ON THE DETERMINANTS OF
FINANCIAL REFORM
AN EMPIRICAL POLITICAL ECONOMY APPROACH
Master's Thesis in Economics
Student: Marcel Klok Student Number: 1481983 Year of Publication: 2010
Supervisor: Prof. dr. J. de Haan
MSc Economics
Faculty of Economics and Business Rijksuniversiteit Groningen
JEL codes: G18, G28, P16
Keywords: Financial reform, political economy
1
TABLE OF CONTENT
TABLE OF CONTENT ... 1
I. INTRODUCTION ... 3
1.INTRODUCTION TO THE THESIS AND ITS OUTLINE ... 3
2.INTRODUCTION TO THE TOPIC:FINANCIAL LIBERALISATION ... 4
II. RESEARCH QUESTION ... 7
III. LITERATURE REVIEW ... 8
1.INTRODUCTION TO THE LITERATURE REVIEW ... 8
2.THEORY AND THEORETICAL LITERATURE ... 9
3.EMPIRICAL LITERATURE ... 32
IV. TESTABLE HYPOTHESES AND VARIABLES ... 39
1.BENCHMARK:PREVIOUSLY TESTED HYPOTHESES AND RELEVANT VARIABLES ... 40
2.BEYOND THE BENCHMARK:ADDED HYPOTHESES AND VARIABLES ... 50
V. FACTOR ANALYSIS ... 61
1.THE METHOD OF FACTOR ANALYSIS ... 61
2.FACTOR ANALYSIS RESULTS ... 67
3.FINANCIAL LIBERTY AND LIBERALISATION INDICES ... 73
VI. METHODOLOGY ... 81
1.ECONOMETRIC MODELS ... 81
2.BENCHMARK SPECIFICATIONS INSPIRED UPON ABIAD AND MODY (2005) ... 81
3.ECONOMETRIC METHODS ... 84
VII. EMPIRICAL RESULTS ... 89
1.BENCHMARK MODEL RESULTS ... 89
2.RECAPITULATION ON RESULTS OF BENCHMARK MODELS ... 99
3.ROBUSTNESS CHECKS ON CRISIS VARIABLES ... 101
4.RESULTS BEYOND THE BENCHMARK ... 104
VIII. CONCLUSION ... 113
2
2.LIMITATIONS AND FUTURE RESEARCH ... 114
REFERENCES ... 117
APPENDICES ... 124
APPENDIX 1:COUNTRY LIST AND COUNTRY CATEGORISATION... 124
APPENDIX 2:VARIABLES AND SOURCES ... 125
APPENDIX 3:QUESTIONS GUIDING CODING OF SUBINDICES ... 128
APPENDIX 4:CODING OF TOPTAX ... 129
APPENDIX 5:CODING OF PRESIDENT ... 130
APPENDIX 6:CODING OF LIEC ... 130
APPENDIX 7:CODING OF EIEC ... 131
3
I.INTRODUCTION
1.Introduction to the thesis and its outline
In front of you lies the product of the project that should finalise my education in Economics at the University of Groningen, my Master’s Thesis. The topic of this thesis is financial reform. Before providing an overview of the content of this thesis, I would like to express my gratitude to the following people who have been helpful one way or another: Rob Alessie, Jakob de Haan, Yongfu Huang, Richard Jong-A-Ping, Jeroen Klomp, and Sunčica Vujić. It almost goes without saying that I remain responsible for the mistakes and anomalies that this thesis contains, even though I suggest otherwise by continuing to write in the plural form.
The outline of this thesis is as follows. After this brief introduction to the thesis, we will provide a short introduction on the importance of finance to illustrate the relevance of zooming in on financial markets and more specifically on the observed trends in financial reform. Second, we will introduce our research question in chapter II.
Third, we will provide a literature review in chapter III. This review contains two sections: a section on our interpretation of the theoretical literature and another on empirical literature. The literature on the liberalisation of financial markets falls within the broader literature on economic reform. We will primarily tap into that literature. Our theoretical focus will more be on the reform issue of financial liberalisation than on the financial side of it. We will however try to make the connection with financial reform. When discussing the empirical literature, we will take the reverse position. We will then solely focus on the empirics of financial liberalisation.
In chapter IV we move on to introducing and criticising data, which serve to operationalise a range of hypotheses. That is because our research is empirical: We will perform an econometric analysis. The novelty of our research is threefold: First, we will test a few hypotheses that according to our knowledge have not been tested before. Second, we will use a dataset with a wide coverage that has not yet been used for this purpose. Third, rather than using an existing indicator as our variable of interest, we will factor analyse a set of indicators each capturing a limited aspect of financial liberty, which as a result together constitute one indicator of financial liberty.
4 The following chapter VI introduces the research methodology that is unrelated to factor analysis. It deals with the econometric model and method.
Chapter VII presents our empirical results. The results will primarily be contrasted with the results of what we believe will turn into an important reference in the literature on financial liberalisation, Abiad and Mody (2005). The final chapter VIII will serve to conclude the thesis.
2.Introduction to the topic: Financial liberalisation
When one makes the step from the reality of an economic textbook to the outside world, one instantly observes that markets contain frictions. Examples are the costs of information and transactions. Financial services may reduce these frictions.
It is therefore that besides real factors financial services can be of importance for the accumulation of physical and human capital, as well as for the development of total factor productivity. In fact, financial services such as monitoring services may even have a positive effect on productive efficiency. A well-developed financial system thus provides important functions to an economy.
Levine (2004) recognises five concrete functions. First, it produces information ex ante about possible investments and allocates capital. Second, it monitors investments and exerts corporate governance after providing finance. Third, it facilitates the trading, diversification, and management of risk. Fourth, it mobilises and pools savings. Fifth, it eases the exchange of goods and services.
Facilitating the development of the financial sector and its services is one of the means by which governments aim to spur economic growth. Financial liberty is sometimes unjustly treated as a synonym for financial development.1 Some degree of financial liberty may be a necessary condition for financial development. Most likely it however is not a sufficient condition. In the same way is the availability of finance a necessary and not a sufficient condition for economic development.
1
5 Nevertheless, the theoretical basis for the importance of finance on economic growth is broad. More developed financial sectors may contribute to higher allocative efficiency as well as to dynamic efficiency. Therefore one would expect all countries to have a financial system that faces few constraints. Yet, some countries chose to highly regulate and control the financial sector.
Abiad et al. (2008) argue that until the 1980s the financial sector was among the sectors in which government intervention was most visible. They moreover showed that liberalisation streaks of individual countries were generally preceded by a long period of status quo maintenance: The level of liberalisation was low and remained constant. But Abiad and Mody (2005) as well as Abiad et al. (2008) also showed that over the last few decades of the twentieth century many countries have liberalised their financial markets.
In an attempt to explain the initial preference for government control on the economy, Anne Krueger (1993, quoted in Rodrik, 1996, p.12) emphasises that there had been “a strong emphasis on the primacy of market imperfections.” If one accepts this point of view, it is only natural that a large a large degree of governmental interference was accepted. Governmental interference can correct market imperfections. It generally takes a long time before doctrines are set aside. Therefore it is possible to explain the occurrence of a long period in which the status quo of financial repression was maintained based on doctrines.
Good economic policy should also be good politics in the sense that politicians get rewarded by the electorate for its favourable economic outcomes (Rodrik, 1996). At the end of the 1980s, striving for financial liberty was clearly seen by economists as a good policy, as it became part of the Washington consensus (Williamson, 2000). A change of doctrine could therefore explain the rising enthusiasm to liberalise that was observed in the 1980s and 1990s.
The public choice movement however emphasises that politicians as well as bureaucrats are self-interested agents. There are many reasons why good economic policy such as financial liberalisation is not good politics from the perspective of the politician. This makes clear that the attachment to a paradigm about what constitutes good economic policy cannot be the only explanation for a long period without reform.
7
II.RESEARCH QUESTION
The central research question of our thesis is as follows: What are the determinants of financial market reform?
When one thinks of determinants of reform, four main topics come forward: the timing, speed, scale and direction of reform. We do not attempt to look at these four topics separately. The reason for this is the fact that it is difficult to fully separate these issues in empirical research. What we will do comes closest to studying the scale of liberalisation. We set a fixed time interval of one year and look to what degree restricting financial market policies have changed. Also the direction of reform will be taken into account, since we allow for positive and negative observations. The timing of reform is implicitly addressed by using the time variation of data in our econometric analysis.
It is not uncommon for scholars to mix up the terminology surrounding financial market liberalisation. Abiad et al. (2008), whose data we will later use, are among the scholars who seem to be haunted by the tendency to e.g. make no clear distinction between financial liberalisation and liberty. For us, financial liberalisation means a reform in the direction of more financial liberty. It therefore is the positive change in financial liberty. Here we define the broad concept of financial liberty as being the degree of absence of government interference that may restrict the professional conduct of players in financial markets.
8
III.LITERATURE REVIEW
1.Introduction to the literature review
In this chapter we will provide an overview of the theories and literature on financial reform. We will first discuss theory and theoretical literature and then continue with summarising the results of empirical research. We do not attempt to provide a full survey of the relevant literature.2 Rodrik (1996) argues that macroeconomic policies aimed at economic stability have different political economy underpinnings than liberalisation policies aimed at structural reform and growth. This suggests that taking the narrow view on structural reform only is theoretically preferable to taking a broad view of economic policy reform in general. This is what we will are trying to do with the choice of our research question: Empirically we restrict ourselves to financial market reform. Nevertheless, we consider it to be too restrictive to always only focus on the literature on financial reform. The theoretical literature on economic reform in general can provide useful insights. Most of what we will discuss will indeed address structural reform rather than reform that may be considered to be aiming at providing economic stability.
Contributions to the theoretical literature often focus on one narrowly defined issue of liberalisation. This especially holds for formal models. The result is a lack of realism. The topic of free trade provides a vivid example: Although a strong consensus exists among economists about the theoretical desirability of free trade, the majority of the electorate as well as politicians often favours protectionism. This may be similar for financial liberty. This is why we consider simple arguments, which are merely inspired upon observations of the outside world, as equally valuable as theoretical predictions arising from formal models. We find plenty of such simple theories in the literature.
One of our limitations is that we will sometimes relate financial reform to factors that correlate with the level of financial liberty rather than necessarily its changes. We justify this by stating that for the period on which we focus, countries tended to start from a level of financial liberty that is not close to the maximally attainable level. Our approach is problematic once a country has reached a high degree of liberty. Now liberalisation is not wanted or beneficial anymore.
2
9 We attempt to keep our discussion of the empirical literature as clean as possible in the sense that we will not discuss literature that addresses the determinants of financial development. Furthermore, we also tend to neglect the literature that solely focuses on the regulation of international capital flows.
2.Theory and theoretical Literature
Many considerations in this section stem from the literature on economic reform. Some theoretical considerations may therefore only sound weakly relevant in the context of financial liberalisation. Financial reform nevertheless remains a subset of economic reform. In order to make theories compatible with our topic, we will sometimes provide an introduction to or an elaboration on the presented literature on economic reform. This means that this section reflects our interpretation of theories of others, not necessarily the interpretation of the scholars themselves.
The political economy approach generally starts with assuming that the agents involved in decision making are rational and self-interested. Also often informational assumptions are necessary. The assumption of perfect information is not uncommon. However, since this thesis addresses an empirical issue, we will also make use of less formal arguments which implicitly rely on assumptions about imperfections, e.g. imperfect information and imperfect alignment of the interests of decision makers and voters.
10 We will start with addressing the so called "war of attrition" model (Alesina and Drazen, 1991). We do so because many of our theories/hypotheses can be explained on the basis of this model. It is mainly relevant for situations in which agents agree that a certain policy should be adopted due to the fact that it is known to be welfare enhancing, but in which they initially disagree on the distribution of the costs of reform. Other models will be presented as well, for instance to explain why agents (fail to) agree on financial reform. Different theories are not necessarily compatible with each other. This chapter will provide an enumeration of possible determinants of financial reform.
2.1.War of attrition
Alesina et al. (2006) explain the timing of reform on the basis of the war of attrition model. This model is a game between interest groups and was originally developed by Alesina and Drazen (1991) for macroeconomic stabilisations. The model can be explained as follows. There is agreement on what reform is beneficial for the public at large. The reform is assumed to necessarily impose ex-post costs on some agents. Agents, which may be political groups, disagree about the distribution of these costs. These parties, who have reform veto power, therefore initially fail to come to an agreement which would initiate the reform. The agents hold each other up and wait for the other party to give in. They are assumed to have waiting cost, political or economic, as long as reform is not executed. For instance, the government in office may experience costs by having to counter lobbying whereas the opposition incurs the costs of performing the lobby in order to remain visible on the relevant subject.
In the full information case, the agent with the highest waiting cost would instantly give in, leading to reform to be initiated instantly. That is, no waiting is Pareto superior to waiting. However, the waiting costs are private information. With this information asymmetry, reform will take place when for one party the marginal cost of waiting equals the marginal benefit of reform. Eventually the reform will eventually occur. However, the timing of this reform is a priori for both parties uncertain. The passage of time is to reveal which party has the highest waiting costs, because its political or economic capital dries up eventually.3
3
11
2.2.Government strength
Due to its formal nature, the war of attrition model is very stylised. It for instance assumes that one group is completely uncertainty about the waiting costs of the other group. Yet, in practice one can attempt to make an estimation of the groups’ relative waiting costs by looking at indicators of the strength of the groups, such as the number of seat parties occupy in the legislative or executive branch of the government. Similarly, a higher degree of agreement between parties within the government may be considered to contribute to the relative strength of the government versus the opposition and to lower relative waiting costs. Stronger parties are assumed to have lower waiting costs: They run less quickly out of support. It is therefore that the strength of the government and opposition may be predictors of the likelihood of reform. When the observed strength of decision making parties is highly skewed, one would expect that the subjective expectation of the relative waiting costs to be such that beneficial reform is initiated early.
The strength of the parties is not constant. In practice, because of its privileged position, the government can generally be considered to be the strongest group. Yet, elections may redistribute political clout. Therefore, in our view, the uncertainty about which groups are strongest is highest just before elections. As a result, the rulers may have higher costs of waiting and become more inclined to be accommodating with reform when election draws nearer. On the other hand, we can argue that the waiting costs are perceived to be less skewed. This means that the weakest party will be more inclined to block liberalisation proposals.
2.3.Coordination problems and lobbying for special interests
Perotti and Kontopoulos (2002) argue that the harder it is for decision makers to coordinate, the more likely it is that welfare suboptimal policies are adopted. Governments with few coordination problems will appear stronger and should thereby be able to more easily agree on initiating financial liberalisation. Coordination problems for instance arise because there are more decision makers or decision makers of different political background involved. The former because of free riding behaviour, the second because when agents have different interests or opinions there is not a single equilibrium at which a consensus may end up.
12 reverse argument hold as well. Gains of reform can be concentrated, while the costs are diffuse (Fernandez and Rodrik, 1991). In this latter case, losers suffer to a large extent from free riding in lobbying efforts. Hence, a small group of gainers may be able to coordinate and induce a government to go ahead with a reform.
Interest group explanations of financial liberalisation or the lack of it are likely to be of considerable relevance. Pleasing a passionate minority that lobbies either for or against reform may raise support for a politician while hurting a silent majority may not cost political support. This could be based on the assumption that the silent majority either is unable to recognise the small costs or that it irrationally does not care about small costs. The passionate minority may for instance effectively communicate via the media what are the benefits of their proposal, while the silent majority fails to communicate the disadvantages. It is also possible that the trading of votes on different issues can make a proposal that benefits a minority at the expense of a majority get adopted (Oppenheimer, 1979).
Girma and Shortland (2005) argue that the sectors which are often favoured by politicians tend to be oligopolistic. As a result, these sectors find it relatively easy not to suffer from the collective action problem. Furthermore, the rents that these sectors obtain from the favourable treatment by the government can be used to improve their lobbying. For these two reasons, long favoured sectors are expected to be powerful lobbyists. According to Haggard and Maxfield (1993, quoted in Girma and Shortland, 2005, p.8) the manufacturing and agricultural sectors tend to be the favoured sectors. Sectors that were long favoured be the government should be expected to lobby for the status quo since reform may change power balances.
Some interest groups are too weak and other groups are not represented at all in the political process (Olsson, 1965, quoted in Girma and Shortland, 2007, p.5)). Firms are probably better able to influence government policy than individuals, for instance due to technical competence or economies of scale. But also among individuals, there are political inequalities. A point made by Hausmann (1994) is that often the poor are so poor as to lack political importance. And when opponents are weak, it is particularly attractive to lobby since the costs will be low.
2.4.Political system
2.4.1.Discretion: short-term checks and balances
13 increase the government’s political cost of waiting for reform. For instance, fighting lobbies may be more costly in political systems where the executives have less discretion. On this basis, Alesina et al. (2006) hypothesise that presidential systems will delay reforms less than parliamentary system. Presidential systems have more powerful executives, who have relatively low waiting costs.
Persson (2002) indeed point to such a crucial difference between presidential systems and parliamentary systems. In presidential systems the executives’ likelihood for survival is relatively high. That is because parliamentary systems have executives that necessarily rely on the confidence of the parliament during the entire term. Therefore short-term checks and balances are better in a parliamentary system. As a result, political costs of consolidation during a war of attrition are more skewed in presidential systems, ceteris paribus. This leads to earlier agreement on the distribution of the reform burden in presidential systems.
2.4.2.Broad reform and agenda setting power
Rodrik (1996) shows that a specific type of reform, financial in our case, may come in a package of broad reform. If the package contains popular policies, undesirable reforms can be adopted due to the fact that the package has to be accepted or rejected as a whole. This shows that agenda setting is an important power.
Persson (2002) argues that the separation of powers is more strict in presidential systems. Parliamentary systems often provide governments with agenda-setting power. Following Rodrik (1996), we can then argue that countries with parliamentary system are more able to execute reform that might be in the interest of the general public but that nevertheless are unpopular. Governments in presidential systems can with more ease use their entire term to execute a reform which in a parliamentary system would have resulted in its collapse in the short term due to the loss of support in the legislature.
At the same time, Persson (2002) argues that this higher degree of power concentration in a parliamentary system allows politicians to collude at the expense of the electorate. That is, they might also use the power concentration to their own benefit. If this effect is dominant, we would expect presidential system to perform more financial liberalisation.
2.5.Learning
14 Even looking backwards, we cannot be certain about what would have been the optimal strategy. This is because we cannot observe the counterfactual; i.e. that is unknown what would have happened without reform.
When a country based on historical experience, either irrationally or not, associates financial freedom with terrible a phenomenon such as hyperinflation, it will be more likely to oppose financial liberalisation. In such a case, it does not matter whether financial liberty is actually related to the bad event. Liberalisation will be less likely. This argument is primarily inspired upon Germany’s support for strict money growth based upon the association between hyperinflation and the rise of fascism (Haggard and Webb, 1993).
Thus observed outcomes are often used, justly or not, as evidence for the effectiveness of a certain policy. This is because people have the tendency to provide causal explanations for historical processes that were in fact not causally linked (Taleb, 2007). In fact, people are often biased to such an extent that they do interpret outcomes that seem to confirm their prior beliefs while they fail to interpret outcomes that disproof their prior beliefs as valid evidence. Taleb (2007) argues that such mental biases are especially prevalent among experts such as economists. Fernandez and Rodrik (1991) do not stretch the argument that far. The simply assume that new information is revealed when a reform is initiated and that no new information is revealed when decision makers fail to agree to reform. Thus, more experience with reform makes a country better informed about the consequences of reform. Now assume that financial liberty is beneficial for society's welfare. We can then conclude that liberalisation is positively related to liberty: The more financial freedom a country has, the more knowledge its agents have about the benefits of it and the more inclined decision makers will be to opt for further financial liberalisation.
15
2.6.Status quo bias
As has been described, it was observed that many countries had had a long period of a (constant) high degree of financial repression before eventually a liberalisation sequence was initiated. This is often referred to as a status quo bias. There is a number of reasons why such a status quo bias may exist.
2.6.1.Individual uncertainty
Fernandez and Rodrik (1991) explained the status quo bias based on the basis of uncertainty. In their formal model, reforms are assumed to be efficiency enhancing. An important assumption is that losers cannot be compensated with transfers. Compensation packages generally are dynamically inefficient, i.e. that decision makers cannot credibly commit to execute the transfer after the expected losers have voted in favour of the reform. Haggard and Webb (1993) mention that economic agents occupy several positions simultaneously; they can be producers, consumers, and recipients of transfers and a lot more. The interests of these positions are often conflicting. This makes the assessment of whether someone will gain or lose from a reform difficult, especially when there is rapid economic change and resulting uncertainty is large. If the loser of reform are difficult to identify, a compensation package is likely to lack political support.
Ex ante, the distributional consequences of a proposed reform are known for groups in the model of Fernandez and Rodrik (1991). However, individuals do not know whether they will gain or lose. It is therefore possible that aggregate support for the reform is insufficient even though a majority is bound to gain.
To understand this outcome, we need to make a distinction between the ex post distribution and the ex ante expected distribution faced by the individual. Individuals make their decision whether or not to support the reform based on the ex ante expected outcome. Individuals belonging to the minority group of certain winners are certain to support the reform. However, the majority of individuals belonging to the other group have to take into account that there is a chance that they are not among the winners. This possibly makes the ex ante expected individual outcome negative. Sufficient support for welfare enhancing reform may thus be absent.
16 Now we turn to the status quo bias. A crucial assumption for the status quo bias to exist in this model is that information about the gains of reform is only revealed when reform is performed. When the initial decision was against reform, no additional information about the reform will be generated. Then, it cannot be that during the first period, there is insufficient support for reform, while reform will be executed in the second period. In the long run, it is therefore most likely that the initial policies will prevail: Bad policies get reversed while superior policies often lack electoral support due to individual uncertainty.
2.6.2.Risk aversion
The model of Fernandez and Rodrik (1991), which has risk-neutral agents, suggests that reforms with large gains that benefit the majority should always get enough support to be executed. Yet, when we assume that agents are risk averse, this prediction may no longer hold. Risk-averse agents prefer the certain suboptimal outcome of current policies over the superior, but uncertain outcome of reform. A high degree of risk aversion may thus contribute to less financial reform. 2.6.3.Lack of rewards
Benevolent reform is not always rewarded. Incumbent may lose office because their political competitors are more “attuned to the political consequences of structural transition” (Przeworski, 1991, p.138). If incumbents are aware that their opponents will become relatively stronger as the results financial liberalisation, they will have the incentive to block the liberalisation and to maintain the status quo.
Voters are often assumed to be backward looking. The incumbent rulers are assumed to be responsible for the current state of the economy. Elections then serve as the final judgement and the incumbent is assumed to be rewarded for favourable economic outcomes. Fidrmuc (1998) however argues that this did not apply for countries in Central and Eastern Europe during the first post-communist elections. This would suggest that governments have few incentives to initiate reform in the first place.
17 2.6.4.Fairness
One has to take into account that real agents are not perfectly rational self-interested agent. On the one hand, losers of reform can also support reform simply based on a normative commitment towards it (Przeworski, 1991). On the other hand, those who gain from reform can oppose it on the basis of perceived fairness of the distribution of the gains and costs of the policy change. 2.6.5.Political bargains
Quinn and Inclan (1997) argue that policy inertia is caused by political bargains made in the past. Breaking promises may more strongly backfire than refraining from initiating welfare increasing reforms.
2.6.6.Short term transition costs
Change brings about transition costs. These costs tend to arise early in the process of policy execution. Transition cost thus constitute an investment. Tommasi and Velasco (1995, p.31) explain that this is politically troubling: “losers immediately take to the streets, while winners emerge and acquire a political voice with substantial delay.” Partial reversal or a reduction in the scale of liberalisation is therefore not uncommon.
2.7.Radical, gradual reform or no reform
If society decides to reform, it has to decide upon the extent and speed of reform. A radical reform is a quick reform which produces the adjustment costs early. Radical reform also yields the assumed positive results after a shorter period than gradual reform. Gradual reforms are slower, meaning that the social costs are spread over longer period of time. Radical reform is assumed to have higher total social costs than gradual reform (Przeworski, 1991). If this were not the case, then there would be less reason to initiate a gradual reform. Despite this assumption, why would rulers opt for radical reform?
Przeworski (1991) provides an explanation based on risk.4 Assume that certainty of events to occur in the near future is higher than the certainty about events in the more distant future. The more certainty there is, the more weight distant events have. Events can be the realisation of
4
18 reform gains and costs. Assume reform will actually be beneficial. The more certain society is about distant events, the more inclined it is to support reform. In the case agents have much confidence regarding the future, radical reform can be preferred to gradual reform despite its higher social costs. That is because now the expected present value of the gains is high. If confidence is moderate, society will opt for gradual reform. With low confidence, the status quo is the optimum: The expected gains do not outweigh the expected costs.
Support for reform tends to erode over time (Przeworski, 1991). This for instance because politicians have a tendency to make unrealistic promises on the effects of reform in the short run. A decline in the confidence of the successfulness of reform is therefore not unheard of.5 Reform is thus not always executed to the initially intended extent.
The fact that gradual reform or the previous status quo is preferred after radical reform has been initiated can be explained by the fact that some people do not like to wait for the costs they will have to incur. They rather get it over with instantly (Loewenstein, 1987).6 However, when they actually experience the social costs, they dislike the costs so much that they will try to reduce them.
There is another reason based on the erosion of support that leads to radical reform being preferred to gradual reform. A gradual reform, compared to a radical reform, is more vulnerable to reversal simply due to the fact that the opposition has more time to fight for reversal before the gains have been observed (Przeworski, 1991). Pushing reforms rapidly through the system can also weaken interest groups that are tied to the status quo and give antireform forces little time to mobilise (Douglas, 1990 cited by Haggard and Webb, 1993, p.159).
Przeworski (1991) assumes that returning to the status quo is more costly during a radical reform than during a gradual reform. Having started a radical reform may therefore prevent a society which prefers to completely reverse the reform to actually do so. Instead, it may opt for a slowdown of reform, ultimately choosing the gradual liberalisation option. For this reason, reforms can contribute to increasing the credibility of the government, assuming that the promises
5
This requires expectations to be based on e.g. the communication of politicians. 6
19 of the reformers become reality.7 Namely the government is shown to have carried out the promise to reform and the promised positive effects also occur eventually. Radicalness signals the commitment of the reformer (Przeworski, 1991). Therefore, it may be optimal to overshoot one's actually reform target. Rodrik (1989) provides formal proof that on the basis of signalling, overshooting may be an optimal policy.
2.8.Inequality
Przeworski (1991) addresses the influence of (income) inequality on the likelihood of reform. The early stages of reform may be life-threatening for the weakest in society because it brings about short term adjustments costs. First let us assume that reform does not have redistributional effects. In an egalitarian society, a large part of the population may fall below the subsistence level during the early stages of reform when the adjustment costs materialise. In an unequal society with a small part of the population being poor, only a small part of the population will see its life threatened. Therefore, we should expect reform to be more prevalent in countries with high income inequality. The reverse expectation holds when the income distribution is more realistically skewed: skewed such that a large part of the population has low income. In this case we would expect more inequality to relate to less reform. Obviously, this theory is most relevant in developing countries.
Now assume that reform also has redistributional effects in the sense that the costs of reform are affecting some relatively more than others. If the most vulnerable of society suffer relatively less reform costs, fewer people will be eliminated due to reform. Hence, reform will get more support, ceteris paribus. The opposite conclusion holds if those with the lowest income face reform costs which are relatively high. Worse, in voting models the median voter may not even support a welfare enhancing reform if the distribution of the gains are such that the median voter is not among the winners (Rodrik, 1986).
We would like to emphasise that the theory of Przeworski (1991) should not be taken too literally. The economic agents distributionally affected could e.g. be firms, which have the possibility of going bankrupt. If politicians let their policies be much affected by the pressure of lobby groups of firms, a similar prediction holds.
7
20
2.9.Identity of reformers
2.9.1.Honeymoon hypothesis
The honeymoon hypothesis states that reform is more likely to be adopted immediately after a new government is elected. Tommasi and Velasco (1995) argue that simply the fact that a new government is "in control" can reassure people almost independently of what the reform looks like. Change is key. Przeworski (1991) stresses the importance of policy failure of the previous government that might have even caused a crisis. Haggard and Webb (1993) explain that the new government can blame the early costs of reform on the previous government.
2.9.2.Partisan orientation/ideology
Quinn and Inclan (1997) discuss the effect of partisan policy on financial liberalisation and that of international financial openness in particular. Right-wing politicians are expected to represent citizens with an above average amount of financial capital. This constituency directly benefits from financial liberty by being able to utilise their capital in the way it considers to be optimal. It also benefits indirectly because financial liberty forces the government to adopt macroeconomic policies which are preferred by investors in order to prevent capital flights. Anti-inflation policies serve as examples of such macroeconomic policies.
The previous argument is dominant in capital rich countries. In capital poor countries, financial liberty will attract much foreign capital. This drives down the returns to capital for the incumbent domestic capital owners. This price effect suggests that in capital poor countries, left wing governments may be interested in financial liberalisation. That is because labour, which is assumed to be represented by left-wing governments, will benefit from the drop in the price of capital. The benefits arise when we assume that capital and labour are complements in production.
21 According to Alesina et al. (2006), constituents may be more willing to accept policies which are against their interest if they are initiated by the party they support rather than when they are proposed by parties representing other constituents. This would imply that leftist governments would find it easier to implement financial liberalisation. They would get support from right-wing parties, which ideologically support financial liberty, as well as some support from their own constituents. Similarly, right-wing governments would have fewer difficulties with reversing financial liberalisation if they would want to. The latter is unlikely when we assume that financial liberty is beneficial to society at large.
Cukierman and Tommasi (1998) pose that voters sceptically judge policy proposals of incumbents. Their scepticism is based on the fact that they are unaware whether the rulers propose reform for the improvement of overall welfare or for the sake of pleasing their constituents. Yet, voters making their judgement have to rely in the policy proposals, since the results of policy will only be visible in the future. This is due to the fact that politicians are better informed about the effects of policies than is the public. Rulers have an informational advantage because by law they have access to classified information, technical assistance and because they deal with policy issues every day. The asymmetry in information makes that the ruling government is able to use policy proposals as a signalling device. If a left-wing government proposes a right-wing policy, the electorate has a larger tendency to believe that the proposal was inspired upon benevolence than when the same policy would have been proposed by a right-wing party.8 The reverse is also true; a right-wing government will not be able to credibly communicate the benefits of financial liberalisation and hence it will be unable to execute extensive liberalisation.
2.10.Crises
Alesina et al. (2006) suggest that the war of attrition model seems especially applicable during times of crisis. When no crisis has occurred, nobody may consider it necessary to perform a reform. During normal times, the costs of waiting may be too low. However, with crises, agents may feel that trying to change the situation is better than staying idle. The waiting costs will rise.
8
22 Drazen and Grilli (1993), making use of a war of attrition model, provided formal proof that a crisis may even increase welfare. This is likely when one party has high waiting cost as the consequence of the crisis, since it will lead to early reform. The benefit of preventing long reform delay is larger than the negative effects of the crisis. According to Corrales (1997), in practice international policy advisors even went as far to recommend politicians to wait for the economic conditions to become disturbing before initiating reform.
Crises are especially useful when parties are asymmetrically hurt. Haggard and Webb (1993) confirm that crises change the power balance of interest groups. This may even happen if a crisis only has distributional effects. The more skewed distribution of waiting costs become, the earlier is reform.
An example of a crisis can be high inflation. The poor, who tend to own little capital and therefore may not support far reaching reform at low levels of inflation, may be hit hardest by high inflation. That is because the poor have less access to financial technologies which yield protection against inflation (Labán and Sturzenegger, 1994). Continuing high inflation weakens the relative bargaining position of the poor and hence will eventually lead to reform.
Crises do not only speed up the attrition of those blocking current reform proposals; they may also quicken the attrition of the reformers. A government that is overwhelmed by a crisis might technically be unable to execute its desired reform. Corrales (1997) therefore stresses that crises are responded upon in a variety of ways. Not every crisis will make reform more likely.
Also in the presence of a government with a status quo bias and an opposition favouring reform may a crisis become a factor that facilitates reform. Krueger (1993, p.353) suggests that crises undermine parties supporting the status quo, rendering “politics as usual” no longer sustainable. Rodrik (1996, p.16-17) indeed wrote with regard to debt crises in Africa, Asia and Latin America in the 1980s: “Orthodox economists who had the ear of policy makers now had their chance to wipe the slate clean and mount a frontal attack on the entire range of policies in use.”
23 Due to information imperfections, the majority may not be aware of the benefits of more liberal policies and of the disadvantages of prevailing restrictive policies. Agents likely base the perceived effectiveness of policies on the economic outcomes they observe. A crisis may be considered as an outcome that may be perceived as the result of bad economic policy, such as financial repression: a crisis contributes to the learning about the “right” model of the world (Tommasi, 2002). Then, a crisis may generate the support for reform that was lacking during normal times.
Initiating reform is a risk, since the outcome will a priori never be known for certain. Tommasi (2002) points out that people are more willing to take risks in the prospect of high losses than when confronted with favourable prospects. Hence, during crises people may be more willing to opt for the risky option of financial liberalisation.
2.11.Political business cycle
In political business cycle models it is (often implicitly) assumed that voters have a short memory, or do not care about the more distant past. It is likely that the costs of reform that is in the interest of the general public dominate the gains during the early stages. This means that executives who strive to be re-elected would like to refrain from reforming close before elections. The myopic electorate is here believed to blame the incumbent for the reform costs, while failing to appreciate the gains that follow later.
Shortly after having obtained office is the best moment for reform, since it is most far away from the next election. The cost of reform might be forgotten during this second election, while the gains, which have just arrived, are currently being highly appreciated by the electorate. Thus, liberalisation may be more likely at the start of a government’s term.9
The classical version of political business cycle models take the approach that the benefits of policies are delivered instantly, but that the negative consequences arise later (Haggard and Webb, 1993). For instance, a government may rise spending on social programmes close to election day and face the problems of fiscal deficits afterwards. If policies are offered in a broader package, then it is possible that financial liberalisation may be facilitated by political business
9
24 cycles: The electorate considers the short term costs of liberalisation as acceptable, since the short term gains of additional government spending outweigh these short term costs. Especially an incumbent regime that believes its days are numbered will be strongly tempted to drum up support through expansionist policies and delays of (the short term costs of) reform, even if this policy is self-defeating over the longer run (Haggard and Webb, 1993).
Haggard and Webb (1993) suggest that the very poor may have a very short time horizon in a struggle for subsistence. One would therefore expect the electoral cycles to be most common in developing countries.
Rogoff and Sibert (1988) and Rogoff (1990), as cited by Haggard and Webb (1993, p.140), provide models in which an information asymmetry between the government and the rational electorate can also be the cause for a political business cycle.
2.12.Political competition, rent-seeking and time horizon
Olson (1993) discusses the role of politicians' time horizon. If a government expects to leave office soon, it is in its leaders' interest to maximise extraction of short term rents. Executing less extractive policies (such as more liberal financial market policy) may potentially provide higher rents to the leaders. However, because the expected value of this type of policy, which takes into account the possibility of losing office soon, is lower than the expected value of major extraction, the latter policy of high extraction will be preferred by governments with a short time horizon. Governments with longer time horizons will see the fruits the financial liberty it allowed for, meaning that more rent can be divided.
2.13.Democracy versus autocracy
2.13.1.Corruption
Olson (1990 cited in Haggard and Webb, 1993, p. 146) shows the importance of the degree of accountability for the severity of corruption. Bureaucrats in authoritarian systems due to a lack of accountability may provide much reform resistance in an attempt to protect their own bribe income. For this reason, democracies may liberalise more.
2.13.2.Time horizon
25 lead to more reform in financial markets, since liberalisation will deliver some of its effects only in the medium or long term.
2.13.3.Political replacement theory
Acemoglu and Robinson (2006) formally showed that innovations that are beneficial to society as a whole, such as financial liberalisation, are most likely to be adopted by countries that are either very democratic or very autocratic. In their model, the rulers make a trade off between the additional rents caused by the innovation and the probability of remaining in office. Financial liberalisation enriches non-elite groups, which may be a threat to the incumbent rulers. Acemoglu and Robinson (2006) assume that non-elite groups will gain political power at the expense of the incumbents. In democratic societies, rulers will be punished with being kicked out of office when they have only slightly deviated from optimal policies. In very autocratic countries, there is no system that allows outsiders to remove the incumbents from office, even though the outsiders gained power from a liberalisation. The government does not have to worry about losing future rents. It will therefore maximise the expected value of its own rents by adopting policies that are good for economic development. However, when the government faces an intermediate degree of checks and balances, it will opt against financial liberalisation. In this case the benefit from keeping the relative political power of the outsiders constant outweighs the cost of sacrificing the forgone additional rents.
On the assumption that human capital is complementary to economic development, Acemoglu and Robinson (2006) also found that the political elite has a higher tendency to block beneficial innovations when the countries human capital is low. For our case this means that this is because the beneficial effects from the financial liberty are assumed to be higher when society is skilled. 2.13.4.Dynamic inconsistency
26 that the democratic safeguards are especially effective in preventing reversals. As a result the time inconsistency problem of temporarily using liberalisation as bait for extra investment is less profound.
2.14.Tying the hand of current and future executives
Rajan and Zingales (2003) touched upon the idea that the legislature may want to give the financial sector freedom in order to reduce the scope to which the executives can use their power in an unwanted manner. Similarly, we can imagine that current governments in office may want to restrict their own discretion in order to tie the hands of the next government. Restriction oneself may have an instant benefit. Haggard and Webb (1993) argue that sometimes rules are better than discretion: Delegating adds credibility to a government's promise not to expropriate financial investments. These arguments addressed in this paragraph either assume that performing reverse liberalisation costs effort, time or that imposing more financial repression is politically more difficult once society is used to a high level of financial liberty.
2.15.Fiscal deficits and policy scrutinisation
Girma and Shortland (2005) predict that high fiscal deficits reduce the likelihood of financial liberalisation based on the argument that it will disrupt the status quo and may cause capital flights. On the one hand, controls on capital shield governments’ policies from the scrutiny of investors. On the other hand will financial liberalisation make it more attractive for foreign investors to transfer capital into the country. This allows the government to more easily finance its fiscal deficit. The latter argument will be dominant when despite the fiscal deficit the government policies are not considered as a threat to a profitable investment in government bonds.
2.16.International factors
2.16.1.Systemic convergence
27 argument economically and political powerful countries may have provided political pressure for other countries to liberalise.
Simmons and Elkins (2004) explains that countries are not all alike, but that groups of countries have similarities that makes them compete for the same part of international capital that is available on a global basis. Countries with similar characteristics and risk profile compete in the same pool of capital and may therefore deviate little in their liberalisation strategies.
Girma and Shortland (2005) also emphasise the importance of international political pressure. They see a key role for the United States, the OECD, NAFTA, GATS, the EU and the international financial institutions such as the World Bank and the IMF. Political pressure is especially relevant due to the fact that these players occasionally have an own interest in reform. This is often the case because they have provided loans or aid themselves.
Boockmann and Dreher (2003) studied the effect of the involvement of international institutions on economic reform for a number of reasons. Accepting conditional loans from the IMF or the World Bank obliges a country to make policy adjustments. Because of this but also due to the fact that the executives can blame the IMF for the costs of reform, reform may be more likely when loans are accepted. IMF involvement reduces the government’s political costs of reform.
Additional funds received from international organisations provide budgetary room for the government. Hence, those in power have deeper pockets for fighting the war of attrition. This induces the opposition to realise that it will probably lose this war. Hence, it will go give in sooner and reform will be more likely to happen early.
On the other hand, with additional funds, the government may feel less need to perform reforms (Boockmann and Dreher, 2003). The loans soften the government budget constraint. In this case, additional funds may lead to postponement of reform. This is especially so if the government has a short time-horizon. Additionally, governments sometimes act in this manner, because they cannot credibly commit to fulfil the IMF conditions.
28 2.16.2.Openness
Rajan and Zingales (2003) take a vested interest perspective on liberalisation.10 In order to make their theory more applicable to the topic of financial liberalisation we assume that more liberal financial systems have a tendency to develop more financial transactions which happen at arm's length. A characteristic of arm's length markets is that these do not respect incumbency.
Rajan and Zingales (2003) argue that incumbent firms in the real sector as well as incumbent financial firms sometimes have an interest in opposing more liberty. However, their power and interest to do so is not static; it is related to a country’s openness to trade.
Incumbents in the real sector in countries with an undeveloped financial sector can during normal times finance projects with the return to past projects or through attracting external capital against collateral. This means that incumbents have a financing advantage over new entrants. This advantage will be reduced when the financial sector develops due to financial liberalisation. That is because more developed financial sectors are able to supply more and more types of funding. Incumbents in the real sector may thus oppose financial liberalisation.
Incumbents among the financial firms may oppose liberalisation because it indirectly reduces the importance of direct relationships. The more developed financial sector that resulted from liberalisation will deal more at arm's length, which is not the in line with the competitive advantage of incumbent financial firms. Thus, liberalisation will develop the financial sector, and changes relationships, competitive advantages as well as rents.11
In countries that are relatively closed, incumbents are powerful enough to protect their position by preventing liberalisation. Incumbent firms in the real sector face more competition from abroad (and as a result extract lower profits) and have more investment opportunities abroad when openness is increased. The first aspect necessarily reduces their reliance on internally generated funds due to lower rents. This as well as the latter aspect increases the firms' demand for more external funding. Higher trade openness however also allows firms in the real sector to more easily perform activities such as transfer pricing among its foreign and its domestic
10
More correctly, their theory is about financial development. It however contains elements that can also apply to financial liberty.
11
29 subsidiaries. In this way, the incumbents in the financial sector may be worse off due to increased openness.
Furthermore, foreign firms may be important lobbyists for more domestic liberalisation since these newcomers only have potential gains and no rent losses from more domestic financial liberty. Increase openness therefore increases lobbying in favour of financial liberalisation. Finally, Rajan and Zingales (2003) emphasise that the effect of increased openness may not exist if the rest of the world does not provide access to capital.
As a limitation we would like to note that the theoretical considerations of Rajan and Zingales (2003) mainly treat openness as exogenous. Openness is in fact also an outcome of the political process. This limitation is probably less relevant for small countries, which are more dependent on the outside world. In such countries the decision on trade as well as capital openness probably is less influenced by politics. In countries that more easily can afford to live autarky, the degree of openness might simultaneously be decided upon as on the degree of financial liberty.
2.16.3.Repression in exchange for trade openness
Rajan and Zingales (2003) explain the lack of financial development, which occasionally proxies for financial liberty, that could be observed between the 1930s and the 1980s by referring to major event such as the World Wars. The United States, in contrast to numerous other developed countries, came out of World War II with a competitive advantage. It thereby had a strong interest to lobby for trade openness. The authors argue that the less competitive developed countries may have allowed for trade openness only in return for more restrictions on finance.
2.16.4.Competition for financial centre
Some countries may in their industrial policies aim to turn their financial sector into the leading financial sector of the world. Competition between cities such as New York, London and Tokyo were therefore a likely cause of financial liberalisation in these countries (Helleiner, 1994, cited in Rajan and Zingales, 2003, p.25).
2.17.Trend
30
2.18.Price incentive
Assuming that lower prices increase the quantity demanded is basic economics. Central governments are often concerned about the country's ability to attract capital from abroad. In order to smooth the inflow of capital and thereby facilitate the exploitation of low external interest rates, governments can make investment more attractive by taking a hands-off approach (Bartolini and Drazen, 1997). In short, one can assume that less government interference in financial markets can serve as a substitute for interest rate premia.
2.19.Demand for insurance
Rajan and Zingales (2003) argue that as a result of World War I and the Great Depression, the demand for insurance initially experienced a strong increase and as a consequence governments were given control over the financial sector. Perhaps the fading of the memory of these historical events reduced the demand for insurance can then partially explain the recent liberalisation trend.
2.20.Religion
Stulz and Williamson (2003) discuss the role of religion. First, in contrast to Protestantism, Catholicism tended to be organised in a hierarchical manner. Power can therefore be expected to be more centralised in Catholic countries. Protestantism has a stronger focus on individuals, which may indicate that Protestant countries have a stronger preference for liberty. Second, Catholicism as well as the Islam have been hostile against the payment of interest, which provides an indication for a bias in favour of financial repression.
2.21.Bureaucrats
Financial reform is a deep change, which according to Tommasi and Velasco (1995) requires expert advice, lengthy discussion and participation. Technocrats such as bureaucrats then also are important. If the bureaucracy opposes reform, it is unlikely to be adopted. According to Przeworski (1991), members of the (economic) reform team are technocrats who primarily care about the overall reform results and do not care about social cost. Societies’ tolerance for reform and politicians’ courage can be assumed to be limited (Nelson, 1984, cited in Przeworski, 1991, p.165). Technocrats would therefore have a preference for radical reform.
31 Increasingly these remaining bureaucrats start believe that it is unlikely that the reform will offer them better jobs. Hence, they will remove their support for reform.
2.22.Quality of law enforcement
According to La Porta et al. (1997), richer countries provide better law enforcement than poorer countries. This means that financial market regulation that is in place in rich countries translates into more de facto repression. It is therefore possible that these countries prefer to have a limited amount of financial market regulation in order to prevent themselves from having a high degree of de facto government interference.
2.23.Complementary government interference
Rajan and Zingales (2003) argue that some government interference is good for the development of the financial sector.12 One example is regulation that prescribed a minimum degree of transparency. When financial a financial market is repressed, additional desirable regulation will have no or only a small effect on financial development. But with a high degree of financial liberty, some regulation may be beneficial. Hence, some forms of government interference are complementary to liberalisation. Liberalisation may then be desired since it is the optimal response to prudent government interference. This also suggests that if a country's government lacks the technical ability or "social capital" to provide this type of interference, then financial development may not arise despite to liberalisation of other aspects in financial markets. If this is expected, the liberalisation efforts will not be made. Haggard and Webb (1993) suggest that the implementation of reform takes a while because it can only be fruitful if it is complemented with institutional change.
2.24.Legal system
Based on a similar argument about the complementarity to liberalisation, a Common Law system may be more desirable for financial liberalisation than a Civil Law system since it may provide better investor protection against expropriation by insiders (La Porta et al., 1997). That is, liberalisation may only be a profitable strategy when it is complemented with an increase in the
12
32 quality of investor protection. Indeed, Tressel and Detragiache (2008) provide some indirect empirical evidence for this theory.
There are additional reasons why the legal origin of a country may matter. Law making tends to be more centralised in a Civil Law system than in a Common Law system. Rajan and Zingales (2003) argued that Civil Law is therefore probably better fit for executing the will of the government. This suggests that countries with a Civil Law system would reform more easily. On the other hand, interest group lobbying against liberalisation will also be more effective since lobbying efforts can be more focussed. This would suggest that liberalisation is more common in Common law systems. This argument however requires that the most effective lobbyists have interests which are not in line with national interests. I.e. that lobbying groups oppose financial liberalisation.
Beck et al. (2001b) stress the importance of the legal system's ability to adapt legislation and jurisprudence to changing conditions. Financial liberty most likely brings about financial innovations. Common Law is said to be more flexible. Hence, Common Law countries may more easily profit from potential financial innovation and thus from financial liberalisation.
3.Empirical literature
3.1.Lora and Olivera (2004)
Lora and Olivera (2004) created an indicator of measures capturing the liberty to the market allowed for by financial policies. It was created on the basis of four subindices which are given equal weight. These measured the liberty of interest rates on deposits, freedom of interest rates on loans, the real level of reserves of bank deposits, and finally quality of oversight in the financial sector. The financial policy index was normalised from 0 to 1, with 0 corresponding with the minimum value in the sample and 1 with the maximum value. The data of Lora and Olivera (2004) is annual data for 1985-1995 covering 19 Latin American countries. The first difference of the financial policy index was used as the dependent variable in regressions testing a number of hypotheses.
33 of real GDP per capita with the previous maximum real GDP level, GDP growth rate during recessions, fiscal deficits when larger than 3 percent of GDP, and a measure for inflation volatility. Furthermore, political variables did not produce a significant influence. A dummy for the second year of the presidential term and a variable measuring government strength or otherwise called "political fragmentation" did not have a coefficient which significantly differed from zero. The bundling with other policy reforms, such as tax and trade reforms, was also insignificant. Financial reform also seemed unrelated to state efficiency according to Lora and Olivera (2004).
3.2.Abiad and Mody (2005)
Abiad and Mody (2005) created an indicator for financial liberty, covering 35 countries for the period 1973-1996. This indicator is a normalised version of an aggregation of six subindices. These subindices capture the degree of credit control, interest rate control, entry barriers into the banking sector, operational restrictions and the establishment of security markets, privatisation of banks, and restrictions on international financial transactions. After aggregation the financial market policy index was normalised to a 0-1 scale. The scholars then used the first difference of the normalised index as their measure of financial liberalisation. With an ordered logit technique, Abiad and Mody (2005) estimated a number of models in an attempt to investigate what factors may have served as determinants of financial liberalisation.
They included variables which can be summarised under the categories of learning, shocks, ideology and structure. The level of financial freedom appeared as robustly positively related to reform. The square of this term generated a significant negative coefficient in most regressions. The authors concluded that the bias for the status quo decreases with financial liberalisation and that reform is most profound at intermediate levels of liberty. Hence, what they found is an inverse U-shaped relationship between liberty and reform. Yet, it is disputable whether the conclusions concerning the effects of financial liberty may be interpreted in the way Abiad and Mody (2005) did.13
The level of economic development, as proxied by GDP, was interacted with financial liberty. This term was positively related to reform. Abiad and Mody (2005) explained this by stating that
13
