Financial Liberalization, Saving, and Growth*

Klaus Schmidt-Hebbel Central Bank of Chile Luis Servn The World Bank

Abstract Saving rates display significant variation across world regions and over time. This survey first reviews the main determinants of saving rates, with particular attention to the effects of financial liberalization. While financial reform can affect saving through various potential channels, on the whole its net effect is ambiguous. The international evidence suggests a positive association between financial development and saving across countries. However, empirical studies controlling for other saving determinants reveal that the effect of financial development on saving is, if anything, negative. Regarding the growth implications of financial reform, the international evidence confirms the theoretical prior of a positive effect of financial liberalization on growth, channeled through higher investment and improved resource allocation leading to an increase in productivity. Finally the paper analyzes the Chilean experience, where financial liberalization was key among the structural reforms that gave a boost to growth, investment, and saving.

* For presentation at the Banco de Mxico Conference on Macroeconomic Stability, Financial Markets, and Economic Development, Mexico, DF, Nov. 12-13, 2002. We thank Matas Tapia for excellent discussion and efficient assistance.

1 Introduction Saving provides the resources for investing in physical capital, an important growth determinant. But saving rates show disparate regional trends, with potentially important implications for growth and development. What explains these trends and which policies are most effective in raising saving rates? One major policy reform that has been adopted by an increasing number of industrialized and developing countries alike has been the deregulation and opening of banking and other financial service industries. In many cases, these reforms have been advocated on the grounds of their favorable impact on saving. However, according to theory, financial liberalization can affect saving through various channels, which operate in different directions. What does the empirical evidence say about the effects of financial liberalization on saving? Looking beyond saving, has financial liberalization raised growth? This survey paper addresses the three questions posed above by drawing from both the empirical evidence for the world at large and a specific country experience. Recent cross-country studies suggest policy-relevant answers to the questions we pose. We complement these findings with a detailed review of the evidence for Chile, a country that has applied deep policy reforms for a sufficiently extended period to allow useful inferences on the role of financial liberalization on saving, investment, and growth. Section 2 analyzes the main stylized facts of saving rates in the world, as well as the determinants and policies affecting them. Section 3 discusses the impact of financial reform and development on saving, based on theoretical and empirical literature and presenting some broad world trends. Section 4 reviews the impact of financial development on growth. Section 5 analyzes Chiles experience in terms of financial reform, saving rates and growth. Section 6 concludes. 2 - What do we know about saving in the world? Policy makers and academics have a long-standing interest in aggregate saving. Should higher saving be considered a policy priority? If the answer to the latter question is positive, which are the main determinants of saving, and which role could policies play in boosting saving? How is saving related to growth? To start with the latter question, there is strong evidence of a positive and robust relationship between saving and growth rates. Countries that save more also tend to grow faster (Schmidt-Hebbel and Servn, 1999). But causality between saving and growth could go in either direction and, moreover, both variables may be simultaneously determined by third factors. However, international evidence tends to suggest that Granger causality (in the sense of time precedence) goes from growth to saving rather than the other way around. Yet the latter dominant causality should not lead to dismissing the role of saving, as the saving-growth relationship could become a virtuous cycle, with more saving spurring growth to some extent, which in turn would further raise saving.

Moreover saving is often considered beneficial for its financial dimensions. In open economies, raising national saving is a way to reduce the dependence on foreign saving, protecting the economy from external shocks. This is an important policy concern in a world of increasing financial integration. Together with a strong and well-capitalized financial system, saving represents a form of self -insurance to reduce the economys vulnerability to unexpected reversals of international capital flows. In this manner, saving can help reduce macroeconomic volatility, which empirically has been shown to hamper growth (Ramey and Ramey 1995; Fats 2000). Thus either because of its direct association with growth or its indirect association through the role of financial buffer, the concern with saving seems to be justified when looking at the international experience. Countries that have saved more have often benefited from higher growth and have proved more resilient when facing international financial crises. 2.1 Stylized facts about saving in the world Loayza, Schmidt-Hebbel, and Servn (2001) summarize the main results of the World Banks Saving Across the World project, the largest and most recent research project on the behavior of saving worldwide. They report seven stylized facts reflected by the data of the World Saving Database. 1 While some of these facts deal with time trends and regional differences in saving rates (beginning with the gross national rate), other assess simple correlations between private saving rates and related macroeconomic variables. In this paper we use the latter database that covers through 1994, complemented by more recent national saving data through 2000 gathered by Shankar (2002). The world gross national saving (GNS) rate has declined since the 1970s (Figure 1). After standing at 20.6% in 1967-73, the world GNS rate fell to 20.0% in 1974-1982 and declined further to 18.3% in 1983-1992, with a marginal increase to 18.5% in 1993-2000. GNS rates differ significantly among regions. In industrialized economies, the median GNS rate rose between 1960 and the mid-1970s, prior to the first oil shock. Saving rates have declined since, reaching an all-time low in the latter years of the data series. A similar trend is observed in developing countries, where saving rates are consistently lower than in industrial economies. Saving rates in Latin America have remained stagnant, only above the levels observed in Sub-Saharan Africa. Saving rates in APEC countries have grown steadily to exceed the developed countries saving rate during the 1990s (Figure 2).

The main issues addressed by the project included an exploration of the significant differences between saving rates worldwide, a clear assessment of the alleged relationship between saving and growth, and an evaluation of the main determinants of saving, as well as the potential role of specific policies in promoting national saving rates. The projects database represents a major improvement over previous information. It covers a significant amount of data, both in the time and cross-country dimensions, covering a maximum of 35 years (from 1960 to 1994) and 112 developing and 22 industrial countries. Using complementary information for national saving data assembled by Shankar (2002), this paper extends the national saving rates up to 2000. Although the national saving rates calculated from 1994 to 2000 are not strictly comparable to those for 1960-1994, we are confident that they are useful for pointing out the main trends and correlations up to 2000.

Public saving has declined in industrial economies since 1975, but has been rising in developing countries since the early 1980s. Private saving has remained roughly constant in industrial countries, while it declined and then recovered in developing nations (Figure 3). Figure 4 presents a scatter plot of private saving rates in the 1970s and 1990s. The figure reveals both a large degree of diversity in saving rates across countries and a remarkable degree of persistence in saving rates across decades. Saving rates and per-capita income levels are positively correlated. The combined time-series and cross-country correlation between the gross private saving rate (as a share of gross private disposable income) and per-capita income is 0.52 and is highly significant. The correlation is even higher for developing countries. Saving rates and income growth rates are also positively correlated. Although the relationship is weaker than that observed for income levels, economies that grow faster on average also save more. For the world sample, correlations are 0.36 (using cross-country data) and 0.21 (combining time series and cross-country data) and both are significant. Finally, saving rates and investment rates exhibit a positive correlation, confirming the Feldstein-Horioka (1980) puzzle. Cross-country correlations between GNS and investment are 0.71 for the whole sample, 0.72 for developing economies and 0.47 for industrialized economies. Correlations are smaller for private saving and investment rates but remain significant. 2.2 The determinants of saving The above stylized facts on world saving are based on simple correlations and hence could be misleading as they do not control for other determinants. Moreover, they provide no basis for any assessment of causality. For this we need to resort to empirical analyses using a multivariate framework. Table 1 lists a number of saving determinants pointed out by theory and for which empirical evidence has been found. The table identifies expected signs of causal effects of individual variables and their signs, as well as identification of significant effects and their signs found in seven empirical panel-data studies of private saving rates for both industrial and developing economies (Masson, Bayoumi and Saimei, 1995; Edwards, 1996; Bailliu and Reisen, 1998), industrial economies alone (Haque, Pesaran and Sharma, 2000), and developing economies alone (Corbo and Schmidt-Hebbel, 1991; Dayal-Ghulati and Thimann, 1997). The table also presents the results obtained by Loayza, Schmidt-Hebbel and Servn (2000), who conduct a comprehensive study using the information contained in the World Saving Database. Although the empirical studies differ widely in specification, data coverage and definitions, and estimation techniques, many of the estimated coefficients are consistent among them and with theoretical priors. The variables for which consistent and statistically significant coefficients are found include the terms of trade, foreign credit restrictions, fiscal policy variables, and pension system variables. Results are ambiguous for those

variables for which theory is itself ambiguous, such as income growth or interest rates. Some variables for which the theoretical priors are clear also exhibit differences across different studies, like income, inflation, and dependency ratios. The evidence presented in Table 1 allows to focus now on two main issues: the determinants of saving and the role of specific policies in affecting the level of saving rates. Persistence Saving rates exhibit significant inertia, with a high serial correlation even after controlling for other relevant variables. Thus, the full impact on the saving rate of a change of its determinants only occurs after several years. Specifically, long run effects are almost twice as large as the short run (less than a year) impact. Income A positive correlation between income and saving is found when empirical equations test income as a determinant of private savings. As discussed earlier, the effect is stronger in developing countries, and disappears in middle and upper levels of income. Loayza, Schmidt-Hebbel and Servn (2000) estimate that doubling per capita income raises long term saving rates by 10%. Thus, policies that foster income growth are an indirect but effective manner to affect savings. Related elements include the effect of income distribution (no significant effects of income concentration are found), as well as the distinction between temporary and permanent shocks (although an extreme version of the permanent income theory is rejected, the impact on saving of a temporary income shock is stronger than the one caused by a permanent change). Growth A simple version of the permanent income hypothesis predicts a negative impact of higher income growth on current saving. The relationship, however, is more ambiguous under a life-cycle framework. Moreover, and as discussed earlier, reverse causality might be at work, with saving affecting the growth rate through increased capital accumulation. Consistently with what was described in the stylized facts, a positive relationship between both variables has been largely reported. However, the interpretation still remains controversial, with some authors claiming that growth induces saving (Carroll and Weil, 1994) and others defending the view that saving leads to growth through its impact on investment (Levine and Renelt, 1992). Loayza, Schmidt-Hebbel and Servn (2000) analyze this issue once again, testing for the impact on income growth of the level of private savings. Their results indicate that a one- percent increase in the growth rate increases savings by a similar magnitude, although the effect may be temporary.

Attanasio, Picci and Scorcu (2000) find that growth Granger-causes saving, although the effect seems rather weak. They also report that increases in saving rates do not always precede increases in growth, and that a negative relationship exists between lagged saving rates and the current level of income. This could reflect that agents will save if they anticipate a negative shock that may affect their income in the future). Deaton and Paxson (2000) study the relationship between savings and growth using household data, and they find that the causality runs from income to saving. Rodrik (2000) finds that increases in saving that last in time are followed by persistent increases in growth that die out after some years, while persistent increases in growth lead to permanent increases in saving rates. Demographics The life-cycle hypothesis is based on the age of individuals, predicting that saving follows an inverted-U shaped path. Although the evidence suggests that the theory is unable to account for some of the empirical findings (see Loayza, Schmidt-Hebbel and Servn, 2001 for further description), it does find some patterns that are consistent with it. An increase in the dependency rate of the young and the elderly is associated with a reduction in saving, implying that countries that are in a demographic transition could experience significant movements in their saving rates over time. Uncertainty In the face of higher uncertainty, risk-averse agents would increase their saving rates as a precaution to hedge themselves against possible adverse shocks in the future (Skinner, 1988, Zeldes, 1989). Uncertainty can explain the close relationship between consumption and income observed among young agents facing a positive, yet uncertain path of future income. The same applies to excessive saving by the elder, uncertain about their health-care expenses and the time span left until their death. Preliminary empirical evidence (Carroll and Samwick, 1995) suggests that precautionary motives can indeed be an important determinant of saving rates. Among the results in Table 1, macroeconomic uncertainty (measured by inflation) is found significant only by Loayza, Schmidt-Hebbel, and Servn (2000). 2.3 Can policies promote paving? A related issue is the role played by policy variables. What does the evidence say on the effect of specific policies on the saving rate? Public saving Ricardian equivalence states that, given a certain set of assumptions, an increase in public saving is fully offset by an equivalent reduction in private saving, thus leaving the GNS rate unaltered. The empirical evidence shows that this offsetting is only partial, however. , and this implies that public saving is one of the most powerful, direct tools that can be used to

increase the level of national saving. However, the magnitude of the impact differs widely among estimations, and also among countries. For example, while Loayza and Shankar (2000) find a compensation coefficient below 30% for India, Burnside (1998) finds figures above 80% for Mexico. Lpez, Schmidt-Hebbel, and Servn (2000) present international evidence that points out at liquidity constraints, rather than finite horizons, as the main explanation for the absence of full Ricardian equivalence. The international evidence also suggests that increasing public saving through an increase in taxes has a weaker impact on national saving than a similar change obtained through a reduction in public spending (Lpez, Schmidt-Hebbel and Servn, 2000). The evidence regarding a popular instrument used to boost savings tax exemptions for specific financial instruments is far from conclusive. Estimated effects are heterogeneous, a result of the theoretical ambiguity regarding the effect of return rates on saving decisions (more on this will be discussed on the next section). Direct evidence shows that results from tax incentives to voluntary saving instruments for the old-age are mixed and, if positive, small. Pension reform The replacement of pay-as-you-go pension systems by fully -funded systems based on individual capitalization has been defended as a saving-enhancing reform. However, an analytic evaluation suggests that the impact may vary, depending on how the transition deficit is financed, and on the efficiency gains associated to the new system. In the long run, the reform can impact savings through its mandated contributions, which could raise the saving rates of lower-income workers above their desired levels. In such a scenario, it is doubtful that the increase in saving involves a positive welfare effect. However, funded pension systems can have positive effects besides their direct impact on savings, enhancing growth and income by reducing labor distortions and contributing to the development of capital markets. Empirically, countries with fully funded systems seem to save more than those with pay-as-you-go schemes. Schmidt-Hebbel (1999), analyzed the experience of Chile (the first country to reform its pension system) and found that more than 30% of the 13% increase in the saving rate observed since 1986 can be attributed to the pension reform. However, Samwick (2000) finds that savings increased only in Chile, and not in the other four countries included in his study. However, he also finds, using cross-country data from 94 countries, that PAYG systems have a negative impact on saving. External debt The impact of foreign saving on national saving is still unclear. One problem that complicates the evaluation of this issue is the simultaneity between both variables. Controlling for that problem, Loayza, Schmidt-Hebbel and Servn (2000) find that an exogenous increase in external debt reduces private saving by roughly half the increase.

One major policy affecting saving rates is still left financial liberalization. As it constitutes the central topic of this paper, the next section is devoted exclusively to analyze it. 3- Financial liberalization and saving Financial liberalization grants market forces a dominant role in setting financial asset prices and returns, allocating credit, and developing a wider array of financial instruments and intermediaries. All these changes are aimed at improving the efficiency of financial intermediation, (possibly) raising saving and investment, improving the efficiency of investment, and spurring growth. However, financial liberalization is not risk-free and should be carefully implemented to attain its benefits. Excessively rapid financial reform often leads to unsustainable credit and activity booms, which then lead to financial crises. These risks increase significantly in the absence of prudential regulation and strong supervision of banks and other liberalized capital market segments. Although this issue goes beyond the scope of this paper, it should be kept in mind when analyzing the potential impact of financial reform on saving and growth. Section 3.1 analyzes the various channels through which financial reform can affect the saving rate, both in the short and long run. Section 3.2 presents the empirical evidence that we obtain from cross-country correlations using data on saving rates and financial liberalization. Section 3.3 surveys the empirical literature on the links between saving and financial reform. 3.1 Theory From a theoretical perspective, financial liberalization, and the ensuing increase in financial development, has an ambiguous effect on the level of saving, which depends on the net impact of several simultaneous channels. Furthermore, as discussed by Bandiera et al. (1999), the long-term effect of financial liberalization on saving can be significantly different from the impact effect observed in the aftermath of financial reform. A financial reform typically comprises several key phases, often separated by several years. Thus an evaluation of the potential effect of financial liberalization on national saving implies analyzing the different channels through which this impact can take place, as well as distinguishing between short and long-term effects involved in the transmission process. Financial reform, translated into more developed financial intermediation, alters significantly a countrys financial system. The reformed system is typically characterized by higher interest rates, fewer credit constraints (or a reduced share of agents subject to them), increased saving opportunities, a larger portfolio of investment instruments, and a larger number and diversity of financial institutions. The combined effect of the latter changes on saving is uncertain, as pointed out by the theoretical and empirical literature.

Two effects can be distinguished: a direct short-run effect, that reduces saving, and an indirect long-run effect that boosts saving. The direct effect can be split into two parts. A credit channel related to the impact of financial reform on interest rates and a quantity channel involving availability of credit. We shall explore each in turn. Together with allowing households to consume at the optimal level determined by their life-cycle position, a larger supply of consumer credit could lead them to revise their precautionary saving levels. This would allow young households that had been constrained to consume more than they would over a full unconstrained lifetime. The two latter effects suggest that saving can fall temporarily below its steady-state level, leading to a temporary consumption boom (Muellbauer, 1994). In the long run, however, saving could be raised not directly through structural changes to the financial system, but by one of the byproducts of financial development: GDP growth and higher levels of income. We briefly analyze some of the channels through which changes in the level of financial development and intermediation may affect short and long-term saving levels. Interest rates The rate of return at which financial resources are transferred in time is an obvious reference point in the analysis of the impact of changes in the financial environment on households saving decisions. However, and as noted above, the sign of the interest rate elasticity of saving is ambiguous, both theoretically and empirically. Although optimization models suggest that saving could be sensitive to interest rates, the existence of opposing effects does not lead to a clear conclusion regarding the sign of such elasticity. Higher interest rates increase saving through the substitution effect, but could ultimately reduce the saving rate if the associated income and wealth effects are sufficiently strong. This theoretical ambiguity has not been solved, and the direction of the response of aggregate saving to an exogenous increase in the interest rate still remains vastly controversial. Financial reform has been equated to a shift towards higher interest rates, as it implies reducing the constraints on the level of interest rates that go along with financial repression. However, the theoretical controversy stated above forbids building a strong case for an increase in saving associated to the increase in the rate of return for savers. If anything, a review of the literature (see, for example, Fry 1995) reveals that the number of positive elasticities found in the studies exceeds the number of negative ones, but that both types of coefficients are generally small and non-significant. Honohan (1999) suggests that the empirical puzzle could imply the existence of different elasticities across countries, depending on each nations specific characteristics, such as per capita income, growth rates, or levels of indebtedness. Another possible complexity of the relationship between interest rates and saving is the existence of non-linearities, associated to non-linear effects on the Euler equation of intertemporal consumption allocation and on the influence of the interest rate on the households wealth. From a macroeconomic perspective, the level of interest rates might

not only be associated with the extent of liberalization and financial opportunities, but also reflect the existence of political uncertainty or macroeconomic instability. The controversy on the empirical impact of interest rates may be related to the omission of relevant saving instruments. Domestic interest rates may misrepresent the opportunity cost of consumption, if a wide array of financial and non-financial assets is available for saving. Honohan (1999) indicates that a complete account should analyze not only the behavior of financial interest rates, but also the rate of return of owner-occupied housing and other real estate investments. As discussed by Koskela and Viren (1994), financial reform ha been associated to significant property booms. The rates of quoted equities, bonds, and foreign-currency denominated assets should also be considered. Lifting credit constraints One of the main channels through which the financial sector can affect borrowing is the extent to which agents are allowed to borrow. As long as households are not permitted to borrow as much as they would wish from an intertemporal maximization perspectiveor if they are not allowed to borrow at all-changes in credit conditions will necessarily shift actual saving rates, allowing agents to behave closer to their maximization schemes. One of the methodological approaches to test for the existence of credit constraints, and their impact over households consumption and saving decisions, has been to look at the role of current income in the intertemporal allocation of consumption, as dictated by the Euler equation. If agents are able to freely transfer resources subject to their intertemporal income restriction, lending and borrowing would be used as devices to smooth out consumption in the presence of temporary income variations. As long as there is no uncertainty, and no wedge between borrow ing and lending rates, consumption should depend solely on interest rates, not on the behavior of current income. However, and as discussed earlier, current income does appear as a relevant variable in estimated saving functions. The significance of current income in explaining consumption is seen as a result of the incapacity of agents to borrow as much they would chose according to their welfaremaximizing optimization (see for instance, Campbell and Mankiw, 1989,1991; Zeldes, 1989). 2 In practice, liquidity constraints can be associated with two distinct phenomena. The first is the large gap between borrowing and lending rates. In this scenario, the household is not directly constrainedit has the potential of acquiring debt if it chooses to but the premium (relative to deposits) of doing so is so high that it chooses to consume only current income. It is evident that, as long as financial development leads to higher competition among financial intermediaries, interest rate gaps (associated to abnormal returns) should decrease, thus leading households to borrow more. This result was found by King (1986) for the UK. Alternatively, the household's inability to borrow at wholesale market interest rates may be a rationing phenomenon, whereby households are directly unable to borrow at any rate (or can do so only at a rate that approaches infinity).
2

Strictly speaking, it could also suggest that any of the other assumptions involved in deriving the Euler equation might be wrong.

10

Thus, as financial reform relaxes credit constraints, increasing the availability of borrowing alternatives, in the aggregate it might reduce rather than increase private savings (Muellbauer and Murphy, 1990, Jappelli and Pagano, 1994). One last element to be kept in mind is that credit constraints can affect saving in two ways. The first is the one discussed herein, that current constraints prevent agents from borrowing as much as they would, thus forcing them save more than they would freely choose to. The second one involves the risk of facing credit constraints in the future, even if the household is currently free to borrow. The possibility of being constrained in the future could reduce current consumption through an increase in precautionary saving. Raising saving opportunities Financial development can imply an increase not only in the availability of credit funds and instruments, but also in saving instruments and devices. A deeper financial system should be capable of providing alternative saving instruments that more adequately match individual preferences, risk-aversion and income profiles. Credibility in the soundness and sustainability of banks, for example, should increase the incentives for households to participate in the financial system. The risk of expropriation or default can severely harm household trust in the financial system, as the Argentinean crisis has cruelly exemplified. As mentioned above, the evolution of saving opportunities should not only include financial instruments such as bank deposits, but also the developments of markets for savings in the form of physical assets, durable goods, foreign-currency denominated instruments, and the like. A related process refers to the evolution of convenience, as discussed by Honohan (1999). Apparently simple developments associated with financial development, such as the geographical location of bank branches,3 queuing time at bank offices , or minimum deposit requirements can play a significant role in the willingness of persons to deposit their savings in the financial system. 4 However, the latter developments are more likely to affect the allocation rather than the volume of saving. Portfolio diversification The increase in the range of available financial instruments as a result of higher financial development will not only imply a wider set of saving instruments, but also a broader portfolio that will allow for diversification and risk-hedging. If that is the case, risk-averse agents can reduce their exposure to shocks that would otherwise threaten their
3

Fry (1986, 1988) provides some evidence of the positive impact of an increase in the density of rural bank branches on saving rates. Again, in the absence of proper financial instruments to save, people would probably resort to alternative saving devices as durable goods.

11

income and wealth levels. This reduction in uncertainty and exposure should reduce the level of precautionary saving. International credit flows The process of domestic portfolio adjustment can not only can lead to transitory decreases in the volume of domestic saving due to an increase in available credit, but it may also induce large capital inflows. The liberalization of the dome stic financial system has typically been only one element in a package of reforms that have been associated with these inflows, and the inflows have proved to be easily reversible. The impact on saving comes through the associated changes in credit availability and costs, revised expectations of income growth, and increases in financial wealth, especially due to upward movements in real estate prices. All this may lead to consumption booms and to a fall in the saving rate. Higher growth If the steady-state effect of financial reform leading to increased financial development is associated to higher growth rates, the positive relationship between saving rates and growth discussed in the previous section should imply an increase in saving in the long run. Many studies (King and Levine, 1993; Levine and Zervos, 1993, 1998; Levine, Loayza and Beck, 2000) suggest that this is indeed the case. This effect could manifest itself through two channels: financial development could encourage investment, thus directly affecting growth rates by enhanced factor accumulation; alternatively, financial intermediation keeping investment constant, could allow for a better allocation of resources, thus increasing growth by speeding up total factor productivity for a given rate of factor accumulation. More discussion on this issue will be presented in the next section. Higher income and wealth If financial development increases growth, then it also increases income prospects for the economys households. As discussed in the pre vious section, evidence of a positive link between income and saving rates exists, although the link virtually vanishes at high income levels. Theoretically, the impact on saving should depend on the timing of the increases in income. If the beneficial effect of the financial reform has a comparable impact on both current and future income, the effect is akin to a rise in the level of permanent income that should increase consumption accordingly, with the saving rate staying unaltered. If, however, income is expected to increase in the future, the rise in the level of permanent income would lead to an increase in current consumption even if current income has not risen yet, thus reducing saving temporarily. An additional element here is precautionary saving, through which the increase in wealth would further reduce the saving rate. 3.2 Simple evidence on the relationship between saving and financial liberalization Bandiera et al (2000) have developed a synthetic measure of financial reform for eight developing economies. The authors date major changes in the degree of financial

12

liberalization, yielding zero-one variables for each country, and deriving the principal components of financial reform. 5 More recently, Abiad and Mody (2002) have built an annua l index of financial reform for 35 countries for the period 1973-1995. Their focus is not on analyzing the impact of financial reform on macroeconomic performance but rather on the events and developments that trigger and support reform over time. The database, consistent with the discussion presented in the previous section, involves a multi-dimensional measure of financial liberalization. 6 Using this database in combination with World Saving Database, we estimate some simple correlations between financial liberalization and saving. Figure 5 depicts a scatterplot of financial liberalization and saving rates for the 35 countries included by Abiad and Mody (2002). We find a positive, significant relationship between financial reform and saving (0.43). A higher degree of financial liberalization is observed in countries that on average save more, although simple association does not reveal anything about the direction of causality e.g., countries that save more might feel more inclined to adopt financial reforms, or both saving and financial reform might be related to the income level of each country. In any case, a quick glance at the data in figure 5 provides no evidence of a negative association between the degree of financial liberalization and the private saving rate. The results hold true when the simple correlation between the time evolution of liberalization (measured as the change in the average 1974-84 vis a vis the 1984-93 index) and the time evolution of saving are analyzed (Figure 6). This provide s evidence additional to the cross-country dimension, suggesting that countries which deepened financial reform on average exhibited a (marginal) rise in their saving rate (correlation: 0.17). Note, however, that neither of these correlations is statistically significant. Also confirming the formal evidence derived by Abiad and Mody, there is a positive and significant correlation between the level of financial liberalization across decades (0.55). Does financial liberalization translate into financial development? This can be verified by examining the association between the above index of financial liberalization and standard indicators of financial development. The latter can be obtained from the Beck et al (1999) database. The authors provide various definitions of financial development, using indicators such as the relative size of central bank assets and deposit money assets, the ratio of private credit to GDP, and the share of liquid liabilities.
5

Including exogenous changes in interest rate regulation, reserve requirements, directed credit, bank ownership (privatization), liberalization of securities markets, prudential regulation, and international financial liberalization. 6 The index includes six measures: the use of directed credit and reserve requirements, the existence of interest rate controls, the existence of entry barriers and regulations that undermine competition, the existence of operational regulation and/or the non-existence of prudential regulation, the share of public property in the financial sector, and the extent of controls on international financial transactions. For each variable, the authors classify individual countries into four categories, ranging from full repression to full liberalization. Based on their data, the authors describe stylized facts regarding the process of financial reform, particularly the existence of policy reform inertia.

13

A positive association, reflected by a correlation coefficient of 0.68 that is significantly different from zero, is observed between financial reform and the level of financial development (defined as private credit to GDP). Figure 7 illustrates the positive correlation with a scatterplot for both variables. This should not come as a surprise, as increased financial development and intermediation should be the outcome of financial reforms. This positive association between the reform input and its outcome is exploited in the following section, which covers the literature on the effects of financial development on growth. Additional evidence, for a broader set of countries, can be obtained by combining the financial development information from the Beck et al. database with the World Saving Database. Figures 8, 9 and 10 present scatterplots of financial development and private saving rates (1980s and 1990s averages), as well as their first-differences. Simple correlations for this broader dataset provide a picture similar to the one obtained with the financial reform information. A positive, significant correlation between financial development and private saving is observed in a cross-country comparison (0.52 for the 1980s, 0.44 for the 1990s). However, unlike the result obtained with financial reform, the correlation between the first-differences of private saving and financial development is insignificant (0.04), suggesting that in the dataset of 100 countries, increases in the level of financial development were not associated to a significant change in domestic saving rates. 3.3 Empirical analysis of the effects of financial reform on saving Although reforms have differed in their speed, depth and permanence over time, the average level of financial liberalization has certainly increased across the world. This trend towards financial liberalization has brought about, as a natural consequence, an increased interest in understanding its determinants and main characteristics, as well as its potential impact on macroeconomic variables such as saving, investment and growth. If a general consensus could be derived from the existing literature, it would be summarized as follows: while financial development appears to be a relevant determinant of growth, its effect on saving is unclear, although some evidence suggests that it is negative. This section will review existing studies of the impact of financial reform and development on saving, as well as provide some new evidence using the World Saving Database. The next section will review the evidence regarding the impact of financial development on growth. In the empirical literature, it is hard to disentangle the effects of financial liberalization from those of financial development. In other words, while some studies directly test for the effect of financial reform indexes on the level of national saving, other works rely on the impact of different measures of financial development. Many of these studies focus mainly on the effect of a relaxation of credit constraints. Although both variables should be related (financial liberalization should lead to enhanced financial intermediation), the correlation between both variables is not perfect,

14

and significant differences in timing may exist. These caveats should be kept in mind when analyzing the evidence. Most empirical studies have involved adding one or more variables to standard econometric specifications of saving or of the rate of change in consumption. Some studies simply include dummies identifying pre and post liberalization periods (an early example is de Melo and Tybout, 1986, for Uruguay); while others have relied on a linear trend reflecting the evolution of gradual liberalization (Muellbauer and Murphy, 1993 for the UK). Studies that have assessed the impact of financial de velopment (rather than reform) have resorted to proxy variables such as the volume of consumer credit (e.g. Jappelli and Pagano 1989, 1994). Ostry and Levy (1995) used this variable both independently and in interaction with an interest rate, finding that financial development caused by liberalization lowered saving rates in France, but reversed the negative association between saving and interest rates into a positive one. Bayoumi (1993) found a similar result for the UK. Miles (1992) found that liberalization of the mortgage market in the UK led to a significant decrease in saving. Loayza, Schmidt-Hebbel and Servn (2000) find that a 1% increase in the flow of private credit relative to income reduces the long-run saving rate by 0.74%. An easing of credit market conditions for households that led to a drop in saving was also detected for the 1980s in Scandinavian countries by Koskela et al (1992), and Lehmussaari (1990), Here the effect on saving came indirectly through the impact of increased housing credit on the price of new homes. In their 30-country study, Jappelli and Pagano (1994) found another expression of financial development to be highly significant in explaining the evolution of saving. The normal loan-to-value ratio obtained from mortgage finance institutions, a measure of credit availability, appears to affect saving significantly: a 15 percentage point increase in the loan-to-value ratio reduces the national saving rate by 2.6 percentage points. This substantial effect may not be entirely housing-related, because the variable may be capturing movements in the availability of a wider credit category. In the same strand of literature that studies the association between financial development and saving through the existence and behavior of credit constraints, alternative proxy measures have been used. They include the percentage of home -owners in certain age groups, the interest rate wedge on consumer and mortgage loans (Jappelli and Pagano, 1989), and the rate of consumer credit delinquencies (Carroll, 1992). Confirming the evidence for industrial countries, Vaidyanathan (1993) shows that international variations in consumption sensitivity to income are positively related to financial depth (measured by the ratio of M2 to GDP), suggesting again the importance of liquidity constraints. Bandiera et al (2000) analyze the experience of eight countries that underwent significant reforms in their financial systems, namely Chile, Ghana, Indonesia, Korea,

15

Malaysia, Mexico, Turkey and Zimbabwe. The y estimate an econometric relationship expressing the private saving ratio as a function of the real interest rate and the index of financial liberalization cited earlier, along with income, inflation and public savings. In addition to directly measuring the contribution of liberalization to the volume of aggregate savings, their procedure improves on earlier estimates of the saving-interest rate relation, which limited the role of financial sector liberalization to the real interest rate channel. Their re sults do not provide a clear answer on the impact of reforms on saving, as the effect appears significantly negative in some cases (Korea and Mexico), positive in some others (Greece and Turkey) and insignificant in the remaining countries. This heterogene ity probably reflects that the weight of the different channels discussed in the previous section differs among countries. In a related study, Loayza and Shankar (2000) use detailed saving information from India to find that financial reform has not change d the saving rate, but shifted the composition of saving towards a higher share of durable goods. As mentioned earlier, this conclusion suggests that the picture provided solely by financial saving measures could be misleading. 4 - Financial development and growth Financial intermediation can provide valuable services along several dimensions. They include the trading of risk, allocating capital, monitoring managers, reducing information asymmetries, mobilizing savings and easing the trading of goods, services, and financial contracts (Levine, 1997). All these services could have a significant impact on growth, which should mainly occur through two channels. First, allowing for higher investment, as a result of increased availability of funds for investment coming from either domestic or foreign savings. Second, through higher total factor productivity resulting from more efficient resource allocation and higher efficiency, due to better information about and access to bank credit and capital markets by previously credit-constrained firms with efficient investment projects. This section first provides empirical evidence on the cross-country association between growth rates and financial development (section 4.1). Section 4.2 then presents a survey of the empirical literature that has analyzed the microeconomic evidence of efficiency and investment effects at the firm level, and then the macroeconomic evidence on the growth effects of financial reform. 4.1 Simple evidence on the relationship between growth and financial development As with saving, the information contained in the Beck et al (1999) database can allow us to provide a simple, statistical description of the relationship between financial development and growth rates in a broad set of countries. Figures 11, 12 and 13 present scatter plots of financial development and GDP growth rates (1980s and 1990s averages), as well as between their first differences. The large country sample used here yields a positive but insignificant relationship between

16

growth and financial development, both in levels and first differences. This is confirmed by simple correlations. The (non-significant) correlations in levels are 0.052 for the 1980s and 0.073 for the 1990s, and 0.044 for the variation in decade averages. 4.2 Empirical evidence Microeconomic evidence Several studies have analyzed the impact of financial liberalization on resource allocation and investment decisions at the firm and industry levels. Jaramillo, Schiantarelli and Weiss (1992) use data from Ecuadorian firms, and find that, after liberalization, the flow of funds to more efficient firms as defined by a CobbDouglas production function applied to panel dataincreased, after controlling for other characteristics of the firm. Siregar (1992) finds a similar result for firms in Indonesia. A parallel strand of literature has covered a related issue, namely the availability of resources for small, previously constrained firms. Although these papers do find that such firms have gained greater access to financial resources after liberalization, they do not directly assess whether those firms are actually more efficient, a necessary condition for the new distribution of funds to imply a better allocation of resources. Rajan and Zingales (1998) use industry-level data to show that industries that require more external financing grow faster if they are based on more financially developed countries. A similar story is presented by Demirguc-Kunt and Maksimovic (1998), who find that firms in more fina ncially developed countries grow at a faster rate, relative to a benchmark growth rate that would hold in the absence of external financing. At a more aggregate level, Wurgler (2000) uses industry-level data, defining financial development as the average size of credit and equity markets relative to GDP. He finds evidence that the rate of investment growth is more related to growth in value added in countries more financially developed. In fact, financial development is associated with higher investment i n growing industries and lower investment in declining industries. However, his analysis is a cross-country comparison, rather than an analysis of the timeevolution of the reform and its effects. Other papers (such as Cho, 1988, for the Korean case) have analyzed the change in the volatility of expected marginal returns to capital across industries, before and after financial liberalization. A decrease in volatility is seen as a reduction in uncertainty that facilitates the process by which flows of capital equate returns. More recently, Galindo, Schiantarelli and Weiss (2002) test for the impact of financial liberalization on the allocation of resources, using microeconomic evidence from developing countries. Does liberalization lead to an enhanced outcome in resource allocation? Specifically, the authors test whether financial liberalization has increased the share of investment going to firms with a higher marginal return to capital. Using firm-level

17

data from 12 developing countries7 , the authors construct an index of investment efficiency, which compares the marginal returns obtained across firms in a given year with a benchmark return that would have been obtained had resources been distributed according to the firms capital share. Their results sugge st that, for most countries, the introduction of financial reform has increased their measured level of efficiency. Macroeconomic evidence Many authors have analyzed the impact of financial liberalization on growth. King and Levine (1993) and Levine and Zervos (1993) show that financial intermediation development is a good predictor of economic growth, even after controlling for country characteristics and growth determinants. Time-series studies confirm that financial development predicts growth (Neusser and Kugler, 1998; Rousseau and Wachtel, 1998). However, rather than finding a causal relationship between growth and the level of financial development, these papers establish that financial development (using various definitions) and GDP growth are positively related, andin some studiesthat financial development is a good predictor of future GDP growth. Strictly speaking, however, these studies do not really establish if financial development causes growth or vice-versa. Beck, Levine and Loayza (1999) try to confront the issue of causality, using GMM dynamic panel estimators and cross-sectional instrumental-variable estimators to control for the biases existent in previous studies of the finance-growth relationship. Their results, for a sample of 74 countries between 1960 and 1995, indicate that the exogenous component of financial intermediary development is positively associated with economic growth. Although this does not fully settle the issue of causality, it proves that the positive link found in previous is not the result of an estimation bias. In a related paper, Beck, Levine and Loayza (2000) use the same database to examine the relation between financial development and what they label as sources of growth. These sources include private saving rates, physical capital accumulation, and total factor productivity. The authors find a large and statistically significant relation between financial development and both real per capita GDP growth and total factor productivity growth, which cannot be attributed to simultaneity bias or country-specific effects. Results are rather ambiguous regarding physical capital growth and savings, as they are not robust to the definition of financial development included in the estimation. Thus, the authors conclude that the impact of financial liberalization on growth is mainly channeled through total factor productivity growth (i.e., better allocation of resources) and not through investment and greater accumulation of physical capital. Bekaert, Harvey and Lundblad (2001) provide additional evidence on the positive effect of financial development (measured as stock market liberalization) on the economys GDP growth. Using data for 35 countries between 1965 and 1985 and growth-accounting exercises, Benhabib and Spiegel (2000) and Spiegel (2001) also try to separate between the effect on factor accumulation and total factor productivity. They find that financial
7

Argentina, Brazil, Chile, India, Indonesia, Korea, Malaysia, Mexico, Pakistan, Philippines, Taiwan, and Thailand.

18

development is linked to both investment rates and total factor productivity. The relative magnitude of the impact through each channel depends on the precise measure of financial development the authors use, although the results are more robust regarding physical capital accumulation, suggesting that the main effect is the quantity of investment, rather than its quality. A subset of APEC countries are included in the latter study, with the results suggesting that financial development has a greater impact on TFP growth among those countries than for the rest of the sample. Finally, some recent empirical studies (Ramey and Ramey 1995, Elbadawi and Schmidt-Hebbel 1998, Fatas 2000) find a significantly negative impact of macroeconomic volatility on growth. To the extent that macroeconomic volatility has been found to be negatively related to the degree of financial development (IDB 1995, De Ferranti et al 2000, Fatas 2001), the implied result is that increased financial development also makes an indirect contribution to growth through reduced macroeconomic instability. 5 - A case study: Chiles experience 5.1 Growth, saving, and investment 5.1.2 Main trends Major changes in policies, economic structure, and performance occurred in Chile during the past three decades. These changes resulted in higher economic growth, supported by similarly higher national saving and investment rates and productivity growth. Economic performance, however, has had its share of setbacks and difficulties. Adverse foreign shocks and domestic policy mistakes have punctuated the trajectory that started with the first structural reforms in the mid 1970s. Neither is the countrys current growth path without risks. During the past four years since 1998, Chiles average annual GDP growth has been 2.4%, significantly lower than the average 7.7% it recorded during the golden age from 1986 to 1997. This downturn suggests that the country still faces formidable policy challenges to lock in high growth for the long haul. Schmidt-Hebbel (2002) provides a detailed description of the evolution of Chiles saving, investment, and growth performance in the last four decades. Figures 14 and 15 present the evolution of these variables, while Table 2 shows a summary breakdown by relevant sub-periods. The selected sub-periods reflect different regimes of policies and performance. The pre-reform era was characterized by increasing macroeconomic instability, worsening economic indicators, and growing restrictions on the free operation of markets (1961-74). Serious macroeconomic stabilization efforts, coupled with deep structural reforms, began in 1974-75, continuing to date at varying speeds and intensities. Growth was low during the 1975-85 reforms, mainly as a result of the intense foreign shocks that sank the economy in a recession. Recovery was superbly strong in the golden dozen years of very high growth in 1986-97. More recently, growth has slowed down substantially in the wake of new adverse foreign shocks (1998-2001).

19

It is no surprise that considerable attention has been given to Chiles golden age 8 . The impressive boost in growth rates, accompanied by a similar increase in saving and investment rates, has been widely studied, among others by Gallego and Loayza (2002), Beyer and Vergara (2002), Chumacero and Fuentes (2002), and Schmidt-Hebbel (2002). While the average growth rate rose from a low 2.0% in 1975-85 to 7.7% in 1986-97, the gross national saving rate jumped from 9.6% to 20.8%, with the gross domestic investment rate also increasing from 16% to 24%. One fact noted by most authors is that factor accumulation only provides a partial explanation for Chiles economic expansion. The increase in capital and employment is unable, under a Cobb-Douglas production, to explain the magnitude of the countrys growth jump. Indeed, and as reflected in Table 2, a simple growth accounting exercise suggests that the increase in the growth rate was accompanied by a significant increase in total factor productivity. Rather unsurprisingly, the behavior of saving and investment has been highly procyclical. Positive, significant correlations exist between the growth rate and the evolution of both saving and investment rates. 5.1.2 The effect of reforms on growth As mentioned earlier, the main interest of recent studies on Chiles economic growth has been to find the key determinants of Chiles high rates in the golden period of 1986-1997, in comparison to its performance recorded from the 1960s through the mid1980s (or to the relevant cross-country experience). A review of recent studies of the determinants of Chiles growth is provided by Schmidt-Hebbel (2002). An element that has been stressed in the literature is the role that the structural reforms played in boosting the countrys growth rates, either through factor accumulation or through TFP (or both). While some papers have focused on specific reforms (Barro 1999, Jadresic and Zahler, 1999), others have relied on the impact of overall reforms (Gallego and Loayza, 2000; SchmidtHebbel, 2002). In the former ones, the growth increase in Chile is explained as the combination of a number of specific reforms. In fact, the 5% increase in average growth between 1986-1997 and 1961-1985 can be explained as the combined effect of trade reform (increasing growth by 1.1% in Rojas et al, 1997), the reduction in government size (increasing growth in 0.8% in Barro 1999), more political rights (an increase of 1.6% in Jadresic and Zahler 2000) and tax reforms (an increase of 1.4% in Bergoeing et al 2001). However, the selection of determinants is somewhat arbitrary, as many variables could potentially have a positive impact on either TFP or factor accumulation. With this caveat in mind, in the next section we will present the evidence provided by this type of studies on the specific impact of financial reform on growth. The encompassing approach is presented in this section. Reform is assessed from a broader perspective, in which specific policies are not isolated shifts in particular markets, but part of a consistent trend throughout the economy.
8

Gallego and Loayza (2002).

20

As mentioned, two recent studies provide this type of analysis . Gallego and Loayza (2002) analyze the impact of what they define as policy complementarities, the combined impact of overall reforms (a 1.1% increase in GDP growth, net of the impact of specific policies). Their main conclusion is that an overall set of consistent policies has a higher impact on growth than the sum of the individual impacts of specific reforms. Thus, if financial reform is beneficial for growth, its positive effect will be enhanced if the shift towards higher financial liberalization is accompanied by consistent shifts in areas such as trade, taxes or technology. Gallego and Loayza (2002) also find an interesting result regarding growth and investment. Using growth accounting, investment obviously appears as an important determinant of growth, as it implies capital accumulation. Thus, all growth studies find that the increase in the investment rate played a significant role in explaining growth through factor accumulation, and that further increases in the investment rate would increase current growth. However, investment is not exogenous. The important element is what determines investment, and if investment is itself endogenous to the growth process. Loayza and Gallego use a VAR approach to provide some evidence of this direction, but find no significant impact of investment and saving on growth. Moreover, growth has a significant impact on investment, but not on saving 9 . Bravo and Restrepo (2002) find a similar impact, with the evolution of GDP as a significant determinant of investme nt functions. Returning to the evolution and impact of overall reform, Schmidt-Hebbel (2002) uses a simple reduced-form model to analyze the impact on per capita income of two policy indexes: a macroeconomic index (MAC), basically measuring the behavior of inflation, and a microeconomic index of overall reform (MIC)including financial reformupdated from Morley et al (1999). The corresponding time series for MIC and MAC are depicted in Figure 16, reflecting the massive improvement recorded by both indexes from the early 1970s to date. Both indexes appear significant in explaining the evolution of Chiles per capita income between 1960 and 2001. The author uses his estimated model to simulate the evolution of Chiles growth prospects in different reform scenarios. If no further microeconomic reforms were adopted, Chiles aggregate GDP growth would decline quickly to rates below 3% per year toward the end of the 2002-2015 period. 10 Further micro-structural reforms at the pace of the 1990s would help to slow down the decline in growth rates. If reforms were adopted at the average pace of the last 27 years, GDP growth would exceed no-reform growth by almost 1 percentage point toward 2015. Thus, recent evidence suggests that a significant part of Chile growth miracle was explained by the countrys overall reforms. This is an important lesson to be learned by the designers of economic policies and reform worldwide: coherent, overall efforts augment the beneficial effects of specific reforms, financial liberalization among them.
9

This somehow contradicts the saving literature discussed below. This is not strictly a growth projection, but rather a base case for comparison.

10

21

5.1.3 Saving and its determinants The boost in public and private saving rates observed since the 1980s has received considerable attention in the empirical literature. Figure 17 shows the evolution of both public and private saving rates since 1960, taken from the database constructed by Bennett, Schmidt-Hebbel and Soto (2001). Several authors have recently studied the determinants of Chiles saving rates from a macroeconomic perspective (Hachette, 1998; Morand, 1998; Vergara, 2001, Bennett, Loayza and Schmidt-Hebbel, 2001), discussing the Chilean saving miracle. Table 3 summarizes some of the major macroeconomic studies of the determinants of Chiles private saving rates. Most authors, with the exception of Vergara (2001), suggest, in line with our opening section, that the high average GDP growth rate observed in the same period played a significant role in explaining this result. A positive relationship between saving and the level of current income is also typically found. The authors also find that the impact of certain specific variables was important. Hachette (1998), Morand (1998) and Vergara (2001) suggest that foreign credit restrictions have a positive impact on the saving rate. Regarding interest rates, only Vergara (2001) finds a positive, significant effect, while the remaining studies conclude that this variable does not affect saving significantly. Evidence on the existence of Ricardian equivalence (at least partially) is reflected in the negative, significant effect of public saving on private saving. The evolution of Chiles saving rates has also been analyzed from a microeconomic perspective. Coronado (1998) and Butelmann and Gallego (2000) analyze the determinants of saving at the household level. Their main results are summarized in Table 4, assessing the role of variables such as age, education, income level, and the pension system. 5.2 Financial reform and development in Chile: A brief chronicle Figure 18 presents the liberalization index constructed by Abiad and Mody (2002), to depict the evolution of Chiles financial policies (we extend the index up to 2001 to include the elimination of capital controls). After being an extremely repressed system since the mid-1970s, aggressive financial reform measures were taken. On the domestic realm, successive measures eliminated controls on deposit and lending rates, significantly reduced reserve requirements, diminished entry barriers, abolished credit ceilings, and privatized banks. Regarding international financial markets, the late 1970s saw the elimination of the ceilings in foreign borrowing and of most restrictions on capital movements. After a severe banking crisis in 1982, a partial reversion was experienced, involving the placing of non-solvent banks under special government administration and a toughening of restrictions on foreign capital flows. However, domestic restrictions were relaxed in the mid-1980s, accompanied with a new legal framework providing prudential regulation and strengthening the supervisory system.

22

Restrictions on capital movements were gradually relaxed during the 1990s, to be fully removed in 2000. Gallego and Loayza (2000) provide a detailed description of the evolution of Chiles financial system in different dimensions. Since the 1970s, Chiles financial system has consistently grown in size relative to GDP. The banking sector experienced significant growth in the 1970s, with a more moderate expansion in the last two decades. The bond market expanded substantially since 1980s, while the stock market experienced a huge development in the 1990s. Thus, the financial system experienced a relevant shift not only in terms of its size, but also in terms of its composition, with different markets experiencing temporary booms that were later reversed (for instance, the banking crisis of 1982-83 partially reversed the credit expansion of the previous five years). Figure 19, constructed using the data set developed by Beck, Demigurc-Kunt and Levine (1999), depicts the evolution of financial development in Chile, compared to international standards (a group of countries of comparable income level). A story similar to the one regarding financial reform applies here: after being significantly underdeveloped (relative to the average development in higher middle-income countries). Chiles financial system experienced a rapid expansion in the late 1970s, a partial reversion in the mid1980s 11 , and an upward trend ever since, remaining consistently above the average. This is reinforced by Gallego and Loayza (2000), who analyze how the banking system development is above what they label a development line, basically the world average of the Beck et al (1999) database. Regarding efficiency, Gallego and Loayza (2000) find that gross banking markups fell significantly until the 1990s, after which they have remained relatively stable. Basch and Fuentes (2000) suggest that stability in markups does not imply a stagnation in efficiency, but rather show the higher competition that banks have faced in the form of other financial intermediaries, which has implied an increase in overall costs. A particular feature of the Chilean banking system, which should have a significant impact upon the economy, is its level of overall soundness. Chiles financial indicators more than fulfill Basel standards, with a low percentage of non-performing loans. This condition diminishes the risk of a financial crisis (a relevant issue in the midst of significant regional turbulence). However, the banking system is only one component of the overall financial system. The stock market also plays a significant role, and has evolved quite differently from the banking sector. After a gradual expansion in the 1970s and 1980s, growth accelerated only during the 1990s, with its size reaching a maximum of 111% of GDP in 1995. Notice, however, that Chiles stock market was below the international average before the 1990s. As discussed by Gallego and Loayza (2000), alternative measures of stock market
11

As noted by Gourinchas, Landerretche and Valds (1998), this reversion after the banking crisis was probably a correction of unsustainable credit expansion. This highlights that temporary shifts in nave financial development indexes may not indicate a change in the level of development, but rather fluctuations associated to short-run events.

23

development tell a similar story, experiencing significant increases only in the last decade and sometimes remaining below international standards. This has led Caballero (1999) to classify Chiles stock market as illiquid. As for efficiency, Gallego and Loayza (2000) follow Demigurc-Kunt and Levine (1999) and use the rotation rate as a proxy. They find that efficiency has grown significantly in the 1990s, especially after Chilean companies stocks began to be traded in foreign exchanges in 1992. One of the most relevant actors in Chiles financial system, that played a significant role in its evolution, are pension fund managing institutions (AFPs). Created after the pioneer pension reform that replaced the existent pay-as-you-go system with a fully funded system based on individual capitalization, AFPs invest the economically active individuals savings in a variety of financial instruments, ranging from domestic public bonds to foreign stocks. The growing stock of resources channeled through the pension system has had a positive impact on financial intermediation and development. A similar, yet smaller, role has been played by insurance companies that also mobilize significant resources and have allowed for further market depth and the creation of new financial instruments. Finally, regarding structure, the time trend has implied a decrease in the relative importance of the banking sector, with growing space for stocks and other capital markets (such as bonds). This trend became stronger during the 1990s, and probably responds to the fact that, as mentioned, Chiles banking sector is relatively large by international standards, while the stock market has remained underdeveloped. 5.3 The effects of financial liberalization on saving, investment and growth As discussed in Section 4, it is hard to disentangle financial reform from financial development. Thus, here we present a quick survey of studies that individually assess the case of Chile regarding the impact of either of the two variables on saving, investment and growth. 5.3.1 Effects on saving Among the saving studies described in the previous section, Bennett et al (2001) find a negative, significant effect of one measure of financial development on the private saving rate. The included variable is financial depth, defined as M2 over GDP. The negative sign of the elasticity of saving to financial depth remains robust across different specifications. Moreover, the effect becomes more negative if private saving includes the purchase of durable goods, a result that is counterintuitive . The authors suggest that the increase in financial depth observed between 1977 and 1997 reduced private saving, as a ratio of disposable national income, by 0.12. The authors also analyze the effect of financial development (domestic private credit over GDP) on corporate saving. The estimated coefficient is insignificant, suggesting that higher credit availability does not reduce saving by firms.

24

Bandiera et al (1999) include Chile in the group of 8 countries that they use to test for the impact of financial liberalization on saving. Although the sign of the estimated coefficient varies across different specifications and estimation techniques, it always remains small and insignificant, so that the authors do not detect any significant effect of financ ial liberalization on Chiles private saving rates over the period 1970-1994. Morand (1998) discusses the positive effect of the pension reform, associated with a deepening of the financial system, on the saving rate. However, he empirically tests only the effect of pension reform, not of financial development. The author also finds a positive, significant effect of the level of M1 relative of GDP on the saving rate. He suggests that this provides evidence of the existence of liquidity constraints (that should weaken under higher financial liberalization). Morand (1998), Hachette (1998), and Vergara (2002) find a negative relationship between the level of foreign saving and domestic saving. This can be interpreted as evidence suggesting that the liberalization of international financial flows could lead to a reduction in domestic saving rates. 5.3.2 Effects on investment Gallego and Loayza (2000) explore this issue using a panel of 79 Chilean firms between 1985 and 1995. They find that financial development has been associated with shifts in investment decisions at the corporate level. During the 1980s corporate investment was not empirically related to Tobins q, being positively affected by cash flows and negatively by the level of debt. This suggests that firms were constrained, and financed their investments using internally raised funds. During the 1990s, however, investment became more sensitive to Tobins q and less dependent on cash flows and the level of debt. The authors also provide evidence that the evolution of the banking and stock markets has had a significant, positive effect on the growth rate of firms. Medina and Valds (1997) provide similar evidence, suggesting that investment decisions by Chilean firms are linked to cash flows. Thus, financial development that lifts liquidity constraints should allow investment decisions to be more strongly determined by productivity. 5.3.3 Effects on growth Gallego and Loayza (2002) find that financial developmentmeasured as private credit over GDP had a positive contribution to the increase in Chiles growth rate between 1960-1985 and 1986-1997, that ranged (depending on the specification of the estimated equation) between 0.15% and 0.72% of average GDP growth per year. Notice that this is only the direct effect of financial reform, as its indirect, combined effect through policy complementarities increased GDP growth by an additional 1.1%. The authors also simulate how a further increase in financial development (to the levels observed in the 10% most financially developed countries) would further boost growth by 0.15% over the next 10 years.

25

Although they do not directly evaluate the role of financial liberalization, Coeymans (1999) and Braun and Braun (1999) provide some evidence on the importance of foreign financing (related to the extent of liberalization in the capital account) on Chiles growth prospects. Bergoeing et al (2001), when comparing the different pace in growth recovery observed in Chile and Mexico after their recessions in the early 1980s, argue that the banking and bankruptcy reforms conducted in Chile since the 1970s played a significant role in explaining the countrys superior performance. Generally speaking, the authors state that differences in the degree of banking liberalization led to credit allocation being determined mostly by the market in Chile; while in Mexico it was determined by the government. The evolution of the financial system after the crisis also differed. While Mexican banks remained under government control, Chilean banks were reprivatized. Private credit in Chile subsequently recovered, while in Mexico it remained stagnant during most of the 1980s. The authors do not provide formal empirical evidence, but they do offer a theoretical framework tha t suggests that the financial reform allowed for more efficient resource allocation in Chile, explaining the differences in growth rates between both countries. Thus, an analysis of Chiles experience over the past three decades reinforces the conclusions drawn from the review of the cross-country evidence. In a context of overall structural reforms, financial liberalization was associated with the countrys golden age, a period of unprecedented growth, saving and investment. Moreover, Chiles GDP growt h expansion was not only the result of an increase in factor accumulation, but mainly the outcome of a significant increase in total factor productivity. Although the empirical evidence does not provide a clear-cut result in terms of the impact of financ ial development on saving, it does contribute some valuable insights on its effect on investment and growth. Financial liberalization has given financially-constrained firms access to financial resources required for investment, and has oriented investment decisions by profit-maximizing concerns, and not by liquidity constraints. Financial development has also been a relevant factor in explaining the boost in GDP growth, both through its direct effect and through its association with a consistent set of overall reforms. 6 - Conclusions There has been a worldwide trend towards financial liberalization in recent decades, and it is unlikely to be reversed in the foreseeable future. Its impact on saving rates has attracted a good deal of interest. After analy zing the various channels through which such an impact can take place, we conclude that the overall net effect is theoretically ambiguous. A nave look at the cross-country data suggests a positive and significant association between saving and financial development. However, after controlling for relevant saving determinants, careful scrutiny of the international evidence suggests that financial liberalization actually tends to reduce saving rates. However, increasing saving is not the ultimate aim of financial liberalization. Financial reform could lead to better financial intermediation and raise the level or

26

efficiency (or both) of investment, thus spurring growth. The international evidence shows that, after considering the influence of other relevant causal factors, financial development does lead to higher growth. The latter positive link occurs both through higher factor accumulation and an increase in total factor productivity, although the various studies do not provide a conclusive answer on the relative strength of these two effects. Chiles experience provides an ideal case study for analyzing the effects of financial reform on the economys overall macroeconomic and growth performance. In the past thirty years, Chile implemented a comprehensive set of structural reforms that contributed to an extended period of high saving, investment, and growth. The empirical evidence shows that financial reform played a significant role in the expansion of growth and investment, although its effects on saving were rather ambiguous. Hence Chiles case illustrates that financial liberalization, combined with adequate prudential regulation and strong supervision of banking and capital markets, can breed a sound and deep financial system able to boost growth over an extended period. It also suggests that larger benefits can be reaped when financial reform does not come as an isolated policy action, but is part of a consistent and comprehensive strategy of stabilization and structural reform.
References Abiad, A. and A. Mody (2002). Status Quo Bias in Financial Reform. Manuscript, IMF, July. Attanasio, O., L. Picci and A. Scorcu (2000). Saving, Growth and Investment. The Review of Economics and Statistics 82 (2): 182-211. Bailliu, J. and H. Reisen (1998). Do Funded Pensions Contribute to Higher Aggregate Savings? A CrossCountry Analysis. Review of World Economics 134 (4): 692 -711. Bandiera, O, G. Caprio, P. Honohan and F. Schiantarekki (2000). Does Financial Reform Raise or Reduce Savings? The Review of Economics and Statistics 82 (2): 239-263. Bayoumi, T. (1993). Financial Deregulation and Household Saving. Economic Journal 103: 1432-1443. Beck,T., R. Levine, and N. Loayza (1999): Finance and the Sources of Growth. Journal of Financial Economics 5 8: 261-300. Beck, T., A. Demirguc-Kunt and R. Levine (2000). A New Database on Financial Development and Structure. World Bank Economic Review 14(3): 597-605. Bekaert, G., Harvey, C.H. and C. Lundblad (2001). Does Financial Liberalization Spur Growth? Manuscript. Duke University. Benhabib, J. and M. Spiegel (2000). The Role of Financial Development in Growth and Investment. Journal of Economic Growth 5: 341 -360. Bennett, H., N. Loayza and K. Schmidt-Hebbel (2001). Un Estudio del Ahorro Agregado por Agentes Econmicos en Chile. In Anlisis Emprico del Ahorro en Chile, edited by F. Morand y R. Vergara . Banco Central de Chile. Bergoeing R., P. Kehoe, T. Kehoe and R. Soto (2001). A Decade Lost and Found: Mexico an Chile in the 1980s. NBER Working Paper Series 8520.

27

Beyer, H. and R. Vergara (2002). Productivity and Economic Growth: the Case of Chile. Central Bank of Chile Working Paper N 174. Braun, J. and M. Braun (1999). Crecimiento Potencial: El Caso de Chile. Cuadernos de Economa 36 (107): 479-517. Burnside, C. (1998). Private Saving in Mexico. Butelmann, A. and F. Gallego (2000). Ahorro de los hogares en Chile: evidencia microeconmica. Economa Chilena 3 (1): 5-24. Campbell, J. and N.G. Mankiw (1989). Consumption, Income and Interest Rates: Reinterpreting the Time Series Evidence. In NBER Macroeconomic Annual 1989, edited by O.J. Blanchard and S. Fischer, Cambridge, MA: MIT Press. Campbell, J. and N.G. Mankiw (1991). The Response of Consumption to Income: A Cross-Country Investigation. European Economic Revie w 35: 715-21. Carroll, C. and D. Weil (1994). Saving and Growth: A Reinterpretation. Carnegie Rochester Conference Series on Public Policy 40: 133-192. Cho, Y. (1988). The Effect of Financial Liberalization o n the Efficiency of Credit Allocation: Some Evidence for Korea. Journal of Development Economics 29: 101-110. Chumacero, R. and R. Fuentes (2002). On the determinants of the Chilean Economic Growth. Central Bank of Chile Working Paper 134. Corbo, V. and K. Schmidt-Hebbel (1991). Public Policies and Saving in Developing Countries. Journal of Development Economics 36: 89-115. Coeymans, J.E. (1999) Ciclos y Crecimiento Sostenible a Mediano Plazo en la Economa Chilena, Cuadernos de Economa 36 (107). Coronado, J. (1998). The Effects of Social Security Privatization on Household Saving: Evidence from the Chilean Experience. Finance and Economics Discussion Series 12. Division of Research & Statistics and Monetary Affairs. Federal Reserve Board. Dayal-Ghulati, A. and C. Thimann (1997). Saving in Southeast Asia and Latin America Compared: Searching for Policy Lessons. IMF Working Paper 110. Washington. Deaton, A. and C. Paxson (2000). Saving and Growth: Another Look at the Cohort Evidence. The Review of Economics and Statistics 82 (2): 212-225. de Melo, J. and J. Tybout (1986). The Effects of Financial Liberalization on Savings and Investment in Uruguay. Economic Development and Cultural Change 34: 561-588. Denizer, C. y H. Wolf. 2000. Aggregate Saving in the Transition: A Cross-Country Study. World Bank Economic Review 14 (3):445-456. Demirguc-Kunt, A. and Maksimovic, V. (1998). Law, Finance, and Firm Growth. Journal of Finance 53: 2107-2137. Edwards, S. (1996). Why Are Latin Americas Saving Rates so Low? An International Comparative Analysis. Journal of Development Economics 51 (1): 5-44.

28

Elbadawi, I. and K. Schmidt-Hebbel (1998). "Macroeconomic Policies, Instability and Growth in the World", Journal of African Economies , 7: 116-68, 1998. Fats, A. (2000). Do Business Cycles Cast Long Shadows? Short-Run Persistence and Economic Growth Journal of Economic Growth 5 (2): 147-162. Fry, M.J. (1995). Money, Interest, and Banking in Economic Development. Baltimore: Johns Hopkins. Galindo, A., F. Schiantarelli and A. Weiss (2002). Does Financial Liberalization Improve the Allocation of Investment? Micro Evidence from Developing Countries. Inter-American Development Bank Working Paper 467. Gallego, F. and N. Loayza (2000). Financial Structure in Chile: Macro-economic Developments and Microeconomic Effects. Central Bank of Chile Working Paper 75. Gallego, F. and N. Loayza (2002). The Golden Period for Growth in Chile- Explanations and Forecasts. Central Bank of Chile Working Paper 146. Hachette, D. (1998). Ahorro Privado en Chile. Cuadernos de Economa 35: 104-47. Haque, N., M. Pesaran and S. Sharna (2000). Neglected Heterogeneity and Dynamics in Cross-Country Savings Regressions. In Panel Data Econometrics - Future Direction: Papers in Honour of Professor Pietro Balestra. Oxford: Elsevier Science. Honohan, P. (1999). Financial Policies and Saving. In The Economics of Saving and Growth, edited by K. Schmidt-Hebbel and L. Servn. Cambridge University Press. Jadresic, E. and R. Zahler. (2000). Chile's Rapid Growth in the 1990s: Good Policies, Good Luck, or Political Change. IMF Working Papers 153, August. Jappelli, T. and M. Pagano (1994). Saving, Growth and Liquidity Constraints. Quarterly Journal of Economics109: 83-109. Jaramillo, F. F. Schiantarelli, and A. Weiss (1994). Capital Market Imperfections Before and After Financial Liberalization: An Euler Equation Approach to Panel data for Ecuadorian Firms. Journal of Development Economics 51: 367 -386. Koskela, E., H.A. Loikkanen and M. Virn (1992). House Prices, Household Savings and Financial Market Liberalization in Finland. European Economic Review 36: 549-558. Koskela, E. and M. Virn (1994). Taxation and Household Saving in Open Economies Evidence from the Nordic Countries. Scandinavian Journal of Economics 96: 425-441. King, M. (1986). Capital Market Imperfections and the Consumption Function. Scandinavian Journal of Economics 88: 59-90. King, R. and R. Levine (1993). Finance, Entrepeneurship, and Growth. Journal of Monetary Economics 32: 513-542. Lehmussaari, O.P. (1990). Deregulation and Consumption Saving Dynamics in the Nordic Countries. IMF Staff Papers 37: 71-93. Levine, R. (1997). Financial Development and Economic Growth: Views and Agenda. Journal of Economic Literature 35: 688-726.

29

Levine, R. (1998). The Legal Environment, Banks, and Long-Run Economic Growth. Journal of Money, Credit, and Banking 30: 596-620.Levine, R. and D. Renelt (1992). A Sensitivity Analysis of Cross-Country Growth Regressions. American Economic Review 82: 942-963. Levine, R., N. Loayza and T. Beck (2000). Financial Intermediary Development and Growth: Causes and Causality. Journal of Monetary Economics 46(1): 31-77. Levine, R. and S. Zervos (1993). What Have We Learned About Policy and Growth from Cross-Country Regressions? American Economic Review 83: 426-430. Levine, R. and S. Zervos (1998). Stock Markets, Banks, and Economic Growth. American Economic Review 88: 537-558. Loayza, N., K. Schmidt-Hebbel and L. Servn (2001). Una Revisin del Comportamiento y de los Determinantes del Ahorro en el Mundo. In Anlisis Emprico del Ahorro en Chile, edited by F. Morand y R. Vergara. Banco Central de Chile. Loayza, N., K. Schmidt-Hebbel and L. Servn (2000). What Drives Private Saving Across the World? The Review of Economics and Statistics 82 (2): 165-181. Loayza, N. and R. Shankar (2000). Private Saving in India. World Bank Economic Review 14 (3): 571-596. Masson, P., T. Bayoumi and H. Samiei (1995). Saving Behavior in Industrial and Developing Countries. Staff Studies for the World Economic Outlook. Washington, D. C. Medina, J.P and R. Valds (1997). Liquidez y Decisiones de Inversin en Chile: Evidencia de Sociedades Annimas. Central Bank of Chile Working Paper 25. Morand, F. (1998). Saving in Chile. What Went Right?. Journal of Development Economics 57: 201-228. Muellbauer, J. (1994). The Assessment: Consumer Expenditure. Oxford Review of Economic Policy 10: 139. Muellbauer, J. and A. Murphy (1993). Income Expectations, Wealth and Demography in the Aggregate UK Consumption Function. Paper presented to the HM Treasury Academic Panel. mimeo. Neusser, K. and M. Kugler (1998). Manufacturing Growth and Financial Development: Evidence from OECD Countries. Review of Economics and Statistics 80: 638-646. Ostry, J. D. and J. Levy (1995). Household Savings in France: Stochastic Income and Financial Deregulation. International Monetary Fund Staff Papers 4: 375-97. Ramey; G. and V. Ram ey (1995). Cross-Country Evidence on the Link between Volatility and Growth. American Economic Review 85(5): 1138-51. Rajan, R. and L. Zingales (1998). Financial Dependence and Growth. American Economic Review 88: 559586. Rodrik, D. (2000). Saving Transitions. World Bank Economic Review 14 (3): 481-508. Rojas, P., E. Lpez and S. Jimnez (1997). Determinantes del Crecimiento y Estimacin del Producto Potencial en Chile: El Rol del Comercio Internacional. Rousseau, P. L. and P. Wachtel (1998). Financial Intermediation and Economic Performance: Historical Evidence from Five Industrial Countries. Journal of Money, Credit, and Banking 30: 657-678.

30

Samwick, A. (2000). Is Pension Reform Conducive to Higher Saving? The Review of Economics and Statistics 82(2): 264-272. Schmidt-Hebbel, K. (2002). Chile's Growth: Resources, Reforms, Results. Presented at the Central Bank of Chile Conference The Challenges of Economic Growth. Schmidt-Hebbel, K. and L. Servn (1999). Introduction. In The Economics of Saving and Growth, edited by K. Schmidt-Hebbel and L. Servn. Cambridge University Press. Siregar, M. (1992). Financial Liberalization, Investment, and Debt Allocation. Manuscript, Boston University. Skinner, J. (1988). Risky Income, Life Cycle Consumption, and Precautionary Savings. Journal of Monetary Economics 22 (2): 237-255. Spiegel, M. (2001). Financial Development and Growth: Are the APEC Nations Unique? Pacific Basin Working Paper PB01-04. Vaidyanathan, J. (1993). Consumption, Liquidity Constraints and Economic Development. Journal of Macroeconomics 15: 591-610. Vergara, R. (2001). Determinantes del Ahorro Privado en Chile. In Anlisis Emprico del Ahorro en Chile, edited by F. Morand and R. Vergara. Central Bank of Chile. Wurgler, J. (2000): Financial Markets and the Allocation of Capital. Journal of Financial Economics 58: 187-214. Zeldes, S.P. (1989). Consumption and Liquidity Constraints: An Empirical Investigation. Journal of Political Economy 97: 305-46.