1
The Impact of Public Debt on Financial Development: a ‘National
Blessing’ or a ‘Public Curse’?
Master’s Thesis
MSc Economic Development & Globalization
Topic: International Monetary and Macroeconomics
Author: Thomas Bloemberg
Student Number: S2559366
Mail address:
t.r.bloemberg@student.rug.nl
2
Abstract
This paper uses a panel fixed effects model for a total sample of 96 countries, classified as low-, middle- or high-incomelow-, during the period 1988-2017. The results show that public debt and financial development do not have a non-linear relationship. Public debt does impact financial development for high-income countries and the sample as whole robustly, although the direction of the effect is still ambiguous. Public debt has a negative impact on financial development in low-income countries and thus seems to be a ‘public curse’. Concerning middle-income countries, the result is the same as for low-income countries, however, this result is not robust.
3 1. Introduction
Nowadays, it is widely accepted that financial development is a potential key driver of (long-run) economic growth, at least on a non-linear basis (Arcand, Berkes and Panizza, 2015; among others). Consequently, it is important to know what the drivers of financial development are to be able to explain the arising question why some countries have a higher level of financial development -and subsequent levels of economic development- than others. The literature has documented the relationship between financial development and inflation (Boyd, Levine and Smith, 2001), openness (Rajan and Zingales, 2003; Chinn and Ito, 2006) and institutional quality (La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1997; 1998) quite well. Another potential driver of financial development may be public debt. At the end of the 18th century people began to think about the mechanics of public debt. Hamilton (1781, April 30) wrote in a letter to U.S. Founding Father Robert Morris –who in 1781 was the Superintendent of Finance– that “a national debt if it is not excessive will be to us a national blessing: it will be powerfull cement of our union” [sic]. However, not everyone was convinced about the positive impact that public debt could have. The fourth President of the U.S. James Madison once wrote: “public debt is a public curse” (1790, April 13). So, what is the impact of public debt on financial development: a ‘national blessing’ or a ‘public curse’?
The relationship between financial development and public debt has frequently been a topic of discussion in policy circles, despite receiving relatively scant attention in the academic literature (Hauner, 2009). Therefore, this paper aims to examine the public debt-financial development link. Public debt could be a so-called ‘safe asset’. This ‘safe asset view’ is not a new concept, it emphasizes the positive role public debt can play in the development of the financial sector (De Soto, 2000; Reinhart and Sack, 2000; Kumhof and Tanner, 2005). Lower borrowing costs, longer maturities and better allocation of savings will be the result if there is a ‘safe asset’ (Hauner, 2009). On the other hand, a crowding-out effect is possible, credit to the private sector is crowded out by a rise in public debt (Ismihan and Ozkan, 2012; among others).1 Furthermore, some academics highlight a potential non-linearity and a threshold effect in the relationship between public debt and financial development. That is, up to a certain threshold point public debt can stimulate financial development, whereafter the effect becomes negative (Hauner, 2009). Up to this threshold, governments can provide a ‘safe asset’ which stimulates financial development. When the threshold in terms of public debt is reached, the positive effect turns into a negative effect: the ‘lazy bank view’ of Hauner (2009) replaces the ‘safe asset view’. Whereas most studies concerning the public debt-financial development link tend to focus on developing or middle-income countries (Kumhof and Tanner, 2005; Hauner, 2009; Îlgün, 2016), this study looks at low-, middle-, and high-income countries separately and as a whole. Using a fixed effect estimator with an unbalanced panel dataset for 96 countries from 1988 to 2017 the impact of public debt on financial development is empirically tested. The results show that public debt is related to financial development concerning high-income countries and the sample as a whole, although the direction is ambiguous. On top of that, public debt seems to be a ‘public curse’ for low-income countries. Middle-income countries show the same result, although that result is not robust.
The remainder of this paper continues as follows. Section 2 gives a presentation of the existing literature. Section 3 describes the data and methodology. Section 4 provides the results and Section 5 offers a discussion.
1 Credit to the private sector is the most frequently used measure for financial development. So, in other words, a rise in public debt could
4 2. Literature review
First, we take a closer look at financial development itself. Secondly, the relevance and importance of financial development is discussed on the basis of the finance-growth nexus. Thereafter, the drivers of financial development are explained and reviewed, with special focus on public debt. Finally, hypotheses are provided.
2.1 Financial development
The World Bank provides an explanation of key terms in the background section of its Global Financial Development Report 2015/2016. It stresses that financial (sector) development “occurs when financial instruments, markets, and intermediaries ease the effects of information, enforcement, and transaction costs and therefore do a correspondingly better job at providing the key functions of the financial sector in the economy”.2 Another definition of financial
development is provided in The Financial Development Report 2012, issued by the World Economic Forum (WEF). In this study financial development is defined as “the factors, policies, and institutions that lead to effective financial intermediation and markets, as well as deep and broad access to capital and financial services” (WEF, 2012, p. 13).
Financial development is measured in a variety of ways. Traditionally, financial development (i.e. financial depth) has been measured as the formal size of all financial intermediaries divided by total economic activity, i.e. GDP (Goldsmith, 1969; McKinnon 1973). King and Levine (1993) utilized such a measurement of ‘financial depth’, calling it LLY: liquid liabilities to GDP3. They noted, however, that some financial services are not necessarily closely related to
the absolute size of the financial system. Consequently, King and Levine (1993) constructed a second proxy: the ratio of commercial bank assets over commercial banks assets plus central bank assets, which they call BANK.4 This ratio indicates the relative importance of the central bank in the banking system. The rationale behind this measurement is that “banks are more likely to offer better risk management and investment information services than central banks” (King and Levine, 1993, p. 718). They believed that their constructed variable (BANK) would augment and complement the conclusions based upon the traditional measure of liquid liabilities to GDP (LLY). And it did: LLY and BANK are both significant (Table 7, in King and Levine, 1993). The coefficient of BANK was equal to, or even higher than that of LLY.
Several other studies use different measures for financial development. For example, stock market development (Levine and Zervos, 1998; Levine, 2001; Arestis, Demetriades, and Luintel, 2001) and banking system development (Arestis et al. 2001). Similar to King and Levine (1993), Ang and McKibbin (2007) argue that easily available monetary aggregates (like M2 and M3) with respect to GDP are not adequate indicators for measuring ‘financial depth’. This is due to the fact that these indicators “reflect the extent of transactions services provided by financial system rather than the ability of the financial system to channel funds from depositors to investment opportunities” (Ang and McKibbin, 2007, p. 220). A superior measure would be private credit to GDP, which was first used by King and Levine (1993) and has been used very often in prior research. Private credit to GDP “captures the amount of credit channeled from savers, through financial intermediaries, to private firms” (de Haan and Sturm, 2017, p. 12). Beck, Demirgüç-Kunt and Levine (2007) point out that private credit to GDP as
2 The World Bank (2016). Background information on key concepts in the Global Financial Development Report 2015-2016: Long-Term
Finance. Retrieved at: https://www.worldbank.org/en/publication/gfdr/gfdr-2016/background/financial-development.
3 According to King and Levine (1993), this measure equals two money aggregates. Namely, “M3” or sometimes “M2” when information of
“M3” is not available.
4 King and Levine (1993) also used the variable PRIVATE (claims on the nonfinancial private sector to domestic credit) and PRIVY (ratio of
5 a proxy for financial development has advantages over alternative measures, like broad money (M2), as a share of GDP. M2 as a share of GDP “does not measure a key function of financial intermediaries, which is the channeling of society’s savings to private sector projects” (Beck et
al., 2007, p. 31). Similarly, in contrast to measures like M1, M2 and M3, private credit to GDP
is more accurate in real size of the funds channeled to the private sector (De Gregorio and Guidotti, 1995).
Čihák, Demirgüc-Kunt, Feyen and Levine (2012) introduce an extensive dataset of the financial system, called the Global Financial Development Database (GFDD). It has recently been updated and now covers 109 indicators for 214 economies between 1960 and 2017. The World Bank also keeps track of financial sector statistics (FinStats), using 64 indicators. Sahay et al. (2015) create the ‘Financial Development Index Pyramid’ on basis of the matrix of financial system characteristics from Čihák et al. (2012): a number of indices are created to construct the terms depth, access and efficiency which indicate how well-developed a country’s financial institutions and financial markets are. Cumulating these results in the final overall index for financial development (Sahay et al., 2015; Svirydzenka, 2016)5. The resulting Financial Development Index Database (FDID) supplements the data of the FinStats and GFDD with additional data; it could prove to be useful whilst it covers 183 countries from 1980-2013.6 Moreover, this database is more user-friendly and better feasible for empirical work: instead of using 109 distinct indicators in the GFDD or the 64 indicators in FinStats, the FDID aggregates the109 indicators into 9 distinct indices. Therefore, within the FDID, the overall level of financial development is easier to ‘read’ and particular features of the financial system can be viewed more comprehensively (Svirydzenka, 2016). Similarly, as pointed out by Svirydzenka (2016), “the indices allow to pin down where deficiencies in financial development lie or which aspects of financial development affect macroeconomic performance, which could then be investigated in greater detail using the disaggregated data from FinStats or GFDD” (p. 6). Hence, for an empirical examination of financial development, the FDID is the superior database.
2.1.1 The finance-growth nexus
The finance-growth nexus is more than a century old and goes back to the end of the 18th century. Hamilton (1781) stated that to increase the level of economic activity “banks were the happiest engines that ever were invented” (p. 36). In his book The Theory of Economic Development, Joseph Schumpeter (1911) wrote about the finance-growth nexus. He argued that, with respect to fostering economic development (and technological change), financial intermediaries are crucial. On the other hand, Robinson (1952), did not agree with Schumpeter. He argued that if the level of real economic activity is subject to change, financial development merely follows. Goldsmith (1969) was one of the first to empirically establish the finance-growth nexus. Although he wrote that there was not a causal relationship between financial development and economic growth, he did find a positive correlation in comparing financial development and economic activity. As did McKinnon (1973) and Shaw (1973) a few years later. Despite several empirical studies indicating that financial development fosters economic growth, some of the embodied literature does not share this consensus. Adams (1819), the second President of the United States, was among the first to question it: banks are not all
5 Svirydzenka (2016) defines financial development as “a combination of depth (size and liquidity of markets), access (ability of individuals
and companies to access financial services), and efficiency (ability of institutions to provide financial services at low cost and with sustainable revenues, and the level of activity of capital markets)”(p. 5).
6 The FDID is available at the International Monetary Fund and is recently updated to 2017. It can be retrieved at:
6 growth enhancing because they damage the “morality, tranquility, and even health” of nations (p. 36). Lucas (1988) goes even further than Robinson (1952): “I believe that the importance of financial matters is very badly overstressed in popular and even much professional discussion and so am not inclined to be apologetic for going to the other extreme” (Lucas, 1988, p. 6). Later on, in their highly cited paper, King and Levine (1993) presented cross-country evidence that was consistent with the view of Schumpeter (1911): financial development is able to predict subsequent economic growth. Levine (2005) reviews the existing theory and literature on the finance-growth nexus and constitutes five categories on how financial development can contribute to growth. Financial systems: (i) produce information ex ante about possible investments and allocate capital; (ii) monitor investments and exert corporate governance after providing finance; (iii) facilitate the trading, diversification, and management of risk; (iv) mobilize and pool savings; (v) ease the exchange of goods and services (Levine, 2005). Robinson (1952) and Lucas (1988) do not agree with many researchers who stress that there is a positive relationship between finance and growth. In their view, financial development merely ‘follows’ or ‘overstresses’ economic growth. Nevertheless, some influential people have gone even further and noted that financial development has negative impacts on economic development – for instance Marx, Hilferding, Veblen and Lenin (Bezemer, 2019). There is a certain threshold after which financial development is no longer sustainable: a growing body of literature emphasizes that a high level of financial development could be drag on growth or even cause negative growth –i.e. economic contraction– (Shen and Lee, 2006; Cecchetti and Kharroubi, 2012; Bezemer et al. 2014; Arcand et al. 2015). This is especially true for the last couple of decades. Fig. 1 shows this threshold, in a cross section of 50 countries over 1971- 2011, where financial development is no longer good for growth. “The larger the number is on the vertical axis, the more economic growth each euro of credit is supporting” (Bezemer, 2019, p. 108). The solid line represents the point estimates of 5 year non-overlapping averages, the dashed lines represent the 95% confidence interval boundaries (Bezemer et al., 2016). From the beginning of the 1980’s every extra euro of credit supported less and less economic growth. If the trend continues, financial development could definitely be a drag on growth. Furthermore, Arcand et al. (2015) show that the level of institutional quality matters. If credit to the private sector is below 20% of GDP and the quality of institutions is high, financial depth has a significant positive effect on growth. However, when credit to the private sector rises above 70% (95%) the effect becomes (significantly) negative.
Fig. 1. The correlation of financial development Fig. 2. Development of credit composition.
and economic growth, 1971-2011 Different types of bank credit stocks over 1990-2011
7 Fig. 2 depicts the composition of credit over the years 1990-2011 for a balanced panel of 14 advanced countries. Non-financial businesses (credit to the real-sector) was stable during this period. However, the increase in mortgage / real estate credit is astounding: from 20% to 50% of GDP. Even during the crisis, it kept rising.
Financial development –or financialisation- became very rapid at the end of the 20th century, whilst most of it was due to growth in (financial) assets markets and especially the real estate sector (Bezemer et al., 2014). This excessive growth in assets markets is counterproductive and detracts from the real sector, e.g. in the form of rents. In economic sense, rents do no refer to the sum received from renting (out) a house. Rents are the revenue received above the cost of maintaining a non-produced input, like land or patents for example. In other words, what profit is to wages and production, are rents to ownership (Bezemer, 2019). Furthermore, financial development can cause higher crisis frequency, which in turn has a negative impact on growth. For instance, this ‘dark side’ of financial development (Bezemer, 2019) was a key contributor to the subprime-mortgage crisis that took off in 2007, which led to a global recession.
Thus, it is important to note a distinction: credit that goes to the financial sector (asset markets; rents) or the real sector (production and services), where the former could signal towards a potential crisis. For example, reasons why the system is prone to financial crises is explained by Hyman Minsky’s ‘financial instability hypothesis’ (Minsky, 1992). Credit to the real sector usually fosters economic growth, but note that it can cause a crisis as well.7
As can be noted from above, a large body of literature which explores the finance-growth nexus exists. Although apparently financial development is a key element for a country with regard to economic development, whether this relationship is positive or negative –or even there at all– is beyond the scope of this empirical research. However, it is now safe to say that financial development is a very important concept. Hence, it is necessary to take a closer and more specific look at what drives financial development. As Levine (1997) emphasizes: “I believe that we will not have a sufficient understanding of long-run economic growth until we understand the evolution and functioning of financial systems” (p. 719). Thus, to understand long-run economic growth, it is paramount that we first understand how financial development actually works. So, what factors spur financial development?
2.2 Drivers of financial development 2.2.1 Inflation
Inflation is one of the important determinants of financial development. Boyd et al. (2001) are among the first to show empirically that inflation and financial development are negatively related.8 Moreover, there is support that the relationship is highly non-linear and that it might be driven by threshold rates of inflation: “as inflation rises, financial sector performance falls, but the marginal impact of additional inflation on the financial sector also diminishes rapidly” (Boyd et al., 2001, p. 225). In a similar vein, Khan, Senhadji and Smith (2006) emphasize that there is a threshold effect in the relationship between inflation and financial development. Using the model of Boyd et al. (2001), they show that the relationship is positive until the threshold point, whereafter the effect becomes negative. The threshold point is usually in the range of 3
7 Specifying what causes (financial) crises is beyond the scope of this research. For further reading see, among others, Mishkin (1992),
Minsky (1992), Acharya and Richardson (2009), and King (2013).
8 Boyd et al. (2001) use a banking data set for concerning the period 1960-1995 focusing on 65 countries and a stock market dataset from
8 to 6%, depending on which measure of financial depth is used (Khan et al., 2006). However, they only looked at industrialized and developing countries, leaving low-income countries aside.
Although there is a broad spectrum of literature that highlights the potential negative effect of inflation –at least after a threshold- on financial development, there is a way to reduce this effect. Ayadi, Arbak, Ben-Naceur and De Groen (2015) look at Mediterranean countries for the years 1985 to 2005 and highlight that the negative effect of inflation on financial development could be partly offset if an economy is financially open. When a country is financially open, it is more prone to financial products that are linked to currency. As Ayadi et al. (2015) state concerning the effect of inflation on financial development: “the availability of currency-linked financial products could sweep and possibly reverse some of these negative effects of inflation”. So, in their reasoning, inflation can still hurt financial development although this is less the case when a country is more financially open.9
Overall, most studies highlight the negative side of inflation with regard to financial development. So, with respect to financial development, inflation needs to be controlled for. Some authors emphasize this: price stability is an important concept for successful financial development (Boyd et al., 2001; Bittencourt, 2011; Abbey, 2012). Former Vice President of the European Central Bank (ECB) Lucas Papademos (2006) provided a definition of price stability in a speech: “a state in which the general price level is literally stable or the inflation rate is sufficiently low and stable, so that considerations concerning the nominal dimension of transactions cease to be a pertinent factor for economic decisions”.
2.2.2 Openness
Openness can be viewed as opening up the market for goods and service (trade openness) or opening up the capital account for capital to flow more freely across borders (financial openness). Two reasons emerge from the existing literature why openness is important for financial development, the ‘supply side’ (Rajan and Zingales, 2003) and the ‘demand side’ (Svaleryd and Vachos, 2002) reason. Starting with the latter, an economy that opens up can develop financially because it decreases the incentive for incumbent ‘interest groups’ to extract monopolistic rents (Rajan and Zingales, 2003). These ‘interest groups’ are usually established incumbent industrial and financial players who usually stand to lose from financial development. They are opposed to financial development because new firms can enter with more ease, which in turn increases competition and lowers possible rent extraction. When an economy opens up, the power of these ‘interest groups’ is weakened. However, there could be higher rent extraction by banks when opening up. Because of increased competition by opening up, small local banks could merge to remain competitive (Ashraf, 2018). In turn, the following market concentration could reinforce the monopolistic fashion of rent extraction and result in lower banking efficiency (Agénor, 2003; Yoo, 2016).
Second is the ‘demand side’: risk diversification (Svaleryd and Vachos, 2002). Risks may increase when a country opens up because it will be exposed to exogenous shocks. The increased exposure could occur via various channels.10 “To the extent that openness is
associated with greater risks, such as increased exposure to external demand shocks or foreign
9 It should be noted that Ayadi et al. (2015) focus on banking development as a proxy for financial development when looking at the impact
of inflation. However, Kumhof and Tanner (2005) state that focusing on the banking sector is not a good idea, countries who did focus on the banking sector lagged behind in their financial development.
10 Winters (2002) provides a framework for the transmission of trade shock when trade is liberalized/open. Four channels emerge: the factor
9 competition, it will create new demands for external finance” (Huang and Temple, 2005: p.1). So, to be able to deal with these exogenous shocks, finance is demanded. Svaleryd and Vlachos (2002) stress that liberalizing trade could lead to financial market development, which in turn constitutes the ability of agents to diversify the ‘added’ risks of openness. Similarly, Ashraf (2018) utilizes a panel dataset of 287 important banks in 37 emerging countries over the time period 2000-2012 to show that international trade gives banks opportunities for risk diversification.
Rajan and Zingales (2003) propose that trade openness without financial openness is not sufficient to bring forth financial development. Likewise, financial development will not come forth when only financial openness is at play. It is key, therefore, to have trade and financial openness at the same time. Their results show that when cross-border capital flows are high, there is a positive correlation between financial development and trade openness. So, Rajan and Zingales (2003) hypothesize that for financial development to take place, both the trade and capital accounts should be opened up simultaneously. Several other studies (Law 2007; Ahmed 2013) report evidence to support the so-called Rajan and Zingales (2003) hypothesis by means of showing that the interaction variable between financial openness and trade openness is positive. Thereby, Law (2007) looks at different income groups and states that “openness in terms of trade and capital are most potent in promoting financial development in middle-income countries, whereas its influence is relatively small in low-income countries and high-income countries” (p.145). However, there are some objections to the Rajan and Zingales (2003) hypothesis. Baltagi, Demetriades and Law (2009) use dynamic panel estimation techniques to evaluate the hypothesis, using banking sector development as a proxy for financial development. They find only partial support: a closed economy could benefit the most if they open both their trade and capital accounts at once, but with only one openness dimension there could still be banking sector development. Thereby, they also provide evidence for different country income groups. Low-income countries benefit more than developing (middle-income) countries, because the latter is already very open, and the marginal effect of more openness will be limited (Baltagi et al., 2009).
Most of the studies in the existing literature, however, focus on either of the two dimensions of openness. The outcomes are not all in line with each other. Svaleryd and Vlachos (2002) are among the first to empirically examine the relationship between trade openness and financial development. They find positive and significant evidence using a panel regression for 88 countries to support their hypothesis: a positive mutual dependence between financial development and trade liberalization does exist. Thereby, they check for the legal origin (as described in La Porta et al., 1997) of a country by incorporating a dummy variable. It did not affect their results. Several others have reported a positive relation since. Like Huang and Temple (2005), who show according to their estimated model that there are strong effects of openness on financial development. Their evidence persists in the long run, for the sample as a whole and low-income countries. However, the evidence does not hold for high-income countries. Also, Hanh (2010) obtains evidence of bidirectional causality between financial openness/development and trade openness for 29 Asian developing countries over 1994-2008. Two crisis dummies are incorporated as well: the crisis of 1997 affected all dependent variables negatively, but the 2007 crisis did not affect the domestic financial system11.
11 That does not mean that the crisis of 2007 could never affect a developing country’s level of financial development: “due to strong linkage
10 On the other hand, some studies report a negative effect, or even no particular effect at all, like, Menyah, Nazlioglu and Wolde-Rufael (2014) who question in what way causality runs. They look at 21 countries in Sub Saharan Africa (SSA) and report results of their panel causality approach between trade openness and financial development. Instead of finding in what way causality runs, they find that for 17 of the 21 countries there was no causality at all. Consequently, they conclude that trade openness and financial development do not have predictive power on each other and have a limited causal relationship. Gries, Kraft and Meierrieks (2009) find a similar result in their sample of SSA countries.12 Although there are indeed some positive and significant interactions between trade openness and financial development, these interactions are often not stable in the long-run or are found to be unrelated to each other. The findings of Gries et al. (2009) and Menya et al. (2014) are somewhat contrasting to the findings of Svaleryd and Vlachos (2002) and Hanh (2010), where causality runs in both directions.
Finally, Kim, Lin and Suen (2010a) find a non-linearity: trade openness has negative short-run and positive long-run effects on financial development. Kim, Lin and Suen (2010b) perform a pooled mean group (PMG) estimator approach with 87 countries over the period 1960-2005. As Kim et al. (2010a), they also show a non-linearity and state that trade openness and financial development are substitutes in the short-run and complements in the long-run. Thereby, developing countries are affected more by financial development than industrialized countries are (Kim et al., 2010b).
Several other studies examined the relationship of financial openness and financial development, with different measurements. The most commonly used measurement of financial openness is capital account liberalization (i.e. financial liberalization). Capital account
liberalization can stimulate financial development (Levine and Zervos, 1998; Levine 2001;
Ang and McKibbin, 2007)13. Chinn and Ito (2002) perform an analysis on a large sample of countries over the 1977-1997 period examining whether financial development is compatible with capital controls. Their results show a strong and positive relationship. However, Chinn and Ito (2002) stress that legal systems and institutions are of great importance: “while financial liberalization alone may have a zero, or even negative, impact on development of the financial system, when combined with a well-developed legal system or institutions, it may well serve to stimulate financial development” (p. 20).14 Several other studies provide results in the same
vein; institutional quality is important for capital account liberalization in stimulating financial (depth) development (Chinn and Ito, 2006; Klein and Olivei, 2008; Calderon and Kubota, 2009; Ahmed, 2013; Trabelsi and Cherif, 2017). Also, Chinn and Ito (2006) find evidence in line with the hypothesis as reported by Tornell, Westermann and Martinez (2004): trade liberalization is a precondition for financial liberalization. Thereby, Chinn and Ito (2006) constructed an index that captures a country’s de jure degree of capital account openness: also known as the
12 SSA countries are representative when looking at low-income and middle-income countries. As 24 out of the 48 countries in Sub Saharan
Africa are classified as low-income and 23 as middle-income. Concerning high-income countries, SSA is not representative as only one country is classified as high-income (Seychelles). Source: World Bank Country and Lending Groups (2019):
https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups
13 Levine and Zervos (1998) use capital control liberalization as a proxy for capital account liberalization. Levine (2001) uses the
liberalization of restrictions on international portfolio flows to measure capital account liberalization/financial openness. Ang and McKibbin (2007) use the inverse of an index for financial repression to be interpreted as the extent of financial liberalization. They construct a summary measure for financial sector development by taking a logarithm of three variables described in sub-section 2.1 (M3 over GDP, private credit over GDP, and commercial bank assets to commercial bank assets plus central bank assets).
14 Chinn and Ito (2002) show this empirically. They create an interaction variable between legal/institutional development and financial
11 Ito index. The higher the value of the index KAOPEN, the more open the country in consideration is to cross-border capital flows and transactions (Chinn and Ito, 2008). Furthermore, a few studies suggest the existence of threshold effects with respect to capital account liberalization (Klein, 2005; Chinn and Ito, 2006; Prasad, Rajan and Zingales, 2007; Eichengreen, Gullapalli and Panizza, 2011). The threshold effect means that “countries must reach a certain threshold in terms of economic and institutional development before they can expect the benefit from capital account liberalization” (Eichengreen et al., 2011, p. 1105). Gopalan (2016) finds a threshold effect as well while using foreign bank presence as a proxy for financial openness. Similarly, Gopalan (2018) looks at 57 emerging markets and developing economies (EDMEs) between 1995 and 2009 while examining the relationship between foreign bank entry and financial sector depth (proxy for financial development). The results suggest that the presence of foreign banks stimulate financial deepening. However, the higher the level of economic development, the more the marginal effects of foreign banks diminish over time. “In other words, the impact of foreign bank entry tends to diminish as the per capita income of the country rises” (Gopalan, 2018, p. 954). On top of that, Van Horen (2013) finds that, on average, there is a negative relationship between foreign bank presence and private credit. However, when the sample is split the negative relationship is only apparent for developing countries, while it is not the case for emerging markets and advanced economies15. Consequently, an important caveat resulting from Van Horen (2013) and Gopalan (2018) is that the level of economic development should be taken into account.
Despite the measurements for financial openness mentioned above, there are some other measurements and databases constructed in existing literature. The ‘capital variable’ by Quinn (1997) for example, which is a continuous measure of capital account policy. Concerning databases, the Chinn-Ito index is used most frequently and is “widely considered to be the best measure” (Satyanath and Berger, 2007, p. 309), although Lane and Milesi-Ferretti (2007) and Abiad, Detragiache and Tressel (2010) constructed financial liberalization databases as well. The Lane and Milesi-Ferretti (2007) proxy for financial openness is based upon both the Chinn-Ito index and estimates of the de facto ratio of external assets and liabilities to GDP. The estimates focus on 145 countries while covering the period 1970-2004. Abiad et al. (2010) create an index for financial reforms covering seven aspects of financial sector policy to proxy for financial liberalization. The index “summarizes de jure changes in credit controls, interest rate controls, entry barriers for banks, regulations, privatization, and restrictions on international financial transactions” (de Haan and Sturm, 2017, p. 171). The database covers 91 countries over the time period 1973-2005. According to de Haan and Sturm (2017), the index of Abiad et al. (2010) is most frequently used for measuring financial liberalization. Although the Abiad et al. (2010) financial liberalization index was constructed several years after the index of Chinn and Ito (2006), the Chinn-Ito index is the more up-to-date version and includes the most countries.16 The most recent constructed database is the one of Fernández, Klein,
Rebucci, Schindler and Uribe (2016), which is an updated version of the dataset presented in Schindler (2009). The Fernández et al. (2016) database encompasses a lot more dimensions than previous databases, focusing on capital controls and provides an overall capital restriction index which is a potential useful proxy for financial openness.
15 This is in line with the findings of Detragiache, Tressel and Gupta (2008) and Claessens and Van Horen (2014) who both find a negative
relationship between foreign bank presence and private credit for low-income countries. Detragiache et al. (2008) look at 89 low-income and lower-middle income countries and Claessen and Van Horen (2014) look at 137 countries between 1995 and 2009. Thereby, Claessen and Van Horen (2014) also documented a database for foreign bank presence.
16 The Chinn-Ito index encompassed the time period 1970-2000 but has recently been updated to 2017 for 182 countries and can be retrieved
12
2.2.3 Institutional quality
The effect of institutional quality on financial development is researched extensively, with diverging outcomes. Institutional quality is import: “contract enforcement, property rights, investor protection, and the like, matter because they allow agents to overcome frictions that arise when two parties with competing interests enter into a production relationship” (Levchenko, 2007, p. 791). As briefly highlighted in previous sub-sections, institutional quality also matters for the relationship between finance and growth (Arcand et al., 2015) and between financial development and financial openness (Chinn and Ito 2006). A closer look at what institutional quality means directly for financial development is provided below.
The starting point is initiated by La Porta et al. (1997; 1998) and is called the law and finance
literature. It prescribes that the legal environment –both legal rules and their enforcement– is
heterogeneous across countries, and that this heterogeneity is important for the development of financial markets. Thereby, “the law and finance theory predicts that historically determined differences in legal tradition help explain international differences in financial development today” (Beck, Demirgüç -Kunt and Levine, 2003, p. 138). These differences in legal tradition can be divided into two influential forms: British Common Law and French Civil Law. In other words, the law and finance literature points out that countries with strong legal environment have well-developed financial markets (Rajan and Zingales, 1998; Demirgüç-Kunt and Maksimovic, 1998; Beck et al., 2003). Because, countries with a strong legal framework are better in financing firms and increase the lenders’ ability for loan collateralization (Claessens and Laeven, 2003). Another reason why countries with a strong legal environment have well-developed financial markets is that “a good legal environment protects the potential financiers against expropriation by entrepreneurs, it raises their willingness to surrender funds in exchange for securities, and hence expands the scope of capital markets” (La Porta et al., 1997, p. 1149). Thereby, countries have bigger financial markets when they are associated with British Common Law and a more prominent amount of shareholders’ rights (La Porta et al., 1998; Beck et al., 2003). In a similar vein, Claessens and Laeven (2003) and Mishkin (2009) stress that strong property rights and a good legal system –one that enforces contracts quickly and fairly- can stimulate financial development.17 Thereby, Mishkin (2009) lists that corruption, quality of financial information, corporate governance and bank regulation and supervision are important with respect to institutions (and its link with financial development). For example, if corruption is limited, property rights and the legal system are strengthened and, concerning advanced countries especially, corporate governance is a continuing struggle to remain a balance between management and stockholders (Mishkin, 2009).
Besides legal origin and property rights, several other ways to measure institutional quality exist. The International Country Risk Guide (ICRG) is frequently used, or one of its 22 components, to construct an index for institutional quality (Hanh, 2010; Law and Azman-Saini, 2012; Ayadi et al., 2015).18 Another example is the Institutional Quality Database (IQB) created by Kunčič (2014). Three homogenous groups of formal institutions -legal, political and economic- are created by clustering more than thirty indicators of institutional quality. Together they shape a complete formal institutional environment of a country (Kunčič, 2014). However, the database contains data up to 2010 and has not been updated since. Similarly, the Worldwide Governance Indicators (WGI) reports six dimensions of governance: voice and accountability;
17 Mishkin (2009) examines the role of globalization in promoting financial development. He reports that opening trade in goods (trade
openness) and opening financial markets (financial openness) ‘encourages’ financial development. He stresses that global markets play a key role and that globalization “should be one of the highest priorities for developing countries” (p. 168) when they want to improve their standards of living.
13 political stability and absence of violence; government effectiveness; regulatory quality; rule of law; control of corruption (Kaufmann, Kraay, and Mastruzzi, 2010). The WGI has been updated a several times and now contains data from 1996 to 2018 and covers over 200 countries. Governance is used several times as a proxy for institutional quality in existing literature (e.g. Law and Azman-Saini, 2012). Hence, the WGI dataset could be useful for this study as it covers multiple dimensions. Thereby, the studies that use governance as an indicator for institutional quality suggest that better governance could stimulate financial development. Another measure for governance could be the frequently used Freedom House database, which consists of a rating of political rights and civil liberties (PR and CL) to measure political freedom across countries.19 When the combined average rating of PR and CL for a country falls between 1.0 and 2.5 they are considered free, between 3.0 and 5.5 partly free, and between 5.5 and 7.0 not free. The Freedom House indexes are updated every year and consists of a considerable timespan (1972-2017) and includes 195 countries. For example, Huang (2005) uses the Freedom House indexes as a potential determinant of financial development. All in all, there are quite a few databases available to proxy for institutional quality.
2.3 Public debt
Relatively little research is conducted on the topic of public debt levels and financial development, with varying conclusions. It depends which financial development indicators (sub-section 2.1) and which debt levels are used, and which countries. For instance, total public debt can be split in domestic debt (lenders in the home country) and external debt (foreign lenders). Unfortunately, data on domestic and external debt is hard to obtain. A well-suited database could be the one of Panizza (2008), would it not have been publically restricted. Another convenient distinction is the one between short-term (< 1 year) and long-term (> 10 years) public debt. Moreover, the term public debt can be somewhat confusing as multiple (close) substitutes are frequently used in existing literature. Sovereign debt, (central) government debt and national debt are all terms that are used interchangeably with public debt. Although not all authors attach the same meaning to these terms, they are often used to indicate the same matter. Institutions like the World Bank, OECD and the IMF all use the term ‘central government debt’. In this research, I use the generic term public debt and follow the central government debt definition of the World Bank: “debt is the entire stock of direct government fixed-term contractual obligations to others outstanding on a particular date. It includes domestic and foreign liabilities such as currency and money deposits, securities other than shares and loans. It is the gross amount of government liabilities reduced by the amount of equity and financial derivatives held by the government”.20
In the 21th century public debt has grown to immense heights in most of the high-income countries. According to Woo and Kumar (2015) public debt levels reached their highest levels in 50 years: 107% of GDP in advanced economies by the end of 2013. “This raised serious concerns about fiscal sustainability and its economic and financial market impact amid the current European sovereign debt crisis” (Woo and Kumar, 2015, p. 705). We are no longer amid the European sovereign debt crisis nor a global recession, but public debt levels are nowhere near the pre-crisis levels. Fig. 3 indicates that the public debt levels have stagnated just above 100% of GDP for advanced economies. On the other hand, it did not stagnate for low- and
19 Source: www.freedomhouse.org. Since 1972, Freedom House publishes their Freedom in the World report and updates the dataset on
annual basis. The reports per year can be found and read on their website. Thereby, according to the World Bank the Freedom House indexes are used often in published studies, have a high coverage across countries, a high coverage across time, and are based on opinions of experts. Source: http://siteresources.worldbank.org/INTLAWJUSTINST/Resources/IndicatorsGovernanceandInstitutionalQuality.pdf
14 middle-income countries (emerging economies in fig. 3), but instead grew. It is now at the same level as the beginning of the century and it keeps on growing. In fact, it is predicted that their public debt will be almost twice as high in 2024 as pre-crisis levels.
Fig. 3. Public debt (% of GDP) in Advanced and Emerging economies
Note: public debt levels as a percentage of GDP are depicted for different economies. The figure includes predictions by the IMF. Consequently, the time period is 2001-2024. The orange line represents advanced economies and uses the left-hand axis. The grey line depicts emerging market and developing economies (abbreviated to emerging economies) and uses the right-hand axis. Source: International Monetary Fund (IMF). 2019. Data for the figure retrieved at:
https://www.imf.org/external/datamapper/GGXWDG_NGDP@WEO/OEMDC/ADVEC.
We inherited these high public debt levels from the global recession. A popular rationale is to reduce the debt-burden because it is a drag on economic growth. “While causality runs both ways, an important causal channel is taxation: high public debt implies the need to distort economic activity (labor, capital) to service the debt (either through taxation or cuts in productive spending), which dampens economic growth” (Ostry, Ghosh, and Espinoza, 2015, p. 2). The high debt burden has its effects on economic growth. An important distinction could be that of the short- and long-run effects. According to Elmendorf and Mankiw (1999), public debt could positively affect growth in the short run through the increase of total demand and output in an economy; but in the long run, the effect becomes negative whilst output and capital are crowded out. Does the high debt burden –or public debt in general– also have its effects on financial development? The next subsection explores the public debt-financial development link.
2.3.1 Public debt and financial development
Kumhof and Tanner (2005) investigate the relationship between government debt and financial intermediation for developing countries in terms of their exposure. Where exposure is defined as the average ratio of financial institutions’ net credit to the government to their total assets. For industrialized countries exposure is around 10%, however, for developing countries 20-40% and sometimes even 50% for large developing countries (Kumhof and Tanner, 2005). Banks in developing countries usually hold a large portion of their assets in government debt (i.e. are highly exposed). “In such an economy, a government that contemplates a debt devaluation is faced with the insolvency of its banking system, because the typical bank capitalization ratio is less than 10 percent of assets” (Kumhof and Tanner, 2005, p. 7).
20 30 40 50 60 70 60 70 80 90 100 110 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
15 Following the numbers above, they conclude that for developing countries, it is crucial to keep government debt a stable investment -no debt devaluation- for banks because they are heavily exposed to it. At the same time, developing countries are exposed to increased risk in private lending due to their relatively weak legal and institutional frameworks. Furthermore, financial intermediation should be focused on securities markets and not on banks to make steps forward. Countries which did not follow this road lagged behind in their financial development (Kumhof and Tanner, 2005).
Hauner (2009) makes a distinction between the ‘safe asset’ view and the ‘lazy banks’ view, whereby the former is positively associated with financial development and the latter negatively.21 The ‘safe asset’ view is used in prior research and is most prominent in developing financial sectors because of the positive role public debt can play while providing a relatively safe asset. Several reasons are provided in previous literature. First, it can tackle information asymmetries (Kumhof and Tanner, 2005). Second, it can help overcome legal and institutional imperfections that rule out the possibility of using real estate or movable property for collateralization (De Soto, 2000). Thereby, to enhance derivative markets and payment and settlement systems, there should be a sufficient amount of liquid collateral (Hauner, 2009). Third, the private bonds market tracks the government (public) bonds market. Essentially, the government bonds market produces a benchmark yield curve (Reinhart and Sack, 2000). Fourth, the willingness of depositors to have their funds intermediated in a generally risky environment is enhanced by financial institutions that have government debt on its balance sheet, because it acts as a form of security to the depositors (Kumhof and Tanner, 2005). On the other hand, Hauner (2009) proposes an alternative view where public debt is negatively related to financial development. In this ‘lazy bank’ view the banking system entails in large public sector borrowing.22 Banks in developing countries that lend mainly to the public sector are usually more profitable, but less efficient, which reduces the incentive of banks to provided credit to the private sector. This, in turn, is detrimental to financial deepening. Moreover, lower bank efficiency also harms financial development via a deadweight loss created by financial intermediation (Fry, 1995).
The analysis of (Hauner, 2009), concerning 73 middle-income countries based on country- and bank-level data in a fixed-effects model, shows that the ‘lazy banks’ view is the overall more favorable one, indicating that public debt is negatively related to financial development, at least on a non-linear basis.23 Îlgün (2016) provides evidence that government borrowing from domestic banks has a negative impact on financial development, which is in line of the ‘lazy banks view’ of Hauner (2009). His results are based on the Common Correlated Effects Mean Group estimator (CCEMG) of Pesaran (2006) with respect to 18 developing countries between 1987 and 2013. Similarly, Ismihan and Ozkan (2012) show a negative relationship using monetary and fiscal authorities and the banking sector. They are the first to construct a theoretical framework in the public debt-financial development link and try to find support for the conclusions of Hauner (2009). Ismihan and Ozkan (2012) conclude that “in countries where credit to government makes up a major share of total bank lending, public debt is likely to harm financial development, with unfavourable implications for economic activity” (p. 351). The finding of Ismihan and Ozkan (2012) is in line with the ‘lazy bank’ view. A crowding-out effect
21 It is somewhat in line with the reasoning of Kumhof and Tanner (2005): focus should be on securities markets (‘safe asset’ view) and not
on banks (‘lazy banks’ view). If the focus is on banks, financial development will lag behind (‘lazy bank’ view is negatively associated with financial development).
22 “Note that “lazy” does not imply a value judgement here, as it reflects rational behavior on the part of the banks” (Hauner, 2009, p. 171). 23 Institutional quality also matters according to Hauner (2009). The effect of public debt on financial development is affected by the
16 takes place where, in the banking sector’s loan portfolio, private borrowing is replaced by public borrowing (Ismihan and Ozkan, 2012).24 Several other studies also report a crowding out effect of public debt (Caballero and Krishnamurthy, 2004; Emran and Farazi, 2009; Bua, Pradelli, Presbitero, 2014; Ayadi et al., 2015). However, the potential non-linearity should be considered: the positive effect of public debt on financial development (‘safe asset’) could hold until a certain threshold point, whereafter the effect becomes negative (‘lazy banks’) (Hauner, 2009). Because of the potential non-linearity, the ‘safe asset’ view and the ‘lazy banks’ view are not necessarily mutually exclusive.
2.5 Hypotheses
The sample as a whole (96 countries) is tested to see how public debt affects financial development. Several studies explored the public debt-financial development link and found that public debt has an effect on financial development (Hauner, 2009; Ismihan and Ozkan, 2012; Îlgün, 2016). Therefore, I expect to find the same result regarding hypothesis 1a: support for the null hypothesis. I also expect the relationship to be negative, mainly due to the fact that several studies report crowding-out effects (Ayadi et al., 2015; among others). Thereby, as can be read below, I expect the relationship for low- and middle income countries to be negative and, positive for high-income countries. However, the positive effect is potentially not sufficient to offset the negative effect expected from low- and middle-income countries. This results in hypothesis 1b. I expect that null hypothesis will be empirically supported, i.e. that public debt has a negative relationship with financial development.
Hypothesis 1a
H0: Public debt has a causal relationship with financial development.
Ha: Public debt does not have a causal relationship with financial development.
Hypothesis 1b
H0: Public debt has a negative relationship with financial development.
Ha: Public debt has positive relationship with financial development.
Furthermore, as described by Hauner (2009), the relationship between financial development and public debt could be non-linear. A threshold effect could be found as well. However, following the threshold definition of Hauner (2009) the effect of public debt on financial development should be ‘harmful’ after the threshold point. I do not test for a threshold point (i.e. a point where the (possible) positive effect becomes negative), instead I will look if there is a ‘regular’ non-linear relationship between public debt and financial development. This results in hypothesis 2 beneath. Thereby, it is important to know if the relationship is non-linear as it may require different estimation techniques.
Hypothesis 2
H0: The relationship between public debt and financial development is non-linear.
Ha: The relationship between public debt and financial development is not non-linear.
24 An increase in public sector borrowing is a consequence of fiscal expansion and occurred frequently after the crisis. Ismihan and Ozkan
17 Hypothesis 1b is also individually tested for the three income-level country groups, see hypothesis 3a, 3b and 3c below. Most studies report a negative relationship -or at least more prominent than a positive relationship- between public debt and financial development while looking at developing countries (Hauner, 2009; Ismihan and Ozkan, 2012; Îlgün, 2016). Hence, for low- and middle-income countries I expect to find evidence supporting the null hypothesis. However, high-income countries are probably better able to provide a ‘safe asset’ and rely more on securities markets. Thereby, they tend to have lower rates of inflation, better quality institutions in place and are usually more open to trade and finance than middle- and low-income countries. Which are all indicators that could stimulate financial development. However, Hauner (2009) noted that the degree of financial openness did not affect his results in the country-level regression, only in the bank-level regression it did. This study focuses on the country-level and the results could therefore be dampened without the stimulating effect of financial openness. On the other hand, Rajan and Zingales (2003) exert that there should be trade and financial openness simultaneously. If trade and capital accounts are indeed open simultaneously, both types of openness together could stimulate the relationship between public debt and financial development. Thereby, Chinn and Ito (2002; 2006) stress that there should be quality institutions in place for financial openness to have an effect. All in all, I also expect to find evidence supporting the null hypothesis for high-income countries. Compared to low- and middle-income countries, high-income countries usually have greater institutional quality, a higher degree of openness, a more stable debt market, and their financial system is usually more capital-based than bank-based. These all have the ability to strengthen the relationship between public debt and financial development.
Hypothesis 3a
H0: Public debt relates negatively to financial development in low-income countries.
Ha: Public debt does not relate negatively with financial development in low-income countries.
Hypothesis 3b
H0: Public relates negatively with financial development in middle-income countries.
Ha: Public debt does not relate negatively with financial development in middle-income countries.
Hypothesis 3c
H0: Public debt relates positively with financial development in high-income countries.
18 3. Data and method
3.1 Data
The empirical research covers a period between 1988 and 2017. Countries in the sample are divided into three groups according on basis of their gross national income per capita:
low-income, middle-income and high-income. 25 Furthermore, within each income group I tried to
select countries across different continents as much as possible. Selecting different countries around the world is hard, especially for low-income countries as most countries classified as low-income by the World Bank are in Central or East Africa.26 Thereby, I expect the results to
be more homogenous because the 96 countries in the sample are eventually split by their level of income.
For financial development, the index of Sahay et al. (2015) is used. It is ought to be superior to popular measures of financial depth such as private credit to GDP or stock market capitalization to GDP because these traditional measures do not capture the multidimensional nature of financial development (Svirydzenka, 2016). The constructed index encompasses a broad measure of financial development based on financial institutions and financial markets. Thereby, it is said to be superior to the FinStats and the GFDD (Svirydzenka, 2016) and has recently been updated to 2017. To be able to get robust results, the various tests beneath are also tested for the traditional measure of financial development: credit to the private sector scaled by GDP. The World Bank provides data for this measure as domestic credit to the private sector scaled by GDP. For public debt, the Global Debt Database (GDD) (Mbaye et al., 2018) is used. Three indicators for public debt could be used from this database: public sector debt, general government debt, and central government debt (all as a ratio of GDP). Public sector debt is the most ‘complete’ measure, however, only for a few countries this data is collected. Furthermore, a distinction between the use of central and general government debt is given: “long time series on private debt tend to focus on bank credit while public debt aggregates usually refer to the (budgetary) central government” (Mbaye et al., 2018, p. 4). Other databases and debt statistics exist as well, the IMF Historical Debt Database for instance. However, this database is updated till 2015 whereas the GDD is updated till 2017. Hence, the GDD’s central government debt to GDP is used for public debt.
As the database of Chinn and Ito (2006) is most frequently used in existing literature to proxy for financial openness and it is used in this study as well. There are two simple reasons why the index of Fernández et al. (2016) -which is based on a lot more dimensions- is not used instead. One, the updated version of the Fernández et al. (2016) database does not have the timespan that is needed for this research, where data is gather from 1995 forward. Second, the Chinn and Ito (2006) database consists of more countries. Their KAOPEN and ka_open variable measures the degree of capital account openness (Chinn and Ito, 2006) and is used for financial openness.
Ka_open is preferred to KAOPEN because it concerns a normalized version (ranging from 0 to
1) whereas KAOPEN consist of unrounded numbers (ranging between 2.35 and -1.92). Concerning trade openness and institutional quality, my own calculations are used. Trade openness is frequently measured as [(export + import)/GDP * 100%] (Huang and Temple, 2005; Menyah et al., 2014; among others), therefore I use the same measurement. The World Development Indicators of the World Bank are used to retrieve data for the level of export and import per country. Specifically, the exports of goods and services (% of GDP) and imports of
25 The countries and their specific categorization are found in Appendix A. The 96 countries in the sample are split in low-, middle- and
high-income countries according to the classifications of the World Bank. Source:
https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups
19 goods and services (% of GDP).27 These two numbers are combined to get a value for the trade openness. Lastly, the two numerical dimensions -political rights and civil liberties- as
constructed by Freedom House are combined to construct an overall numerical index that proxies for institutional quality. Both dimensions range from 1 to 7. They are added up and divided by two. Thus, political rights and civil liberties receive equal weight. The variable
inflation is retrieved from the World Bank as the Consumer Price Index (CPI). The CPI reflects
the change in costs for acquiring a basket of goods and services, with 2010 being the base level at 100.
All variables mentioned above are computed as 5-year non-overlapping averages. There are two reasons why the variables are constructed in this way. First, the focus of this research is on the long run effect of the public debt-financial development link, and therefore not the short run -business cycle- effect (de Haan and Sturm, 2017). Second, macroeconomic variables are turbulent, i.e. they can fluctuate a lot between a few years. Thereby, a minimum of two years of data is required to construct a 5-year non-overlapping average for the particular timeframe. Thus, countries that have only an observation of one of the five years do not receive a value for that timeframe. The timeframes are: 2017-2013, 2012-2008, 2007-2003, 2002-1998, 1997-1993, and 1992-1988.
In the next sub-section, the econometric model is discussed. The model that is used is a panel fixed-effects model which, according to Allison (2009), have two basic data requirements. “First, the dependent variable must be measured for each individual on at least two occasions. Those measurements must be directly comparable, that is, they must have the same meaning and metric. Second, the predictor variables of interest must change in value across those multiple occasions for some substantial portion of the sample” (p. 2). Concerning the dataset constructed for this paper, the two basic data requirements of Allison (2009) to perform a fixed effect model are met. See also appendix A.
3.3 Econometric model
In order to test the relationship between these variables, a panel fixed effects model is used. The model provides an analysis of across countries as well as over time. Analyses concerning panel data face three potential difficulties: endogeneity, autocorrelation and, heteroscedasticity. For such panel data, a fixed effects (FE) estimator is used most frequently to deal with potentially endogeneity problem of the between variation (Brüderl and Ludwig, 2015). In other words, with a FE estimator in combination with panel data, “it is possible to identify a causal effect under weaker assumptions (compared to cross-sectional OLS or POLS)” (Brüderl and Ludwig, 2015, p.329).28 Thus, with the fixed-effect panel model, within-country variation is focused upon. To tackle potential autocorrelation and heteroscedasticity in a fixed effects estimator, the Huber-White robust standard errors are frequently used. However, according to Greene (2008), it is unnecessary. “If the model is correct and completely specified, then the individual effects should be capturing the omitted heterogeneity, and what remains is a classical, homoscedastic, nonautocorrelated disturbance” (Greene, 2008, p. 200). In sub-section 4.1 the decision to perform a FE estimator is also be empirically tested. Before testing the sample for the public debt-financial development link, the conventional model is tested. That is, testing the ‘standard’ variables that emerges from the literature as determinants of financial development. It also includes the trade openness and financial openness interaction variable. For ‘openness’ to have an effect on financial development, a country should have an open
27 Retrieved from the World Bank: http://datatopics.worldbank.org/world-development-indicators/themes/economy.html.
28 The Pooled Ordinary Least Square (POLS) or Random Effects (RE) estimator requires that both the between and within variation are
20 capital and trade account simultaneously (Rajan and Zingales, 2003). Consequently, model estimation (1):
(1) FDi,t = βi + β1INFi,t + β2TO_OPENi,t + β3KA_OPENi,t + β4IQ,t + β5KO_TOi,t + δt +
єi,t
All variables are calculated as 5-year non-overlapping averages. FD is the indicator for financial development. βi is the intercept and captures the average time-invariant value of the fixed
effects in the model. These time-invariant effects could be historical or geographical factors for example. The control variables INF, TO_OPEN, KA_OPEN, and IQ are, respectively, inflation, trade openness, financial openness and institutional quality. KO_TO is the interaction variable between the two types of openness. The δt variable can be thought of as a year dummy, to control for year fixed effects. It should be noted that not all year dummies should be included at once, to avoid perfect collinearity. Finally, єi,t is the error term. The estimation results should
be in line with previous literature on determinants of financial development. In model estimation (2), public debt is added to the equation as an explanatory variable as well as its square term. The square term is added to the estimation to test hypothesis 2, i.e. to test for a potential non-linearity. This looks as follows:
(2) FDi,t = βi + β1DEBTi,t + β2DEBT2i,t + β3-6CONTROLSi,t + δt + єi,t
Where DEBT is the level of public debt divided by GDP and DEBT2 its square. The
CONTROLS variable is in the model estimation for unobserved potential variables that have an
impact on financial development; inflation, trade openness, financial openness, and institutional quality. On top of the ‘openness’ interaction, three more interaction variables are included. First, the effect of financial openness on financial development is conditioned by the level of institutional quality (Chinn and Ito, 2006). Second, public debt levels may be affected by the level of institutional quality. It should be noted that when a country can provide a ‘safe asset’, it can overcome their legal and institutional imperfections (De Soto, 2000). Third, as highlighted in sub-section 2.2.1, a financially liberalized economy could have a potential offsetting effect on inflation (Ayadi et al., 2015). Respectively, these interaction variables are named KO_IQ, DEBT_IQ, and KO_INF. Including these interaction variables on top of the ‘openness’ interaction variable results in estimation (3):
(3) FDi,t = βi + β1DEBTi,t + β2-5CONTROLSi,t + β6-9INTERACTIONSi,t + δt +єi,t
21
(4) FDi,t = βi + β1DEBTi,t + β2D1DEBTi,t + β3D2DEBTi,t + β4CONTROLSi,t +
β5INTERACTIONSi,t + δt + єi,t
D1 and D2 are the dummies taking a value of 1 or 0. D1 has a value of 1 when low-income
countries are considered. Likewise, a value of 1 for D2 is concerned with middle-income countries. When both D1 and D2 have a value of 0, high-income countries are considered in
the model. The analysis if the relationship between public debt and financial development differs between the country income groups, the dummies are interacted with DEBT. Resulting in D1DEBT and D2DEBT.
