Foreign Bank Activity and Economic
Growth: Assessing the Direct Link and
Volatility Effects*
August 2007
Keywords: foreign bank activity, economic growth, credit growth, volatility
Author Supervisor
R.J.J. Hartsuiker Dr. D.J. Bezemer
s1576720 Faculty of Economics
De Kap 127, 7891 LR Landleven 5, 9747 AD
Klazienaveen, The Netherlands Groningen, The Netherlands
Abstract
Table of Contents
1. Introduction ... 4
2. Literature Review... 6
2.1 The Importance of Banking for Financial Development and Growth ... 6
2.2 Does Foreign Bank Activity Accelerate Economic Growth? ... 7
2.3 The Linkages Between Foreign Bank Activity and Economic Growth... 8
2.4 Does Foreign Bank Activity Cause Economic Growth Volatility? ... 12
3. Empirical Analysis ... 17 3.1 Data ... 17 3.2 Regression specification... 17 3.3 Estimation methodology ... 21 4. Results ... 24 4.1 Direct effect... 24 4.2 Volatility effect ... 25
5. Conclusion and Discussion ... 27
1. Introduction
Over the past 30 years, the international component of banks’ activities has been growing steadily. International claims of banks located in industrial economies rose at an average annual rate of 11.5% between 1977 and 2006. Measured against economic activity, these claims have quintupled since 1980, reaching the equivalent of 50% of world GDP by 2006. Moreover, the share of claims channeled through local offices has grown to more than 40% of total international claims. Financial sector liberalization and privatizations, in many cases in the aftermath of financial crises, have facilitated the opening of branches or the acquisition of local banks and hence induced a shift towards greater local presence (BIS, 2007).
During the 1990s, foreign bank ownership has risen most notably in the Americas and among emerging and transition economies in Europe, and less so in Asia, Africa and the Middle East. As a result, a number of banking systems are now effectively foreign owned, as in New Zealand, Mexico and some Central and Eastern European (CEE) countries (BIS, 2007).
It is widely recognized that foreign participation, often spurred by the liberalization of financial markets, can help develop a financial system and hence accelerate economic growth. In the literature on the implications of foreign bank activity for economic growth scholars generally address two main effects. On the one hand, foreign bank activity may accelerate economic growth by increasing credit supply (direct effect), while on the other hand, foreign bank activity is also likely to foster competition causing lower interest margins and thus higher financial sector efficiency (indirect effect). A tremendous amount of researchers have dealt with the indirect effect and found evidence of its significance. By contrast, research on the direct effect is less available, and the evidence to date is ambiguous.
Another issue that is dealt with in the literature, is whether foreign bank activity causes economic growth instability. Although some scholars argue that foreign banks contribute to stability (e.g. foreign banks are less exposed to default risk due to more globally diversified credit portfolios), others argue that they may cause instability arising from increased competition. In addition, too much exposure to foreign banks may increase instability in the sense that host countries can be affected by cyclical conditions in the home countries of foreign banks (credit supply instability). Finally, foreign banks may also cause credit booms and busts (supplying too much credit for speculative investments relative to the supply of credit for investments in the real sector).
finding out the effect of foreign bank activity on economic growth volatility in general. Researchers dealing with the consequences of financial development for output volatility have dealt mainly with financial liberalization, but to my best knowledge only one attempt has been made to specifically reveal the effect of foreign bank activity on economic growth volatility.
This paper contributes to the literature by reinvestigating the direct link between foreign bank activity and economic growth using a sample with a large number of developing countries1, because exactly those countries often lack sufficient credit for investments and hence an increase in foreign bank activity is likely accelerate economic growth in those countries by increasing credit supply2. Moreover, I will deal with the issue of economic growth volatility, a phenomenon that is often claimed to be caused by foreign bank activity.
The key objective of this study is to reinvestigate the direct link between foreign bank activity and economic growth and to examine the effect of foreign bank activity on economic growth volatility. Its importance lies in the fact that it can be a valuable input in the debate on global financial integration. Moreover, it is also valuable for policy makers who intend to achieve growth by attracting foreign financial direct investment or have recently seen an increase in foreign banking activity in their country and need more knowledge of its consequences for economic growth.
The following research question and investigative questions will be addressed in this paper: Does foreign bank activity lead to stable economic growth?
• Does foreign bank activity accelerate economic growth? How important is the direct
link in this case?
• What is the effect of foreign bank activity on economic growth volatility?
• How do the results change the general view towards full financial liberalization in
developing countries that is advocated especially by the WTO, the IMF and the World Bank?
The rest of this paper is organized as follows. In section two the main literature on the link between foreign bank activity and economic growth (volatility) will be reviewed. Section three will deal with the methodological framework and the dataset employed followed by section four, which discusses the empirical results. Finally, section five concludes.
1 Countries that are indicated by the World Bank as Low Income, Lower Middle Income and Upper Middle
Income
2 My paper will be complementary to Bayraktar and Wang (2006) because I extend their model and use a larger
2. Literature Review
In this section I will review the most important literature on foreign bank activity and economic growth. The first subsection will discuss the importance of the banking system for financial development and economic growth of a country. The second subsection will introduce the linkage between foreign bank activity and economic growth, whereas the subsection following it will discuss the empirical and theoretical evidence on its identified linkages more in detail. Finally, subsection four will discuss the volatility effects of foreign bank activity.
2.1 The Importance of Banking for Financial Development and Growth
Considerable debate exists on the relationship between financial development and economic growth. Historically, economists have focused on the banking sector. Bagehot (1873) and Schumpeter (1912) emphasize the critical importance of the banking system in economic growth and highlight circumstances when banks can actively spur innovation and future growth by identifying and funding productive investments. In contrast, Lucas (1988) states that economists 'badly overstress' the role of the financial system. Empirically, King and Levine (1993) show that the level of financial intermediation is a good predictor of long-run rates of economic growth, capital accumulation, and productivity improvements. This is result is confirmed by Levine and Servos (1998) who find that banking development is positively and robustly correlated with contemporaneous and future rates of economic growth, capital accumulation and productivity growth.
The result that banks spur growth to a great extent by accumulating capital is highly disputed by other researchers3. In line with this, Levine (2001) states that financial systems exert a large, causal impact on economic growth primarily by boosting Total Factor Productivity (TFP), which usually is seen as an indicator for technological change, quality advances and resource allocation enhancements. Easterly and Levine (2000) find that TFP growth accounts for 90% of cross-country growth differences leaving only 10% to be explained by the physical capital stock (capital accumulation). However, Rioja and Valev (2004a) find that finance boosts growth in rich countries primarily by speeding up productivity growth, while in poor countries by accelerating capital accumulation. They also find that countries with very low levels of financial development experience very little acceleration of economic growth from a marginal increase in financial development, while the effect is larger for rich countries and particularly large for middle-income countries (Rioja and Valev, 2004b). In general, theoretical and empirical studies show a strong positive relationship between financial development and growth (Levine, 2004)4.
3 Krugman (1993) shows that physical capital accumulation does not account for much of the cross-country
differences in growth rates and Blomstrom et al. (1996) show that more investment does not cause faster economic growth
2.2 Does Foreign Bank Activity Accelerate Economic Growth?
Since the 1990s a tremendous amount of research has been carried out on the effects of foreign bank activity in the host country. Most attention in this area has been devoted to effects of foreign bank activity on competition and efficiency improvements in the domestic banking sector, financial system stability and credit growth & credit allocation5. Although a lot of researchers have dealt with separate effects of foreign bank activity on the host economy, a much smaller number of scholars has focused on the effects of foreign bank activity on economic growth. In line with Goldberg (2003) I note that we have to distinguish between two strands of literature that specifically question whether financial FDI stimulates emerging market growth. One strand of research looks broader than financial FDI per se, and considers the growth implications of financial liberalization. The second strand of research, the focus of this paper, deals with the effects of foreign bank activity on aggregate growth rates6. Though foreign banks sometimes enter as a component of large scale financial liberalization and banking privatization effort, they may also enter as local governments seek to recapitalize their financial systems in the wake of a crisis.
In general, it is acknowledged that foreign bank activity is likely to have a positive effect on economic growth, especially in the long run. However, foreign bank activity may also cause financial instability and hence economic volatility (this will be discussed in paragraph 2.4). In this respect Kaminsky and Schmukler (2002) talk about the intertemporal trade-off between ‘short-run pain and long-run gain’ as a result of financial openness. In the short-run, the bank fragility induced by financial opening may result in financial crisis, but it can enhance growth potential in the long-run once structural deficiencies are eliminated in the wake of overcoming this crisis (Aizenman, 2002). Using a panel, Loayza and Ranciere (2005) also find that a positive long-run relationship between financial development and growth co-exists with a generally negative short-run link. Still, other scholars (like Demirgüç-Kunt et al. 1998, Claessens et al., 2001 and Bayraktar & Wang, 2006) have found positive effects of foreign bank activity on economic growth in the short run as well7.
Hypothesis 1
An increasing share of foreign banks in a country will, all other things equal, have a positive effect on economic growth (in the short run)
and Damar (2006) even find a negative relationship between financial depth and economic growth for Turkey
5 For an overview of the benefits and caveats of foreign bank activity, see Cardenas et al. (2003), Moreno &
Villar (2005), Uiboupin (2005) and Cull & Martínez Pería (2007)
6 Although this paper is part of the second strand of literature, it does not mean that there are no valuable
contributions in the first strand on financial liberalization that can be used as an input for the second strand regarding the separate effects of foreign bank activity (research within this second strand is less available)
7 Voinea and Mihaescu (2006) found that an increase in foreign bank activity is correlated with an increase in the
2.3 The Linkages Between Foreign Bank Activity and Economic Growth
How does foreign bank activity actually contribute to economic growth? The most detailed analysis in the literature is provided by Eller et al. (2005) studying the consequences of foreign bank activity in 11 Central and Eastern European (CEE) countries. They identify four transmission channels through which foreign ownership in banking may affect economic development: intermediation/efficiency, intermediation/credit volume, corporate governance and institution building, as well as signal effects for total FDI and portfolio investments (see figure 1). In the short run, only the first two channels are important.
Figure 1: Identified Transmission Channels between FSFDI and Economic Growth
Source: Eller et al. (2005), p. 6 Intermediation / Efficiency
The channel that has received the most attention from scholars on the relationship between foreign bank activity and economic growth is definitely the efficiency channel8. It is widely believed that foreign banks are more efficient, because they have lower operating costs9. Moreover, foreign banks are likely to have more efficient credit allocation as well as sound monitoring and thus less risk (Eller et al., 2005).
If foreign banks are more efficient than domestic banks, foreign bank activity may directly affect the efficiency of the host country financial system. Higher competition induced by foreign bank activity causes lower interest margins10 and thus higher financial sector efficiency, which results in an overall reduction of transaction costs triggering investments
8 Bayraktar and Wang (2006) refer to this channel as the ‘indirect effect’ whereas the increase in credit volume is
referred to as the ‘direct effect’
9 This cost efficiency can be realized only after a period of experimental learning in the new market and
restructuring in the case of acquisitions (Eller et al., 2005). Moreover, a recent paper by Havrylchyk (2006) on foreign bank activity in Poland showed that the level of efficiency that can be achieved is higher for Greenfield banks. This result is confirmed by Claeys and Hainz (2006) who relate it to increased competition in contrast with activity through M&A
10 Although foreign banks might increase their margins a bit after gaining market share, the increase in
and economic growth (Holló and Nagy, 2006). Hence, foreign-owned private banks are expected to facilitate technology transfer, competition and efficiency in the banking sector of the host economy, enhancing the efficiency of domestic savings, which may be intermediated into domestic investment (Lensink and Murinde, 2006). Koivu (2004) found evidence that increasing financial sector efficiency measured by interest margins has growth-enhancing effects on economies in transition applying country and time series regressions on nine CEE countries over 1995-2002. However, Haber and Musacchio (2005) found that in the case of Mexico, the activity of foreign banks led to a retrenchment of lending and no improvements in efficiency and competition. In addition, Lensink and Murinde (2006) found a robust inverted U-shaped relationship between FSFDI and gross domestic investment suggesting that foreign bank activity leads to investment expansion only after foreign bank presence becomes large enough as a share of local banking activity11. Eller et al. (2006) also find this hump-shaped relationship between FSFDI and economic growth through the efficiency channel. They state that medium FSFDI supports growth if human capital suffices, whereas above a certain threshold, crowding-out of local physical capital via foreign bank activity slows growth.
It should be noted that banks will need to refocus their business and specialize on (other) new target groups leading to higher costs for the domestic banks. They might succeed by offering services more closely tailored to the needs of the local population instead of trying to compete on price with foreign banks who are usually backed by large financial groups (Eller et al. 2006). They also might be crowded out as foreigners ‘cherry pick’12 and only riskier target groups remain, which may result in risk-taking behavior causing instability (IADB, 2005). Finally, another negative effect is that while competition will increase, especially cross-border mergers and acquisitions contribute to growing concentration in the financial system of emerging market economies13 (BIS, 2004; BIS 2005; Hainz and Claeys, 2005).
All in all the empirical evidence on competition and efficiency suggests that foreign bank activity can bring potential gains in this area except in environments which limit competitive forces such as when bank concentration is high, bank activities are restricted, and bank entry and exit is difficult (Cull and Martínez Pería 2007, p. 12).
11 The authors explained this by the fact that the spillover effects of foreign bank activity may lead to an increase
in costs of domestic banks in the short run. If the value of foreign bank activity is low, and thus competitive pressure is low, domestic banks may be able to pass on increased costs due to spillover effects to their clients in the short term (Lensink and Murinde 2006, p. 573)
12 This ‘cherry picking’ is confirmed by studies such as Berger et al. (2001). Detragiache et al. (2006) and Mian
(2006) suggesting that foreign banks only serve the large and most transparent firms. In contrast, studies by Clarke et al. (2006), De Haas & Naaborg (2005) and Giannetti and Ongena (2005) show that though larger firms generally benefit more, also SMEs benefit from foreign bank activity
13 Mamatzakis et al. (2005) found evidence for monopolistic competition in South Eastern Europe over the
Intermediation / Credit Volume
The impact of foreign bank activity on credit supply in general depends on the form of market activity. In the case of an acquisition, the existing client base is maintained, not necessarily changing the credit volume. If a new subsidiary is established, the number of financial intermediaries will increase. In both cases better risk assessment allows foreign-owned banks to finance higher risk/return projects. They have the ability to provide fresh money to the financial host market because foreign-owned banks are usually backed by their parent companies (Eller et al. 2005, p. 9). The empirical evidence on this channel to date is mixed14. However, there is some evidence that foreign bank presence can have positive effects on credit access, even if foreign banks are not lending to small firms15. Still, after a financial crisis credit volume might drop because in many cases foreign banks acquire distressed banks (Cull and Martínez Pería, 2007).
Credit supply to different target groups has various impacts on economic development. Fink et al. (2004) argue that lending to the private sector is necessary to further support private investment, whereas providing finance to an efficient state and thus improve infrastructure efficiency can be a major way to reasonably foster economic growth during transition. Breyer (2004) complements this view by stating that foreign banks finance budget and current account deficits. However, the most important difference between foreign and domestic banks is found in case of private credit: foreign banks are much more involved in lending to the private sector than domestic banks16 (Eller et al. 2005, p. 10).
Corporate Governance and Institution Building
An important objective of many governments opening the financial market to foreign banks, is to improve the quality of the banking sector and thereby the quality of the whole financial system. Foreign-owned banks usually need to stick to international standards and comply with internal group-wide rules which contributes to a reduction in bad loans. This might contribute to financial sector stability which is positive for economic growth. Moreover, the implementation of international standards, e.g. stricter credit requirements, will create the need for companies seeking external finance to adapt to these standards. This is likely to lead to an overall improvement in corporate governance (Eller et al. 2005, p. 13).
As foreign banks enter emerging markets, the introduction of new types of products or
14 De Haas & Van Lelyveld (2002) show that local credit by foreign-owned banks to domestic credit as well as
to GDP rose in the 1990s, except for Slovenia. Clarke et al. (2006) find that all enterprises, including SMEs, report facing lower financing obstacles in countries having higher levels of foreign bank presence. In contrast, Engerer & Schrooten (2004) find no empirical evidence for an impact of foreign bank entry on financial depth in terms of credit volume in eight CEE countries over 1995–2002. Finally, Detragiache et al. (2006) using data for 89 lower (middle) income countries, show that foreign bank participation may lead to lower aggregate credit
15 Bayraktar and Wang (2006) also found evidence of the direct effect in a panel of 28 countries that were fully
liberalized in the period 1995-2002
16 As is the case within the efficiency channel, the level of banking liberalization (restrictions) also plays a
services is faster and innovation can even be accelerated via FSFDI. This creates the need for supervisors to adapt the legal environment to these developments. If regulations for new services are not in place or are not accurately and fast enough adapted, abuse will occur, which will harm the financial sector and the whole economy (Bonin & Wachtel, 2002). Therefore, local supervisors have to upgrade their knowledge and further adapt regulations in order to secure financial sector stability. Foreign-owned banks seeking to mitigate their own risk act as a catalyst for regulatory changes and implementation of international standards (BIS, 2004). In this way, foreign-owned banks can contribute to institutional quality (Eller et al. 2005, p. 14).
Signal Effects
The final channel mentioned by Eller et al (2005, p. 14) consists of ‘signal effects’. Foreign-owned banks may strive to gain higher market share through product innovation and by offering a variety of new financial services, such as asset management services, which leads to the development of new market segments. Rising national income together with pension reforms may add to the demand for and implementation of tradable securities as well for private investors (BIS 2004, 13). Product innovation and the need of local risk-management to hedge risks locally foster capital market development. In consequence, corporate investors may chose from a greater range of finance possibilities which may spur investment and economic growth. Additional non-financial portfolio investment as well as non-financial FDI might be drawn in, which in turn influences economic growth (Durham, 2003).
From the literature on the relationship between financial development and growth, we know that economic growth is to a great extent caused by productivity improvements, rather than capital accumulation (Bonfiglioli, 2005). This means that the most important effect of foreign bank activity is the indirect effect, as stated by Levine (2001) and Eller et al. (2006), whereas the direct effect is less important. Recall that except Bayraktar and Wang (2006), other scholars have not found robust empirical evidence of this direct effect. However, we also know that Rioja and Valev (2004a) found that finance boosts growth in poor countries by accelerating capital accumulation. Hence, I argue that foreign bank activity can accelerate economic growth through increased credit supply especially in developing countries which often lack sufficient capital for investments. However, the level of institutional development (credit market imperfections) is likely to be crucial for this transmission channel.
Hypothesis 2
2.4 Does Foreign Bank Activity Cause Economic Growth Volatility?
The rise in foreign bank participation in many developing countries has triggered a discussion about the effect of foreign bank activity on the stability17 of the financial system. Theoretically as well as empirically a lot of research has been done in this area. However, most papers on the volatility effects deal with the general effect of foreign bank activity on the probability of a financial crisis or with a specific channel through which banks may cause instability. To my best knowledge, only one paper deals specifically with the effect of foreign bank activity on GDP volatility.
In this subsection I will try to give a comprehensive review of the two major channels causing this volatility: credit supply instability and credit booms. Moreover, I will review the existing evidence on the link between financial openness/foreign bank activity and economic growth volatility. However, I will start with the general literature on whether foreign bank activity causes financial crisis leading to a slowdown in physical capital investments and innovation and hence hampering economic growth.
Proponents of foreign bank activity often argue that foreign banks play a stabilizing role in the domestic financial system. In favor of this, is the notion that foreign banks are well diversified institutions with access to many sources of liquidity that will be less affected by shocks (Cull and Martínez Pería, 2007). Moreover, during periods of banking stress, foreign bank presence could diversify against country-specific systemic risks, because foreign banks are diversified across different countries and thus less sensitive to host country cycles, which could well change the cyclical behavior of the host country financial system. Foreign banks could also be more resilient during currency crises. Not only do they tend to be more aware of currency mismatches, they can also call on their parent organizations to provide foreign currency liquidity. However, opponents of foreign bank activity will argue that a large foreign banks could import shocks from their home countries (contagion), such as those affecting the parent bank (Moreno and Villar 2005, pp. 11-12; Lee, 2002). In line with IADB (2005) they may also state that foreign banks also might cut back on local operations rapidly, because they have lower exit costs, depending on the form of market activity, leading to credit swings.
A number of empirical studies dealt with the specific question whether foreign bank activity is associated with a greater probability of financial crisis. For example, a study by Demirgüç-Kunt and Detragiache (1998) showed that financial liberalization has costs in terms of increased fragility, especially in developing countries where the institutions needed to
17 I use the terms instability and volatility interchangeably, because they basically refer to the same economic
support a well-functioning financial system are generally not well established18. However, those studies do not consider the contributions of banks sorted by ownership structure. As Crystal et al. (2001) show, foreign-owned banks may contribute to the overall soundness of local banking systems via more aggressive screening and treatment of problem loans19. (Goldberg 2003, p. 10).
Empirical evidence on the possible stabilizing influence of foreign ownership is also provided by Engerer & Schrooten (2004) who analyzed eight CEE countries from 1995 to 2002. Their results as well as evidence provided by the IADB (2005) emphasize several advantages of foreign-owned banks in this regard. Foreign parent companies might act as a lender of last resort to their local units and related private institutions. Furthermore, foreign-owned banks are less exposed to local default risk due to the higher degree of global (risk) diversification and their often long-term interest. Beyond it, they argue that less volatile deposits and higher loan quality of foreign-owned banks compared to domestic banks add to it. Better disclosure, accounting, and reporting practices as well as stronger prudential supervision are crucial for positive impacts of foreign bank activity on lending practices and financial sector stability (Eller et al. 2005, p.12). A recent study by Cull & Martínez Pería (2007) also found no evidence that foreign bank activity is associated with greater probability of crisis.
Credit supply instability
One of the channels through which foreign banks influence economic growth volatility is through the volatility of credit supply. Though some studies have found that foreign banks can respond to shocks from their home countries (e.g. Martínez Pería et al., 200520), a larger number of studies have found that they tend to be more stable lenders than domestic banks, in particular during periods of crisis in developing countries. For example, using bank level data for the late 1990s for Argentina, Chile and Colombia, Crystal et al. (2001) show that foreign banks on average exhibited higher loan growth rates than domestic banks and Detragiache and Gupta (2006b) also found no evidence that foreign banks abandoned the local market during the 1997-98 Asian crisis, when examining the behavior of banks in Malaysia. Finally, De Haas and Van Lelyveld (2006) examine how foreign and domestic banks in ten Central
18 When an economy has strong institutions, the impact of financial liberalization on the fragility of banking
system will be mitigated through changes in institutions supporting a better functioning of financial market (Kaminsky and Schmukler, 2002). Bonfliglioli (2005) also found that institutional development may limit the damages of a crisis
19 Taken together, their findings that foreign banks have consistently stronger average credit growth, take more
aggressive action to deal with asset quality deterioration, and possess the capacity and willingness to sacrifice short-term profitability for longer-term soundness, suggest that foreign ownership may have quite positive implications for financial sector stability, development and efficiency
20 They find, studying Colombia, Chile and Argentina in depth, that foreign banks may transmit external shocks
and Eastern Europe (CEE) reacted to business cycle conditions and host country banking crises from 1993 to 2000. They show that while during crises domestic banks contracted their credit, foreign banks maintained their credit supply (Cull and Martínez Pería 2007, pp. 12-13).
Credit booms
Growing credit supply is not enough to guarantee a positive impact on investment activities and thus economic development. Fast credit growth, usually called a lending boom, can also be a warning signal indicating a potential financial crisis (Mehl & Winkler 2003). One of the reasons for the emergence of those credit bubbles might be the increase in competition21 caused by foreign bank activity raising the risk appetite of banks in the fight for market share (Eller et al. 2006). There is evidence that these lending booms did occur in many CEE transition countries after financial liberalization (EBRD, 2005). Kraft and Jankov (2005) also discovered a lending boom after financial liberalization in Croatia, but found that asset quality problems have not yet emerged22. In addition, Adalid and Detken (2007) discovered 42 lending booms in 18 OECD countries in the period 1970 – 2004. They note, that none of these booms resulted in a financial crisis, and that not all the booms led to a decline in real GDP growth in the 3 years after the boom23 (in 22 out of 42 cases this happened).
How do these credit booms actually cause economic volatility? The activity of foreign banks is associated with an enormous increase in credit growth that foreign banks want to invest in the fight for market share. However, foreign banks find that there are limited possibilities in developing countries to spend their money and search for new investment opportunities. Because of this, foreign banks might engage in ‘speculative’ financing instead of financing real sector activities24 (some institutions may even be drawn away from investing in the real sector at all) which leads to a rise in financial asset prices25 (house, land). These lucrative speculative investment opportunities also attract investors making them switch to short-term financing of long-term investment projects. This will make the financial system instable, as viability of projects becomes to rely on short-term interest rates and credit availability.
21 An other possible effect of increased competition is that domestic banks might also be crowded out as foreign
owners “cherry pick” and only riskier target groups remain. This might have a delirious impact on stability if it causes the bank’s charter value to drop, thus reducing the incentives for prudent risk-taking behavior (Mamatzakis et al. 2005)
22 Kraft & Jankov (2005, p. 115) also state that credit booms (asset quality deterioration) are not the sole or the
main predictor of banking crisis
23 Still they may slow growth
24 This sector develops in the aftermath of financial liberalization, generating extremely liquid assets.
Speculative investment in these assets can generate good rewards quickly
25 When lending booms lead to asset price increases, there is the potential for a financial accelerator as increased
Several econometric studies have confirmed the existence of a link between rapid credit growth and banking system fragility. Demirgüç-Kunt and Detragiache (1998) find evidence supporting the idea that lending booms precede banking crises. For example, according to their estimates, in the case of the 1994 Mexican crisis, a 10 percent increase in the initial value of lagged credit growth would have increased the probability of a crisis by 5½ percent. Similarly, Hardy and Pazarbasioglu (1998) find that there is a robust evidence that credit to the private sector follows a boom-and-bust pattern ahead of banking crises. Finally, Gourinchas et al. (1999) examine a large number of episodes characterized as lending booms and find that the probability of having a banking crisis significantly increases after such episodes. Moreover, the conditional incidence of having a banking crisis depends critically on the size of the boom. Nevertheless, they remark that such probability remains below 20 percent indicating that while most banking crisis may be preceded by lending booms, most lending booms are not followed by banking crises (adapted from Cottarelli et al. 2005, p. 44)26. More recent evidence comes from Ranciere et al. (2004) and Tornell et al. (2004) who argue that banking crises may arise as a by-product of the higher growth generated by financial liberalization, in countries with credit market imperfections (many of them are developing countries). However, there are also scholars arguing that these credit booms are just a part of financial deepening eventually benefiting the economy.
Finally, there is also a group of researchers that have dealt with the consequences of financial openness on macroeconomic volatility. However, to my best knowledge there is only one empirical paper dealing specifically with the case of foreign bank activity, which is a paper by Morgan and Strahan (2003). These authors found that foreign bank activity is either unrelated to volatility of firm investment spending or positively related. Moreover they found that the impact of firm capital shocks is amplified by the presence of foreign banks27.
In addition, several authors analyzed the link between financial openness and volatility. For example, Buch et al. (2002) use data for 25 OECD countries to examine the link between financial openness and business cycle volatility. They report that there is no consistent empirical relationship between financial openness and the volatility of output. O’Donnell (2001) examines the effect of financial integration on the volatility of output growth over the period 1971–94 using data for 93 countries. He finds that a higher degree of financial integration is associated with lower (higher) output volatility in OECD (non-OECD)
26 Goldberg (2003) states that foreign banks do not appear to magnify boom-bust cycles in emerging markets 27 By contrast, Caballero and Krishnamurthy (2001) suggest that the severe credit constraints in emerging market
countries28. Bekaert et al. (2005) examine the impact of equity market liberalization on the volatility of output and consumption during 1980–2000. They find that capital account openness increases the volatility of output and consumption in emerging market countries. IMF (2002) provides evidence indicating that financial openness is associated with lower output volatility in developing countries. Finally, Kose et al. (2003) show that financial sector development, proxied by M2/GDP, is associated with lower output volatility.
All in all, the evidence so far shows no increase in economic growth volatility coming from foreign bank activity. However, especially developing countries might be more vulnerable regarding macroeconomic volatility, because of the generally lower level of institutional development in those countries (i.e. credit market imperfections). Moreover, large increases in credit supply have not been found to cause financial crisis, but they are still likely to cause GDP volatility, especially when these credit booms lead to an increase in speculative investments relative to investments in the real sector. So, I argue that foreign bank activity might cause economic growth volatility by ‘overlending’29.
Hypothesis 3
A higher level of foreign bank activity is associated with an increase in economic growth volatility
28 This result is similar to the findings by Easterly et al. (2001)
29 Overlending means that more credit is supplied than can be invested in the real sector, which will induce
3. Empirical Analysis
The aim of the empirical section of this paper is to investigate the direct link between foreign bank activity and economic growth and the link between foreign bank activity and economic growth volatility. In the first subsection, information about the dataset is given. In the second and third subsections, the regression specifications designed and the empirical methodologies to solve them are introduced.
3.1 Data
For my analyses I use two panel datasets: one to estimate the direct effect of foreign bank activity on economic growth consisting of 111 countries and another to estimate the volatility effect of foreign bank activity consisting of 45 countries. Both panels are unbalanced and cover the years 1995 to 200230. Because of this relatively short time period, the focus in this paper will be on short-term economic growth.
The most important data source I use is the Bank Ownership and Performance database31 constructed by Micco et al. (2004, 2006) from the Fitch BankScope database providing level data on the foreign bank asset share, banking sector concentration and country-level averages of bank-country-level variables like the net interest margin, overhead costs and before tax income32.
BankScope contains an ownership code that classifies banks as state-owned, private domestic, and foreign, but the code is available only for a subset of banks and it refers only to the last year in the database. BankScope also provides some historical information about banks, including changes in ownership, but the information is not exhaustive. A bank is classified as foreign if at least 50 percent of its capital is in the hands of non-residents. The share of foreign bank assets to total bank assets in each country is measured by their asset shares in total assets in the banking sector. The asset share of foreign banks in each country is presented inAppendix C.
3.2 Regression specification
I use two different regression specifications to study the direct link between foreign bank activity and economic growth and the link between foreign bank activity and economic growth volatility. The economic model for the direct link is based on a standard way to model economic growth, starting with a Cobb-Douglas production function, which is given by Y = AK L1- , with Output (Y), Capital (K), Labor (L), and A representing a Technology
30 In order to maximize the size of the sample, I also include countries for which data are available in 4-7 years 31 A limitation of this dataset is that BankScope changes over time, so it is possible that changes in measured
foreign bank penetration are just the effect of changes in coverage (Detragiache et al. 2006)
32 Appendix A gives detailed information about all the variables and the samples and Appendix B presents
factor/Profitability shock. Assuming that the production function is already maximized for labor and that there are no exogenous shocks, I focus on the on the capital part alone, so Y = K . One will notice, that this function is non-linear and hence to make it linear I take the logarithm of both sides, which indicates that I will use a log-log model. It is specified by the following equation:
GDPCi,t – GDPCi,t-1 = + 1GDPCi,t-1 + 2FBASi,t + 3CVi,t + 4[CVi,t × FBASi,t]+
5Xi,t + 6Tt + i + i,t (1)
where i is the country index, t is the year index, i is an unobserved country-specific effect
and i,t is the error term. Finally, Tt represents a time dummy. The dependent variable is
economic growth measured as the difference of logged GDP per capita in real terms33, GDPCi,t – GDPCi,t-1. As a first regressor I include initial GDP per capita to control for initial
conditions. Including this lagged dependent makes my equation dynamic. CVi,t indicates
credit volume, FBASi,t represents the foreign bank asset share in each country and Xi,t is a
vector of control variables. The interaction term shows the expected direct effect: foreign banks accelerate economic growth by increasing credit supply34.
The control variables I use are chosen according to their significance in determining growth and their potential effects on growth through private capital accumulation, and include variables to control for the efficiency of financial markets, macroeconomic stability, public infrastructure, public spending, openness, political risk/stability and institutional development35.
In line with Bayraktar and Wang (2006, p. 14) I include three variables to control for the efficiency of financial markets, which may play an important role in reducing the cost of capital and hence in stimulating investment. These are the level of net interest margin, profits before tax, and overhead costs. The net interest margin, defined as the ratio of net interest income to total assets, shows the difference between earnings from interest and expenses on interest and is an important indicator of competitiveness, since as the banking sector becomes more competitive the average lending rate is expected to drop, while the average deposit rate is expected to increase. This leads to a drop in the net interest margin as well. The share of before tax profits in total assets is another control variable for efficiency. It is expected that
33 Note that my dependent will take the form log(GDPC
i,t) – log(GDPCi,t-1) = log (GDPCi,t/GDPCi,t-1)
34 I remark that Bayraktar & Wang (2006) argue that the variables indicating efficiency may also capture
financial depth (the supply of funds available to the private and government sector) and hence they exclude measures like liquid liabilities to GDP and private credit to GDP (measuring credit volume). However, my sample includes a lot of developing countries and hence I do expect to find a significant separate effect of financial credit on economic growth
35 Although human capital development is also important for economic growth, it is excluded because of the
profits will fall with rising competition induced by more efficiency. Banks will now have to pay higher interest rates for funds, they can charge lower interest rates on loans and charge lower service fees compared to what they can in less competitive banking sectors. Finally, the ratio of overhead costs to total assets is also expected to fall with increasing competition.
Macroeconomic instability is known as an important deterrent of economic growth and to control for instability I use the inflation rate. It is expected that higher inflation tends to reduce growth through a higher level of price instability (Bayraktar & Wang, 2006). Moreover, Bordo and Rousseau (2006) argue that high inflation will inhibit long-term financial contracting, causing financial intermediaries to keep their portfolios rather liquid, which in turn retards economic growth.
Government expenditure as a ratio to GDP is also included. According to Barro and Sala-i-Martin (1995), government consumption proxies for political corruption, bad government, as well as for direct effects of non-productive public expenditures. A large component of government expenditures are wages and salaries and they have been showed to be unambiguously associated with lower growth. Moreover, several theories predict that private investments are crowded out by government expenditures (Bonfiglioli, 2005).
I will also control for country risk by using the composite political risk index from International Country Risk Guide (ICRG). This index is constructed in a way that higher numbers indicate lower risk. However, it has been inverted, so a negative link between economic growth and this composite risk index is to be expected36. Trade openness is also expected to be important for economic growth and hence will be controlled for by the ratio of net exports to GDP. Trade may effect the efficiency of an economy through several channels such as specialization according to comparative advantage, access to larger markets with more product variety and increased competition. These effects may stimulate both capital accumulation and productivity growth (Bonfiglioli, 2005).
Finally, I include a control variable for public infrastructure. In line with Bayraktar and Wang (2006, p. 15), the average number of main telephone lines per 1,000 people is included to proxy for public infrastructure, which is an important determinant of the level of investment and hence economic growth given that public and private capital stocks are complements37. Table 1 (on the next page) summarizes all the variables used for estimating the direct effect and the expected sign of the coefficients in the regression.
36 The country risk index I use also includes some measures of institutional quality (like regulatory quality, rule
of law and control of corruption), an important determinant of economic growth. Nallari and Griffith (2006) argue that ‘the right institutions are needed to encourage appropriate policies both for the accumulation of capital and for developments that will improve productivity (new technologies, managerial processes, education), all of which affect economic growth’. However, they also point out that institutions are endogenous to economic growth: good institutions stimulate growth and growth encourages the development of good institutions
37 I only proxy by telephone lines because other measures like roads (in km) and rail lines come with a lot of
Table 1: Overview of Variables – Direct Effect
Variable Expected Sign
Log difference of GDP per capita -
Initial GDP per capita < 0
Foreign bank asset share > 0
Credit volume > 0
Direct link: foreign bank asset share × credit
volume > 0
Trade openness > 0
Inflation rate > 0*
Political risk < 0
Government expenditure level to GDP < 0 Main telephone lines per 1,000 people > 0
Net interest margin > 0*
Overhead costs > 0*
Before tax profit > 0*
Source: author’s table
* These variables had to be inverted to be able to transform them into
logs, which indicates that the positive sign actually represents a negative effect
To address whether foreign bank activity is associated with increased economic growth volatility (also called business cycle volatility) I specify the following equation:
GDPVi,t = + 1FBASi,t + 2Xi,t + 3Tt + 4Fi + i,t (2)
where i is the country index and t is the year index. I also include time dummies and country dummies, which are represented by Tt and Fi respectively. The dependent variable, GDPVi,t,
is economic growth volatility measured on a yearly basis. It is calculated by taking the standard deviation of four quarterly observations on GDP. After the constant , FBASi,t
represents the foreign bank asset share in each country, Xi,t is a vector of control variables and i,t is the error term.
The control variables are chosen according to their significance in determining economic growth volatility (which is related to output volatility). Banking sector concentration is used as a control variable because Morgan and Strahan (2003) found this variable to be significantly causing volatility. They argued that bank risk taking tends to increase as concentration (and the associated rents, or bank charter value) falls. Safer banks may translate into safer, less volatile economies. Moreover, they claim that banking sector concentration will also be likely to affect the political game in determining the barriers of foreign banking, and hence will indirectly affect growth volatility through its effect on deregulation.
financial system might reduce growth volatility on average. However, they argue that this relationship is non-linear (squared term is positive) and hence as the financial system becomes larger relative to GDP, increases in risky behavior act to reduce stability. Findings by Morgan and Strahan (2003) and Aghion et al. (2004) contrast this view and find a positive relationship between financial sector development and economic growth volatility.
Other variables included to control for economic volatility are money growth, trade openness (more open economies are more vulnerable to external shocks), and country risk (IRCG country political risk index). In line with Morgan and Strahan (2003) I will also control for the effects of exchange rate volatility (terms of trade shocks) by including the absolute value of the change in the real (effective) exchange rate for a given country. Table 2 summarizes all the variables used for estimating the direct effect and the expected sign of coefficients in the regression.
Table 2: Overview of Variables – Volatility Effect
Variable Expected Sign
GDP volatility -
Foreign bank asset share > 0 Banking sector concentration > 0 Financial sector development > 0
Trade openness > 0
Political risk > 0
Money growth > 0
Terms of trade shocks > 0 Source: author’s table
3.3 Estimation methodology
I use two different strategies to estimate the direct effect and the volatility effect. Given the dynamic nature of the equation I specified to estimate the direct effect of foreign bank activity on economic growth, I use the Generalized Method of Moments (GMM) estimation method for dynamic panels. The key to GMM is a set of moment (orthogonality) conditions that are derived from the assumptions of the econometric model, which should hold to get a consistent unbiased estimator. These moment conditions are defined under the assumptions that (a) the error term is not serially correlated and (b) the explanatory variables are weakly exogenous: they are assumed to be uncorrelated with future realizations of the error term and thus are not affected by future realizations of the dependent variable. Given data on the observable variables GMM finds values for the model parameters such that corresponding sample moment conditions are satisfied as closely as possible (Pynnönen, 2007).
caused by the possibility that some of my independent variables may be endogenous with economic growth (Bayraktar & Wang 2006, p. 16). GMM controls for the potential endogeneity of all explanatory variables by using lagged values of these explanatory variables as instruments. Another advantage of GMM is that it is usually robust to violations of homoskedasticity and normality (GMM requires nothing about the shape of the distribution of the error terms, so the normality assumption can be dropped)38. Finally, I remark that my panel is ‘short and wide’: the time dimension (T) is small (8) and the cross-sectional dimension (N) is large (111). Judson and Owen (1999) argue that when you have a regression with a lagged dependent variable, an unbalanced panel and a T 10, the LSDV model (fixed effects model) leads to inconsistent and biased estimators (called the Nickell bias). Using a Montecarlo approach, they found that it is better to use GMM in this case, because it produces the least biased estimates.
A first way of applying this method is taking the first difference of equation (1) in order to control for unobserved country-specific effects. In this case, suggested instruments are lagged observations of the lagged dependent and explanatory variables taken in levels. However, this difference estimator has conceptual and statistical shortcomings. Conceptually, I would also like to study the cross-country relationship between foreign bank activity and per capita GDP growth, which is eliminated in the first-difference estimator (Carkovic and Levine, 2002). Moreover, when the explanatory variables are persistent over time, lagged levels of them are weak instruments for the equation in differences, especially in smaller samples (Bayraktar and Wang, 2006; Carkovic and Levine, 2002). Therefore it is suggested in the econometric literature to estimate a system of equations, combining the regression in levels and first differences reducing the possible bias associated with estimating the regression equation in first differences only (Arellano & Bond, 1991; Bond et al., 2001).
Applying this methodology, the following system of equations corresponding to equation (1) is estimated by GMM:
GDPCi,t – GDPCi,t-1 = 1GDPCi,t-1 + 2FBASi,t + 3CVi,t + 4[CVi,t × FBASi,t]+
5Xi,t + 6Tt + i + i,t
(GDPCi,t – GDPCi,t-1) = 1 GDPCi,t-1 + 2 FBASi,t + 3 CVi,t + 4 [CVi,t × FBASi,t]
+ 5 Xi,t + 6 Tt + i,t
While, estimating this system, the following orthogonality (moment) conditions are used: E( i,t × GDPCi,t-1) = 0, E( i,t × FBASi,t-1) = 0, E( i,t × CVi,t-1) = 0, E( i,t × [CVi,t-1 ×
FBASi,t-1]) = 0, E( i,t × Xi,t-1) = 0, E( i,t × GDPCi,t-2) = 0, E( i,t × FBASi,t-2) = 0, E( i,t ×
CVi,t-2) = 0, E( i,t × [CVi,t-2× FBASi,t-2]) = 0, E( i,t × Xi,t-2) = 0. The instrumental variables
for the equation in first differences are the twice-lagged level of the dependent and explanatory variables. On the other hand, the instrumental variables for the equation in levels are the most recent lagged differences of the variables.
To check whether the instruments used in estimating the equations are valid, and hence GMM is consistent, I address the Sargan test39 for over-identifying restrictions which tests the overall validity of the instruments. The null hypothesis of this test is that all instruments are valid. Hence, non-rejection gives support to the estimated model and tells us that the moment conditions hold. In addition, I will test for second-order serial correlation of the error term in the first-difference equation40. Finally, I will test for the sensitivity of my results by using different measures of banking sector efficiency.
To estimate the equation specified for the link between foreign bank activity and economic growth volatility I will use a Least Square Dummy Variable Model (also called Fixed Effects Model)41. Since, my primary interest lies in testing whether the behavioral relationship predicting economic growth is the same across countries and over the 8 years period, the slope coefficients of the prediction equation are assumed not to vary neither from one country to the other nor from one year to the other (implying homogenous coefficients). Beyond it, differences across countries can be captured by differences in the constant term (intercept). The intercept is assumed to vary over the cross-section unit and absorbs in this way country-specific unobservable effects (Eller et al. 2005, p 26). Besides having country dummies, I will also include time-dummies to account for time-specific effects42. Hence, I use a
variable-intercept model with country-fixed and time-fixed effects.
Finally, it is likely that some of the explanatory variables are endogenous and that the sample exhibits country-specific and/or time-specific heteroskedasticity. Therefore I will run Granger causality tests and test for equality of variances in the residuals of the OLS estimation.
39 In Eviews, the econometrical package I use, the Sargan test is represented by the J-statistic distributed as
chi-square(k-p), where k is the number of estimated coefficients and p is the instrument rank
40 I test for second-order correlation because (negative) first-order serial correlation is likely to be present due to
the fact that GDPCi,t-1 is correlated with i,t by definition
41 The fixed effects model is more appropriate than the random effects model: if the individual effect represents
omitted variables, it is likely that these country specific characteristics are correlated with the other regressors violating the main assumption of the random effects model
42 One has to consider that panel data growth regressions based on annual frequency data are often determined
4. Results
This section presents the results of the regression specifications for the direct link between foreign bank activity and economic growth, specified in equation (1) and the link between foreign bank activity and economic growth volatility, specified in equation (2). The results are given in Appendix D.
4.1 Direct effect
The results for the direct link are given in Table D1. Three different systems of regression equations are estimated using a weighting matrix that is robust to heteroskedasticity, contemporaneous correlation and autocorrelation of unknown form. The results are given in columns43. In the first column, the system estimation results are reported using the net interest margin as the efficiency indicator. In the second and third column I estimate a similar system of equations replacing the net interest margin by overhead costs and profit before tax respectively. Both net interest margin and overhead cost have the expected positive sign, implying that higher levels of the net interest margin and overhead costs are associated with a lower level of economic growth due a lower level of efficiency or competition in the banking sector. However, the coefficient of bank’s profit before tax is significant, but has an unexpected negative sign44. Bayraktar and Wang (2006) found the same result for the sign of the coefficient of this variable and argued that if we take higher profits as an indicator of a higher level of financial activities, higher growth means higher level of financial activity and, in turn higher profits for banks. The main results appeared to be robust.
The most important result is the expected positive and statistically significant sign of the foreign bank asset share indicating a direct positive effect of foreign banking activity on economic growth. However, the sign of the coefficients for credit volume and the interaction term indicating the ‘direct effect’ are unexpectedly negative (only in case of the interaction term it is significant). An explanation for this result might be that though foreign banks bring in capital, they are also likely to crowd out domestic banks retarding economic growth. Moreover, foreign banks do bring in credit, but this credit is not available to everyone. For many individuals as well as small companies, it might even get more difficult to acquire credit because foreign banks require clients to fulfill certain requirements for obtaining credit (i.e. business plans and motivations/indications of the credit type needed). Especially, in developing countries the population has little experience with financial products and markets and hence cannot fulfill these requirements (Valev, 2006). Finally, the real sector may be
43 Note that I do not list the R²-value and the Durbin-Watson statistic because their usual interpretation does not
hold in dynamic models: Durbin-Watson is not valid when a lagged dependent is used as a regressor and the usual interpretation of R² does not hold because there is no constant included biasing it downwards (it could even become negative). Windmeijer (1995) has suggested to use the pseudo-R², the squared correlation between the predicted and the actual value of the dependent variable, as a measure of goodness of fit. However, Eviews does not provide an option to predict with a system of equations so I could not use this measure
deprived of credit needed for investment projects as foreign banks might be more eager to finance budget and current account deficits (often less risky) resulting in potentially less efficient allocation of this foreign credit (Ardiç and Damar, 2006). All in all, these results provide support for hypotheses 1 and 2.
The estimated coefficients for initial GDP per capita are negative and statistically significant. This indicates that countries which experienced higher rates of economic growth in the previous period, are likely to experience a bit lower growth in the next period. In the literature this is generally taken as evidence of conditional convergence. Trade openness and the indicator of public capital stock (main telephone lines per 1,000 people) have an expected significant positive sign as well as the inflation rate (except for one specification, but it is insignificant).
Political risk and the level of government expenditure to GDP have an unexpected positive sign, but only the latter was significant. The positive coefficient of government expenditure might be explained by arguing that government expenditure also includes ‘investments’ in education, health care and infrastructure which are undoubtedly associated with higher economic growth (Landau, 1983).
Finally, I remark that both test statistics support the model since we fail to reject the null hypothesis in each case. The Sargan test indicates that the instruments are not correlated with the error term and thus the moment conditions hold. The second-order serial correlation test shows that the error terms in the first-difference regression equation do not exhibit any second-order serial correlation (as expected in every specification I found significant negative first-order serial correlation).
4.2 Volatility effect
The results for the volatility effect are given in Table D2. Three different regressions are estimated. The results are given in columns. In the first column I show the regression results correcting for country-specific heterogeneity by dividing the country growth volatility by the population using this outcome as the dependent variable45. However, the high value of the Durbin-Watson statistic shows that autocorrelation is a serious problem and I have no options to correct for it, because using AR terms is not possible in the case of period fixed effects.
Since autocorrelation is a serious problem I decided to estimate a second model without using period fixed effects. Moreover, I include an AR1 term to deal with autocorrelation and use cross-section weights to deal with the country-specific hetero-skedasticity. The regression results of this specification are shown in the second column. It is immediately noticeable that this specification is an improvement compared to the one in the first column. The R² is higher, the Durbin-Watson statistic is lower and more of the
45 Generally, it is better to correct for country-specific heteroskedasticity by using cross-section weights but this
coefficients have their expected sign. In the third column I estimated the same regression but included an interaction between the foreign bank asset share and a developing country dummy to see if there are differences between developing and developed countries in terms of the volatility caused by foreign banking activity. Although its coefficient was negative, it was not significant.
The results from both models (with and without period fixed effects) are robust and do not provide support for hypothesis 3. In all specifications, the coefficient of the foreign bank asset share has the expected positive sign, but is insignificant. The same is true for banking sector concentration and financial sector development46. Moreover, political risk and trade openness are found to be positively associated with growth volatility, with expected significant coefficients. Besides the unexpected negative sign of the coefficient for money growth (which is insignificant), the sign of the coefficient for terms of trade shocks is also unexpectedly negative, but significant. However, this effect might be asymmetrical: an increase (negative shocks) might be associated with higher volatility and a decrease (positive shock) with less volatility, with the majority of the shocks in my sample being positive.
In addition, I checked the proposition by Morgan and Strahan (2003) that foreign bank activity may be endogenous: foreign banks are more likely to enter a country after a sharp downturn (when volatility is high) to buy up assets cheaply. To test for the endogeneity of this and other dependents I estimated pairwise Granger causality tests47. The results showed no evidence of endogeneity. To be sure, I also estimated the specification in the second column using lagged values of the regressors as instruments. Still, it did not change the main results. Finally, I checked whether using a random effects model would have been more appropriate by estimating a random effects regression and performing a Hausman-test. The test showed that is better to use a fixed effects model.
46 Adding a squared term of financial sector development, as suggested in the literature, did not change the
results (and was insignificant)
47 A time series X is said to Granger-cause Y if it can be shown, usually through a series of F-tests on lagged
