

United States 2.182 3 3 2.91 3 3 3 3 3

Venezuela 1.455 2 2.364 2.819 2.273 2.182 2.273 2.364 2.364

Note: The table shows average of indexes covering reforms in domestic sector, capital ac-count and stock markets for 28 countries for three different periods: The first period is be-tween 1973-1983, second period is between 1984-1994, and the last period is between 1995-2005. The value of the indexes for individual sectors range from 1 to 3. Therefore, the higheraverage number shows higher level of financial liberalisation in the respective sectors.

127

Table B.3: List of Sample Countries

1 Argentina2 Brazil3 Canada4 Chile5 Colombia6 Denmark7 Finland8 France9 Germany

10 Hong Kong11 Indonesia12 Ireland13 Italy14 Japan15 Korea16 Malaysia17 Mexico18 Norway19 Peru20 Philippines21 Portugal22 Spain23 Sweden24 Taiwan25 Thailand26 United Kingdom27 United States28 Venezuela

128

Table B.4: Data Appendix

Variables Description SourceGDP per capita growth rateGDP per capita growth rate World Bank (WDI)Bank Pvt Credit/GDP Private credit by deposit money banks

to GDP IFS, IMFGovernment size General government final consumption

expenditure (% of GDP) World Bank (WDI)Currency crisis Annual depreciation of local

currency of 15% or more Reinhart and Rogoff (2009)Banking crisis Bank runs Reinhart and Rogoff (2009)Capital accountliberalisation Regulation on offshore borrowing

multiple exchange rate andcontrols on capital outflows Kaminsky and Schmukler (2008)

Domestic financial systemliberalisation Regulation on deposit interest rate

lending interest rates, allocationof credit and foreign currencydeposits Kaminsky and Schmukler (2008)

Stock market liberalisation Regulation on the acquisitionof shares in the domestic stockmarket by foreigners, repatriation ofcapital & repatriaion of interest anddividends Kaminsky and Schmukler (2008)

The degree of realcurrency overvaluation Deviation from the fitted trend

using HP filter Darvas (2012)Real interest rate (RIR) Nominal interest rate minus

contemporaneous rate of inflation IFS on the basis ofavailability between nominal intereston short-term government stock, discount rate, or commercialbank deposit rate

NFA/GDP ratio Net foreign assetsas a ratio of GDP Lane and Milesi-Ferretti (2007)

M2/Total Reserves M2 to Total reserves comprise holdings of monetary gold, special drawing rights, reserves of IMF membersheld by the IMF, and holdings of foreign exchangeunder the control of monetary authorities. World Bank (WDI)

129

Table B.5: Criteria to Define Liberalisation Periods

Capital AccountFull liberalisationBorrowing abroad by banks & Corp There are no restrictions on banks and corporation to borrow abroad . Reserve

requirements are less than 10%. The required minimum maturity is no lessthan 2 years. And

Multiple exchange rates There are no special exchange rates for either current account or capital accounttransactions.There are no restrictions on capital outflows

Partial liberalisationBorrowing abroad by banks & Corp There are some restrictions on banks and corporation to borrow abroad . Reserve

requirements are between 10 and 50%. The required minimum maturity might bebetween 2 and 5 years. Or

Multiple exchange rates There are special exchange rates for either current account or capital accounttransactions. There might be some restrictions to capital outflows

No liberalisationBorrowing abroad by banks & Corp There are complete restrictions on banks and corporation to borrow abroad . Reserve

requirements is higher than 50%. The required minimum maturity might bemore than 5 years. Or

Multiple exchange rates There are special exchange rates for current account or capital accounttransactions. There are restrictions to capital outflows

Domestic Financial SectorFull liberalisationLending and Borrowing Interest rates There are no controls (ceilings and floors) on interest rates. AndOther Indicators No credit controls (subsidies to certain sectors or certain credit allocations).

Deposits in foreign currency are likely permittedPartial liberalisationLending and Borrowing Interest rates There might be controls in either lending or borrowing rates(ceilings and

floors) on interest rates. OrOther Indicators There are controls in the allocation of credit (subsidies to certain sectors or

certain credit allocations). Deposits in foreign currency might not be permittedNo liberalisationLending and Borrowing Interest rates There are controls (ceilings and floors) on interest rates.OrOther Indicators There are controls on allocation of credit (subsidies to certain sectors or certain

credit allocations). Deposits in foreign currency are likely not permitted

Stock MarketFull liberalisationAcquisition by foreign Investors Foreign investors are allowed to hold domestic equity without restrictions

(except for certain specific reasons). AndRepatriation Repatriation of capital, dividends, and interest is allowed within two years

of initial investmentPartial liberalisationAcquisition by foreign Investors Foreign investors are allowed to hold up to 49% of each company′s outstanding

equity. There might be restrictions on investment in certain sectors. OrRepatriation Repatriation of capital, dividends, and interest is allowed within two-five years

of initial investmentNo liberalisationAcquisition by foreign Investors Foreign investors are not allowed to hold domestic equity. OrRepatriation Repatriation of capital, dividends, and interest is after five years of initial investment

Source: Adapted from Kaminsky and Schmukler (2008)

Bibliography

Abiad, A. G., Detragiache, E. and Tressel, T. (2008). A new database of financial reforms,IMF Working Papers 08/266, International Monetary Fund.

Abiad, A., Oomes, N. and Ueda, K. (2008). The quality effect: Does financial liberaliza-tion improve the allocation of capital?, Journal of Development Economics 87(2): 270–282.

Acemoglu, D., Aghion, P. and Zilibotti, F. (2006). Distance to frontier, selection, andeconomic growth, Journal of the European Economic Association 4(1): 37–74.

Acharya, V. and Richardson, M. (2009). Causes of the financial crisis, Critical Review21(2-3): 195–210.

Agénor, P.-R. and Montiel, P. J. (2008). Development macroeconomics, Princeton Uni-versity Press.

Aghion, P. and Durlauf, S. N. (2005). Handbook of economic growth, Vol. 1, Elsevier.

Aghion, P., Howitt, P. and Mayer-Foulkes, D. (2004). The effect of financial develop-ment on convergence: Theory and evidence, Working Paper 10358, National Bureau ofEconomic Research.

Akerlof, G. A. and Shiller, R. J. (2010). Animal spirits: How human psychology drivesthe economy, and why it matters for global capitalism, Princeton University Press.

Allen, F., Qian, J. and Qian, M. (2005). Law, finance, and economic growth in China,Journal of Financial Economics 77(1): 57–116.

Andrianova, S., Demetriades, P. and Shortland, A. (2011). Government ownership ofbanks, institutions and economic growth, Economica 79(315): 449–469.

Ang, J. (2008). A survey of recent developments in the literature of finance and growth,Journal of Economic Surveys 22(3): 536–576.

Ang, J. (2011). Financial development, liberalization and technological deepening, Euro-pean Economic Review 55(5): 688–701.

130

131

Ang, J. and McKibbin, W. (2007). Financial liberalization, financial sector developmentand growth: evidence from Malaysia, Journal of Development Economics 84(1): 215–233.

Arcand, J., Berkes, E. and Panizza, U. (2012). Too much finance, IMF Working Papers12/161, International Monetary Fund.

Arellano, M. and Bond, S. (1991). Some tests of specification for panel data: Monte carloevidence and an application to employment equations, The Review of Economic Studies58(2): 277–297.

Arellano, M. and Bover, O. (1995). Another look at the instrumental variable estimationof error-components models, Journal of Econometrics 68(1): 29–51.

Arestis, P. (2006). Financial liberalization and the relationship between finance andgrowth, A Handbook of Alternative Monetary Economics pp. 346–364.

Arestis, P. and Caner, A. (2004). Financial liberalization and poverty: Channels of influ-ence, The Levy Economics Institute Working Paper 411.

Arestis, P. and Demetriades, P. (1997). Financial development and economic growth:Assessing the evidence, The Economic Journal 107(442): 783–799.

Arestis, P., Demetriades, P., Fattouh, B. and Mouratidis, K. (2002). The impact of fi-nancial liberalization policies on financial development: Evidence from developingeconomics, International Journal of Finance & Economics 7(2): 109–121.

Arestis, P. and Glickman, M. (2002). Financial crisis in Southeast Asia: Dispelling illu-sion the Minskyan way, Cambridge Journal of Economics 26(2): 237–260.

Arora, V., Habermeier, K., Ostry, J. D. and Weeks-Brown, R. (2013). The liberalizationand management of capital flows: An institutional view, Revista de Economía Institu-cional 15(28): 205–255.

Arteta, C., Eichengreen, B. and Wyplosz, C. (2001). When does capital account liber-alization help more than it hurts?, NBER Working Paper No. 8414, Cambridge, Mas-sachusetts: National Bureau of Economic Research.

Arun, T. and Turner, J. D. (2002). Financial sector reforms in developing countries: TheIndian experience, The World Economy 25(3): 429–445.

Ayhan Kose, M., Prasad, E. and Terrones, M. (2009). Does openness to internationalfinancial flows raise productivity growth?, Journal of International Money and Finance28(4): 554–580.

Bagehot, W. (1878). Lombard Street: A description of the money market, CambridgeUniversity Press.

132

Bandiera, O., Caprio, G., Honohan, P. and Schiantarelli, F. (2000). Does financial reformraise or reduce saving?, Review of Economics and Statistics 82(2): 239–263.

Barro, R. (1991). Economic growth in a cross section of countries, The Quarterly Journalof Economics 106(2): 407–443.

Barro, R. J. (1996). Determinants of economic growth: A cross-country empirical study,NBER Working Paper No. 5698, Cambridge, Massachusetts: National Bureau of Eco-nomic Research.

Barro, R. J. (2001). Economic growth in East Asia before and after the financial cri-sis, NBER Working Paper No. 8330, Cambridge, Massachusetts: National Bureau ofEconomic Research.

Barro, R. J., Mankiw, N. G. and i Martin, X. S. (1992). Capital mobility in Neoclas-sical models of growth, NBER Working Paper No. 4206, Cambridge, Massachusetts:National Bureau of Economic Research.

Barth, J., Caprio, G. and Levine, R. (2004). Bank regulation and supervision: What worksbest?, Journal of Financial Intermediation 13(2): 205–248.

Barth, J. R., Caprio, G. and Levine, R. (2006). Rethinking bank regulation: Till angelsgovern, Cambridge University Press.

Barth, J. R., Caprio Jr, G. and Levine, R. (2001). Banking systems around the globe: Doregulation and ownership affect performance and stability?, Prudential supervision:What works and what doesn’t, University of Chicago Press, pp. 31–96.

Bartholomew, P. F. and Whalen, G. (1995). Fundamentals of systemic risk, Research inFinancial Services: Banking, Financial Markets, and Systemic Risk 7: 3–18.

Baumol, W. J. (1986). Productivity growth, convergence, and welfare: What the long-rundata show, The American Economic Review 76(5): 1072–1085.

Beck, T., Demirguc-Kunt, A. and Levine, R. (2009). Financial institutions and marketsacross countries and over time - data and analysis, Policy Research Working PaperSeries 4943, The World Bank.

Beck, T. and Levine, R. (2004). Stock markets, banks, and growth: Panel evidence,Journal of Banking & Finance 28(3): 423–442.

Beck, T., Levine, R. and Loayza, N. (2000). Finance and the sources of growth, Journalof Financial Economics 58(1): 261–300.

Becker, G. S., Murphy, K. M. and Tamura, R. (1994). Human capital, fertility, and eco-nomic growth, Human Capital: A Theoretical and Empirical Analysis with SpecialReference to Education (3rd Edition), The University of Chicago Press, pp. 323–350.

133

Bekaert, G., Harvey, C. and Lundblad, C. (2011). Financial openness and productivity,World Development 39(1): 1–19.

Bekaert, G., Harvey, C. R. and Lundblad, C. (2005). Does financial liberalization spurgrowth?, Journal of Financial Economics 77(1): 3–55.

Bencivenga, V. and Smith, B. (1991). Financial intermediation and endogenous growth,The Review of Economic Studies 58(2): 195–209.

Benhabib, J. and Spiegel, M. M. (1994). The role of human capital in economic devel-opment evidence from aggregate cross-country data, Journal of Monetary Economics34(2): 143–173.

Bernanke, B. S. (1983). Nonmonetary effects of the financial crisis in the propagation ofthe great depression, The American Economic Review 73(3): 257–276.

Bhagwati, J. (1982). Import competition and response, University of Chicago Press.

Bhagwati, J. (1998). Capital myth-the difference between trade in widgets and dollars,Foreign Affairs 77: 7–12.

Blecker, R. A. (1999). Taming global finance: A better architecture for growth and equity,Economic Policy Institute, Washington DC.

Blundell, R. and Bond, S. (1998). Initial conditions and moment restrictions in dynamicpanel data models, Journal of Econometrics 87(1): 115–143.

Bond, S. R., Hoeffler, A. and Temple, J. (2001). GMM Estimation of Empirical GrowthModels., Nuffield College (University of Oxford).

Bordo, M., Eichengreen, B., Klingebiel, D. and Martinez-Peria, M. S. (2001). Is the crisisproblem growing more severe?, Economic policy 16(32): 51–82.

Broner, F. A. and Ventura, J. (2010). Rethinking the effects of financial liberalization,Working Paper 16640, National Bureau of Economic Research.

Brunnermeier, M. K. and Pedersen, L. H. (2009). Market liquidity and funding liquidity,Review of Financial studies 22(6): 2201–2238.

Buch, C. M., Koch, C. T. and Koetter, M. (2013). Do banks benefit from internationaliza-tion? Revisiting the market power–risk nexus, Review of Finance 17(4): 1401–1435.

Buera, F. J., Kaboski, J. and Shin, Y. (2009). Finance and development: A tale of twosectors, NBER Working Paper No. 14914, Cambridge, Massachusetts: National Bureauof Economic Research.

Byrne, J. P. and Davis, E. P. (2003). Financial structure: An investigation of sectoralbalance sheets in the G-7, Vol. 43, Cambridge University Press.

134

Calomiris, C. W. and Mason, J. R. (1994). Contagion and bank failures during the greatdepression: The June 1932 chicago banking panic, NBER Working Paper No. 4934,Cambridge, Massachusetts: National Bureau of Economic Research.

Calvo, G. (1989). Incredible reforms, Debt, Stabilization and Development: Essays inMemory of Carlos Diaz-Alejandro, (eds.) G. Calve, R. Findlay, P. Kouri, and J. Bragade Macedo, Oxford: Basil Blackwell .

Calvo, G. A. (1987). On the costs of temporary policy, Journal of Development Eco-nomics 27(1): 245–261.

Calvo, G. A. (1996). Capital flows and macroeconomic management: Tequila lessons,International Journal of Finance & Economics 1(3): 207–223.

Calvo, G. A., Borensztein, E., Masson, P. R. and Kumar, M. S. (1993). Financial sectorreforms and exchange arrangements in eastern europe, IMF Occasional Papers 102,International Monetary Fund.

Calvo, G. A. and Mendoza, E. G. (1996). Mexico’s balance-of-payments crisis: A chron-icle of a death foretold, Journal of International Economics 41(3): 235–264.

Caprio, G. (2003). An update of the caprio-klingebiel database (1996, 1999), World Bankwebsite www1. worldbank. org/finance/html/database_sfd. html .

Caprio, G., Atiyas, I. and Hanson, J. A. (1996). Financial reform: Theory and experience,Cambridge University Press.

Caprio, G., Klingebiel, D., Laeven, L. and Noguera, G. (2005). Banking crisis database,Systemic Financial Crises: Containment and Resolution pp. 307–40.

Caprio, G. and Summers, L. H. (1993). Finance and its reform: Beyond laissez-faire, Vol.1171, World Bank Publications.

Caprio Jr, G., Atiyas, I. and Hanson, J. A. (1994). Financial Reform: Theory and Practice,Cambridge University Press.

Carlin, W. and Mayer, C. (2003). Finance, investment, and growth, Journal of FinancialEconomics 69(1): 191–226.

Cartapanis, A., Dropsy, V. and Mametz, S. (2002). The asian currency crises: Vulnerabil-ity, contagion, or unsustainability, Review of International Economics 10(1): 79–91.

Caselli, F., Esquivel, G. and Lefort, F. (1996). Reopening the convergence debate: A newlook at cross-country growth empirics, Journal of Economic Growth 1(3): 363–389.

Cavallo, A. and Cavallo, E. (2010). Are crises good for long-term growth? The role ofpolitical institutions, Journal of Macroeconomics 32(3): 838–857.

135

Cerra, V. and Saxena, S. (2008). Growth dynamics: The myth of economic recovery, TheAmerican Economic Review 98(1): 439–457.

Chakrabarty, B. and Zhang, G. (2012). Credit contagion channels: Market microstructureevidence from Lehman Brothers’ bankruptcy, Financial Management 41(2): 320–343.

Chamberlain, G. (1979). Analysis of covariance with qualitative data, NBER WorkingPaper No. 325, Cambridge, Massachusetts: National Bureau of Economic Research.

Cho, Y. J. (1988). The effect of financial liberalization on the efficiency of credit alloca-tion: Some evidence from Korea, Journal of Development Economics 29(1): 101–110.

Choksi, A. M. and Papageorgiou, D. (1986). Economic liberalization in developing coun-tries / edited by Armeane M. Choksi and Demetris Papageorgiou, Basil Blackwell Ox-ford, UK ; New York, NY, USA.

Clemens, M. A., Radelet, S., Bhavnani, R. R. and Bazzi, S. (2012). Counting chickenswhen they hatch: Timing and the effects of aid on growth, The Economic Journal122(561): 590–617.

Cojocaru, L., Hoffman, S. D. and Miller, J. B. (2011). Financial development and eco-nomic growth in transition economies: Empirical evidence from the CEE and CIScountries, Working Papers 11-22, University of Delaware, Department of Economics.

Collins, J. (1995). Chile’s free-market miracle: A second look, Food First Books.

Conley, J. P. and Maloney, W. F. (1995). Optimal sequencing of credible reforms withuncertain outcomes, Journal of Development Economics 48(1): 151–166.

Cooley, T., Marimon, R. and Quadrini, V. (2004). Aggregate consequences of limitedcontract enforceability, Journal of Political Economy 112(4): 817–847.

Crafts, N. and Fearon, P. (2013). The Great Depression of the 1930s: Lessons for Today,Oxford University Press.

Crotty, J. (2002). From the golden age to global neoliberalism, Seeking shelter on thePacific Rim: Financial globalization, social change, and the housing market pp. 21–41.

Crotty, J. (2009). Structural causes of the global financial crisis: A critical assessment ofthe ‘New Financial Architecture’, Cambridge Journal of Economics 33(4): 563–580.

Darvas, Z. (2012). Real effective exchange rates for 178 countries: A new database,Bruegel Working Papeer 2012/06, Bruegel.

Davis, G. and Kanago, B. (1996). On measuring the effect of inflation uncertainty on realgnp growth, Oxford Economic Papers 48(1): 163–175.

136

De Gregorio, J. and Guidotti, P. (1995). Financial development and economic growth,World Development 23(3): 433–448.

De la Torre, A., Ize, A. and Schmukler, S. L. (2011). Financial development in LatinAmerica and the Caribbean: The road ahead, The World Bank, Washington DC.

Deidda, L. and Fattouh, B. (2002). Non-linearity between finance and growth, EconomicsLetters 74(3): 339–345.

Demetriades, P. and Hussein, K. (1996). Does financial development cause economicgrowth? time-series evidence from 16 countries, Journal of Development Economics51(2): 387–411.

Demetriades, P. and Rousseau, P. L. (2011). The changing face of financial development,MMF Conference, University of Birmingham.

Demirgüç-Kunt, A. and Detragiache, E. (1998a). The determinants of banking crisesin developing and developed countries, Staff Papers-International Monetary Fund45(1): 81–109.

Demirgüç-Kunt, A. and Detragiache, E. (1998b). Financial liberalization and finan-cial fragility, In Proceedings of Annual Bank Conference on Development Economics.Washington, DC, 20-21 April.

Demirguc-Kunt, A. and Levine, R. (2004). Financial structure and economic growth: Across-country comparison of banks, markets, and development, MIT press.

Devereux, M. B. and Smith, G. W. (1994). International risk sharing and economicgrowth, International Economic Review 35(3): 535–550.

Diaz-Alejandro, C. (1985). Good-bye financial repression, hello financial crash, Journalof Development Economics 19(1): 1–24.

Dooley, M. P. (1995). A survey of academic literature on controls over international cap-ital transactions, NBER Working Paper No. 5352, Cambridge, Massachusetts: NationalBureau of Economic Research.

Ductor, L. and Grechyna, D. (2011). Excess financial development and economic growth,WP Fama, E. and K. French (2010):âAIJLuck versus skill in the cross section of mutualfund a estimates,âAI Journal of Finance 65(5).

Easterly, W., Islam, R. and Stiglitz, J. E. (2001). Shaken and stirred: Explaining growthvolatility, Annual World Bank Conference on Development Economics, pp. 191–211.

Easterly, W., Levine, R. and Roodman, D. (2003). New data, new doubts: A commenton burnside and dollar’s" aid, policies, and growth"(2000), NBER Working Paper No.9846, Cambridge, Massachusetts: National Bureau of Economic Research.

137

Easterly, W., Loayza, N. and Montiel, P. (1997). Has Latin America’s post-reform growthbeen disappointing?, Journal of International Economics 43(3-4): 287–311.

Ebbinghaus, B. and Manow, P. (2004). Comparing welfare capitalism: Social policy andpolitical economy in Europe, Japan and the USA, Routledge.

Echeverria, C., Darbar, S. M. and Johnston, R. B. (2004). Sequencing capital accountliberalization-lessons from the experiences in Chile, Indonesia, Korea, and Thailand,International Monetary Fund 97/157, International Monetary Fund.

Edison, H. J., Klein, M. W., Ricci, L. and Sloek, T. (2002). Capital account liberalizationand economic performance: Survey and synthesis, NBER Working Paper No. 9100,Cambridge, Massachusetts: National Bureau of Economic Research.

Edison, H. J. and Warnock, F. E. (2003). A simple measure of the intensity of capitalcontrols, Journal of Empirical Finance 10(1): 81–103.

Edwards, S. (1984). The order of liberalization of the external sector in developing coun-tries, International Finance Section, Department of Economics, Princeton UniversityPrinceton, New Jersey.

Edwards, S. (1990). The sequencing of economic reform: Analytical issues and lessonsfrom Latin American experiences, The World Economy 13(1): 1–14.

Edwards, S. (2001). Capital mobility and economic performance: Are emergingeconomies different?, Working Paper 8076, National Bureau of Economic Research.

Edwards, S. (2009). Sequencing of reforms, financial globalization, and macroeconomicvulnerability, Journal of the Japanese and International Economies 23(2): 131–148.

Eichengreen, B. (2001). Capital account liberalization: What do cross-country studiestell us?, The World Bank Economic Review 15(3): 341–365.

Eichengreen, B. J., Mussa, M. and Dell’Ariccia, G. (1998). Capital account liberalization:Theoretical and practical aspects, IMF Occasional Papers 172, International MonetaryFund.

Favarra, G. (2003). An Empirical Reassessment of the Relationship Between Finance andGrowth, IMF Working Papers 03/123, International Monetary Fund.

Federici, D. and Caprioli, F. (2009). Financial development and growth: An empiricalanalysis, Economic Modelling 26(2): 285–294.

Fernández-Arias, E. and Montiel, P. J. (1997). Reform and growth in Latin America: Allpain, no gain?, IADB, Research Department Working Paper (351).

Fischer, B. (1993). Success and pitfalls with financial reforms in developing countries,Savings and Development 17(2): 111–135.

138

Fischer, S. (1979). Anticipations and the nonneutrality of money, The Journal of PoliticalEconomy 87(2): 225–252.

Forbes, K., Fratzschner, M. and Straub, R. (2013). Capitals controls and macroprudentialmeasures: What are they good for?, Technical report, CEPR Discussion Paper No.DP9798.

Fry, M. (1997). In favour of financial liberalisation, The Economic Journal107(442): 754–770.

Galbis, V. (1994). Sequencing of financial sector reforms-a review, IMF Working Papers94/101, International Monetary Fund.

Galindo, A. J. and Maloney, W. F. (2002). Second moments in speculative attack models:Panel evidence, Journal of International Economics 56(1): 97–129.

Gelb, A. H. (1989). Financial policies, growth, and efficiency, Policy Research WorkingPaper Series 202, The World Bank.

Gibson, H. D. and Tsakalotos, E. (1994). The scope and limits of financial liberalisa-tion in developing countries: A critical survey, The Journal of Development Studies30(3): 578–628.

Giesecke, K. (2004). Correlated default with incomplete information, Journal of Banking& Finance 28(7): 1521–1545.

Glick, R. and Hutchison, M. (2005). Capital controls and exchange rate instability indeveloping economies, Journal of International Money and Finance 24(3): 387 – 412.

Glick, R., Moreno, R. and Spiegel, M. M. (2001). Financial crises in emerging markets,number 2001-2007, Cambridge University Press.

Goldsmith, R. W. (1969). Financial structure and development, Vol. 1, Yale universitypress New Haven.

Grabel, l. (1995). Speculation-led economic development: A post-keynesian interpreta-tion of financial liberalization programmes in the third world, International Review ofApplied Economics 9(2): 127–149.

Grant, J. K. (2009). What the financial services industry puts together let no person putasunder: How the Gramm-Leach-Bliley Act contributed to the 2008-2009 Americancapital markets crisis, Albany Law Review 73: 371.

Green, R. K., Malpezzi, S. and Mayo, S. K. (2005). Metropolitan-specific estimates ofthe price elasticity of supply of housing, and their sources, American Economic Review95(2): 334–339.

Greenaway, D., Morgan, W. and Wright, P. (1997). Trade liberalization and growth indeveloping countries: Some new evidence, World Development 25(11): 1885–1892.

139

Greenwood, J. and Jovanovic, B. (1989). Financial development, growth, and the distribu-tion of income, NBER Working Paper No. 3189, Cambridge, Massachusetts: NationalBureau of Economic Research.

Griffith-Jones, S. and Tobin, J. (1998). Global capital flows: Should they be regulated?,St. Martin’s Press.

Guariglia, A. and Poncet, S. (2008). Could financial distortions be no impediment toeconomic growth after all? Evidence from China, Journal of Comparative Economics36(4): 633–657.

Hadi, A. S. (1992). Identifying multiple outliers in multivariate data, Journal of the RoyalStatistical Society. Series B (Methodological) 54(3): 761–771.

Hadi, A. S., Imon, A. and Werner, M. (2009). Detection of outliers, Wiley Interdisci-plinary Reviews: Computational Statistics 1(1): 57–70.

Hall, M. (2003). The international handbook on financial reform, Edward Elgar Publish-ing.

Han Kim, E. and Singal, V. (2000). Stock market openings: Experience of emergingeconomies, The Journal of Business 73(1): 25–66.

Hansen, L. P. (1982). Large sample properties of generalized method of moments estima-tors, Econometrica 50(4): 1029–1054.

Harrison, L. E. and Huntington, S. P. (2000). Culture matters: How values shape humanprogress, Basic Books.

Hellmann, T., Murdock, K. and Stiglitz, J. (1994). Addressing moral hazard in banking:Deposit rate controls vs. capital requirements, Unpublished manuscript. Washington:World Bank .

Hellmann, T., Murdock, K. and Stiglitz, J. (2000). Liberalization, moral hazard in bank-ing, and prudential regulation: Are capital requirements enough?, American EconomicReview 90(1): 147–165.

Hellmann, T., Murdock, K. and Stiglitz, J. E. (1995). Financial Restraint: Towards a NewParadigm, number 1355, Graduate School of Business, Stanford University.

Hicks, J. R. (1969). A theory of economic history, OUP Catalogue .

Hill, T. (1964). Growth and investment according to international comparisons, The Eco-nomic Journal 74(294): 287–304.

Hnatkovska, V. and Loayza, N. (2004). Volatility and growth, Policy Research WorkingPaper Series 3184, The World Bank.

140

Holtz-Eakin, D., Newey, W. and Rosen, H. (1988). Estimating vector autoregressionswith panel data, Econometrica: Journal of the Econometric Society 56(6): 1371–1395.

Holzmann, R. (1997). Pension reform, financial market development, and economicgrowth: Preliminary evidence from chile, Staff Papers-International Monetary Fund44(2): 149–178.

Hornbeck, J. (2009). Financial regulation and oversight: Latin American financial crisesand reform lessons from chile, Congressional Research Service pp. 1–13.

Huang, Y. (2011). Determinants of Financial Development, Palgrave Macmillan.

Hutchinson, M. and Glick, R. (2000). Banking and Currency Crises: How Common areTwins?, University of California, Santa Cruz.

Jappelli, T. and Pagano, M. (2000). Saving, growth, and liquidity constraints, The Quar-terly Journal of Economics 109(1): 83–109.

Jarrow, R. A. (2014). Financial crises and economic growth, The Quarterly Review ofEconomics and Finance 54(2): 194–207.

Jayaratne, J. and Strahan, P. (1996). The finance-growth nexus: Evidence from bankbranch deregulation, The Quarterly Journal of Economics 111(3): 639–670.

Johnston, R. B. (1998). Sequencing capital account liberalization, Finance and Develop-ment 35: 20–23.

Johnston, R. B. and Sundararajan, V. (1999). Managing Financial Sector Liberalization:An Overview, In Johnston, R. B. and V. Sundararajan.(eds). Sequencing Financial Sec-tor Reforms: Country Experiences and Issues. IMF.

Jones, L. E. and Manuelli, R. E. (2005). Neoclassical models of endogenous growth: Theeffects of fiscal policy, innovation and fluctuations, Handbook of Economic Growth1: 13–65.

Judson, R. and Orphanides, A. (1999). Inflation, volatility and growth, InternationalFinance 2(1): 117–138.

Kaminsky, G. L. (2006). Currency crises: Are they all the same?, Journal of InternationalMoney and Finance 25(3): 503 – 527.

Kaminsky, G. L. and Schmukler, S. L. (2008). Short-run pain, long-run gain: Financialliberalization and stock market cycles, Review of Finance 12(2): 253–292.

Kaminsky, G., Lizondo, S. and Reinhart, C. M. (1998). Leading indicators of currencycrises, Staff Papers-International Monetary Fund 45(1): 1–48.

Kaminsky, G. and Reinhart, C. (1999). The twin crises: The causes of banking andbalance-of-payments problems, American Economic Review 89(3): 473–500.

141

Kaminsky, G. and Schmukler, S. (2003). Short-run pain, long-run gain: The effects offinancial liberalization, NBER Working Paper No. 9787, Cambridge, Massachusetts:National Bureau of Economic Research.

Keeley, M. C. (1990). Deposit insurance, risk, and market power in banking, AmericanEconomic Review 80(5): 1183–1200.

Keynes, J. (1930). A Treatise on Money: Volume 1: The Pure Theory of Money, Macmillan& Company.

Khadraoui, N. and Smida, M. (2012). Financial development and economic growth:Static and dynamic panel data analysis, International Journal of Economics and Fi-nance 4(5): 94–104.

King, R. and Levine, R. (1993a). Finance and growth: Schumpeter might be right, TheQuarterly Journal of Economics 108(3): 717–737.

King, R. and Levine, R. (1993b). Finance, entrepreneurship and growth, Journal of Mon-etary Economics 32(3): 513–542.

Klein, M. and Olivei, G. (2008). Capital account liberalization, financial depth, and eco-nomic growth, Journal of International Money and Finance 27(6): 861–875.

Kormendi, R. C. and Meguire, P. G. (1985). Macroeconomic determinants of growth:Cross-country evidence, Journal of Monetary Economics 16(2): 141–163.

Kose, M. A., Prasad, E., Rogoff, K. S. and Wei, S.-J. (2006). Financial globalization:A reappraisal, NBER Working Paper No. 12484, Cambridge, Massachusetts: NationalBureau of Economic Research.

Kose, M. A., Prasad, E. S. and Terrones, M. E. (2003). Financial integration and macroe-conomic volatility, IMF Staff Papers 50: 119–142.

Krugman, P., Baldwin, R. E. and Branson, W. H. (1993). International finance and eco-nomic development, in A. Giovannini (ed.), Finance and development: issues and ex-perience, Cambridge University Press, pp. 11–28.

La Porta, R., Lopez-de Silanes, F. and Shleifer, A. (2002). Government ownership ofbanks, The Journal of Finance 57(1): 265–301.

Laeven, L., Igan, D., Claessens, S. and Dell’Ariccia, G. (2010). Lessons and policyimplications from the global financial crisis, IMF Working Papers 10/44, InternationalMonetary Fund.

Laeven, L. and Valencia, F. (2008). Systemic banking crises: A new database, IMFWorking Papers 08/224, International Monetary Fund.

142

Lane, P. R. and Milesi-Ferretti, G. M. (2007). The external wealth of nations mark ii:Revised and extended estimates of foreign assets and liabilities, 1970–2004, Journal ofInternational Economics 73(2): 223–250.

Lensink, R. (1996). The allocative efficiency of the formal versus the informal financialsector, Applied Economics Letters 3(3): 163–165.

Levi, M. D. and Makin, J. H. (1980). Inflation uncertainty and the Phillips curve: Someempirical evidence, The American Economic Review 70(5): 1022–1027.

Levine, R. (1997). Financial development and economic growth: Views and agenda,Journal of Economic Literature 35(2): 688–726.

Levine, R. (1998). The legal environment, banks, and long-run economic growth, Journalof Money, Credit and Banking 30(3): 596–613.

Levine, R. (1999). Financial development and growth: Where do we stand?, Estudios deEconomia 26(2): 113–136.

Levine, R. (2001). International financial liberalization and economic growth, Review ofInternational economics 9(4): 688–702.

Levine, R. (2011). Regulating finance and regulators to promote growth, Proceedings -Economic Policy Symposium - Jackson Hole pp. 271–311.

Levine, R., Loayza, N. and Beck, T. (2000). Financial intermediation and growth: Causal-ity and causes, Journal of Monetary Economics 46(1): 31–78.

Levine, R. and Zervos, S. (1998). Stock markets, banks, and economic growth, AmericanEconomic Review 88(3): 537–558.

Li, H. and Zou, H.-f. (2002). Inflation, growth, and income distribution: A cross-countrystudy, Annals of Economics and Finance 3(1): 85–101.

Lipinska, A. and De Paoli, B. (2013). Capital controls: A normative analysis, 2013Meeting Papers, number 861, Society for Economic Dynamics.

Loayza, N. and Ranciere, R. (2006). Financial development, financial fragility, andgrowth, Journal of Money, Credit and Banking 38(4): 1051–1076.

Lucas, R. (1988). On the mechanics of economic development, Journal of MonetaryEconomics 22(1): 3–42.

Luis, C., Milesi-Ferretti, G.-M. et al. (2013). External liabilities and crises, IMF WorkingPapers 13/113, International Monetary Fund.

Mankiw, N. G., Romer, D. and Weil, D. N. (1992). A contribution to the empirics ofeconomic growth, The Quarterly Journal of Economics 107(2): 407–437.

143

Massad, C. (1998). The liberalization of the capital account: Chile in the 1990s, In: Fis-cher, S. , Should the IMF pursue capital-account liberalization?, Essays in InternationalFinance No. 207, International Finance Section, Princeton University, Pinceton, NJ.

McCleskey, S. (2010). When Free Markets Fail: Saving the Market when it Can’t SaveItself, John Wiley & Sons.

McKinnon, R. I. (1973). Money and capital in economic development, Brookings Insti-tution Press.

McKinnon, R. I. (1982). The order of economic liberalization: Lessons from chile andargentina, Carnegie-Rochester Conference Series on Public Policy, Vol. 17, North-Holland, pp. 159–186.

Michaely, M., Choksi, A. M. and Papageorgiou, D. (1991). Liberalizing foreign trade,Basil Blackwell.

Mishkin, F. S., Stern, G. and Feldman, R. (2006). How big a problem is too big to fail? areview of gary stern and ron feldman’s" too big to fail: The hazards of bank bailouts",Journal of Economic Literature 44(4): 988–1004.

Móczár, J. (2010). Anatomy and lessons of the global financial crisis, Public FinanceQuarterly 55(4): 753–775.

Mullineaux, D. J. (1980). Unemployment, industrial production, and inflation uncertaintyin the United States, The Review of Economics and Statistics 62(2): 163–169.

Nathan, H. S. K., Mishra, S. and Reddy, B. S. (2008). An alternative approach to measureHDI, Technical report, Indira Gandhi Institute of Development Research, Mumbai,India.

Neusser, K. and Kugler, M. (1998). Manufacturing growth and financial development:Evidence from oecd countries, Review of Economics and Statistics 80(4): 638–646.

Noy, I. (2004). Financial liberalization, prudential supervision, and the onset of bankingcrises, Emerging Markets Review 5(3): 341–359.

Odedokun, M. (1996). Alternative econometric approaches for analysing the role of thefinancial sector in economic growth: Time-series evidence from ldcs, Journal of De-velopment Economics 50(1): 119–146.

Ostry, J. D., Ghosh, A. R., Chamon, M. and Qureshi, M. S. (2012). Tools for manag-ing financial-stability risks from capital inflows, Journal of International Economics88(2): 407–421.

Pagano, M. (1993). Financial markets and growth: An overview, European EconomicReview 37(2-3): 613–622.

144

Pollin, R. and Zhu, A. (2006). Inflation and economic growth: A cross-country nonlinearanalysis, Journal of Post Keynesian Economics 28(4): 593–614.

Popov, V. (2000). Shock therapy versus gradualism: The end of the debate (explaining themagnitude of transformational recession), Comparative Economic Studies 42(1): 1–58.

Porta, R., Lopez-de Silane, F., Shleifer, A. and Vishny, R. (1996). Law and finance, NBERWorking Paper No. 5661, Cambridge, Massachusetts: National Bureau of EconomicResearch.

Pritchett, L. (2001). Where has all the education gone?, The World Bank Economic Review15(3): 367–391.

Qin, X. and Luo, C. (2014). Capital account openness and early warning system forbanking crises in {G20} countries, Economic Modelling 39(0): 190 – 194.

Rajan, R. G. (2006). Has finance made the world riskier?, European Financial Manage-ment 12(4): 499–533.

Rajan, R. G. and Subramanian, A. (2005). What undermines aid’s impact on growth?,NBER Working Paper No. 11657, Cambridge, Massachusetts: National Bureau of Eco-nomic Research.

Rajan, R. G. and Zingales, L. (1996). Financial dependence and growth, NBER WorkingPaper No. 5758, Cambridge, Massachusetts: National Bureau of Economic Research.

Ranciere, R., Tornell, A. and Westermann, F. (2006). Decomposing the effects of financialliberalization: Crises vs. growth, Journal of Banking & Finance 30(12): 3331–3348.

Reinhart, C. M. and Rogoff, K. (2009). This time is different: Eight centuries of financialfolly, Princeton University Press.

Rioja, F. and Valev, N. (2004). Does one size fit all?: A reexamination of the finance andgrowth relationship, Journal of Development Economics 74(2): 429–447.

Rioja, F. and Valev, N. (2007). Finance and the sources of growth at various stages ofeconomic development, Economic Inquiry 42(1): 127–140.

Robinson, J. (1952). The rate of interest and other essays, Macmillan London.

Rodrik, D. (2005). Why we learn nothing from regressing economic growth on policies,Technical report, Mimeo, Kennedy School of Government.

Rodrik, D. and Velasco, A. (1999). Short-term capital flows, NBER Working Paper No.7364, Cambridge, Massachusetts: National Bureau of Economic Research.

Rodrik, D. et al. (1998). Who needs capital-account convertibility?, Essays in interna-tional finance pp. 55–65.

145

Roe, M. and Siegel, J. (2008). Political instability’s impact on financial development,Harvard Law and Economics Discussion Paper (570).

Roodman, D. (2006). How to do xtabond2: An introduction to difference and systemGMM in STATA, Center for Global Development working paper (103).

Roodman, D. (2007). The anarchy of numbers: Aid, development, and cross-countryempirics, The World Bank Economic Review 21(2): 255–277.

Roodman, D. (2009). A note on the theme of too many instruments, Oxford Bulletin ofEconomics and Statistics 71(1): 135–158.

Roubini, N. and Sala-i Martin, X. (1992). Financial repression and economic growth,Journal of Development Economics 39(1): 5–30.

Rousseau, P. L. and Wachtel, P. (2011). What is happening to the impact of financialdeepening on economic growth?, Economic Inquiry 49(1): 276–288.

Rousseau, P. and Wachtel, P. (1998). Financial intermediation and economic performance:Historical evidence from five industrialized countries, Journal of Money, Credit andBanking 30(4): 657–678.

Rousseau, P. and Wachtel, P. (2000). Equity markets and growth: Cross-country evidenceon timing and outcomes, 1980–1995, Journal of Banking & Finance 24(12): 1933–1957.

Sarma, M. (2008). Index of financial inclusion, Technical report, ICRIER Working PaperNo 215.

Schumpeter, J. (1934). The theory of economic development: An inquiry into profits,capital, credit, interest and the business cycle (R. Opie, Trans.), Harvard UniversityPress, Cambridge, MA.

Shan, J., Morris, A. and Sun, F. (2003). Financial development and economic growth: Anegg-and-chicken problem?, Review of International Economics 9(3): 443–454.

Shaw, E. S. (1973). Financial deepening in economic development, Vol. 39, Oxford Uni-versity Press New York.

Sinha, A., Kumar, R. and Dhal, S. (2012). Financial sector regulation and implicationsfor growth, BIS Papers (62).

Soto, M. (2009). System GMM estimation with a small sample, Working Papers 395,Barcelona Graduate School of Economics.

Stern, G. H. and Feldman, R. J. (2004). Too big to fail: The hazards of bank bailouts,Brookings Institution Press.

146

Stiglitz, J. (2000a). What I learned at the world economic crisis, Globalization and thepoor: Exploitation or Equalizer pp. 195–204.

Stiglitz, J. E. (2000b). Capital market liberalization, economic growth, and instability,World Development 28(6): 1075–1086.

Stiglitz, J. E. (2002). Globalization and its Discontents, New York.

Stiglitz, J. E. (2004). Capital-market liberalization, globalization, and the IMF, OxfordReview of Economic Policy 20(1): 57–71.

Stulz, R. (1999). Globalization of equity markets and the cost of capital, NBER WorkingPaper No. 7021, Cambridge, Massachusetts: National Bureau of Economic Research.

Taylor, J. (2009). The financial crisis and the policy responses: An empirical analysisof what went wrong, NBER Working Paper No. 14631, Cambridge, Massachusetts:National Bureau of Economic Research.

Tressel, T. and Detragiache, E. (2008). Do financial sector reforms lead to financialdevelopment? Evidence from a new dataset, IMF Working Papers 08/265, InternationalMonetary Fund.

Van Wijnbergen, S. (1983). Credit policy, inflation and growth in a financially repressedeconomy, Journal of Development Economics 13(1): 45–65.

Velasco, A. (1991). Liberalization, crisis, intervention: The chilean financial system,1975–85, Banking Crises: Cases and Issues, edited by V. Sundararajan and Tomas J.T. Balino. Washington DC: International Monetary Fund .

Vyas, S. and Kumaranayake, L. (2006). Constructing socio-economic status indices: Howto use principal components analysis, Health policy and planning 21(6): 459–468.

Wacziarg, R. and Welch, K. H. (2008). Trade liberalization and growth: New evidence,The World Bank Economic Review 22(2): 187–231.

Wei, S.-J. (1997). Gradualism versus big bang: Speed and sustainability of reforms,Canadian Journal of Economics 30(4b): 1234–1247.

Weller, C. (2001). Financial crises after financial liberalisation: Exceptional circum-stances or structural weakness?, Journal of Development Studies 38(1): 98–127.

Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step gmm estimators, Journal of Econometrics 126(1): 25–51.

World Bank (1993). East Asian Miracle: Economic Growth and Public Policy, WorldBank Group.

World Bank (2012). The financial development report 2012, Technical report.

147

Xu, Z. (2000). Financial development, investment, and economic growth, Economic In-quiry 38(2): 331–344.

Zeleny, M. (1974). A concept of compromise solutions and the method of the displacedideal, Computers & Operations Research 1(3): 479–496.

essays on financial liberalisation, financial crises and

Documents