The effect of gross capital inflows on the central government debt-to-GDP ratio

(1)

The effect of gross capital inflows on the central government

debt-to-GDP ratio

Master thesis IE&B

Coen Dijkema – s2574616 c.e.dijkema@student.rug.nl

(2)

2

Abstract

The level of gross capital inflows in advanced economies has risen steeply since 2000. At the same time, central government debt-to-GDP ratios in these countries have been increasing. This thesis aims to establish whether gross capital inflows have a significant effect on this ratio, through their impact on real GDP growth, long-term interest rates, exchange rates, inflation, and the creation of banking crises. Additional analysis is done by disaggregating gross capital inflows into direct investment, portfolio investment, and other investment inflows, and by identifying stable, surge, and stop episodes. The results show that an increase in gross capital inflows result in a reduction of the central government debt-to-GDP ratio, but only during stable episodes. During surge episodes, only increases in portfolio investment inflows have a

significant effect on the reduction of government debt, but the vulnerability of the economy to shocks also increase. This vulnerability materializes during stop episodes, during which the reductions in portfolio and other investment inflows sharply increase the central government debt-to-GDP ratio, through macroeconomic effects and banking crises.

(3)

3

Introduction

Since the end of last century, there has been a massive increase in international capital flows, due to a rise in financial integration, which has led to global financial markets. Capital inflows can contribute to the financing of investment, but can also lead to financial instability and economic crises through growth of credit extension, exchange rate volatility, and an increase of the inflation rate. Several of these cases have occurred throughout history.

In the 1990s, Mexico and several other emerging economies in Latin America and South-East Asia experienced a huge increase in capital inflows. This led to a significant increase in bank lending, which mostly went to the real estate sector. Real estate prices increased, leading to the creation of real estate price bubbles. When the bubble burst and real estate prices fell, the banks were stuck with a large amount real estate loans that borrowers were unable to repay, which hurt the banking sector (Carstens & Schwartz, 1998). The problems in the banking sector then had to be solved by using public finances.

More recently, advanced economies have also experienced the dangerous effects that international capital flows can have on the economy. Capital inflows surged throughout the mid-2000s and was the greatest between advanced economies (OECD, 2011). The huge inflows of capital into advanced economies fueled a credit boom which led to a banking crisis after asset prices, mainly in real estate, fell sharply. As has been the case in other countries, the banking sector was left with many failing real estate backed loans. These are prime examples of the negative impact that capital flows can have on the economy.

(5)

5 -150,000.00 -100,000.00 -50,000.00 0.00 50,000.00 100,000.00 150,000.00 200,000.00

Average Direct Investment Inflows Eurozone Average Portfolio Investment Inflows Eurozone Average Other Investment Inflows Eurozone

Figure 1: Average gross capital inflows Eurozone and Non-Eurozone 1975-2012, mil USD

Source: Broner et al., (2013), IMF, and author’s calculations.

Figure 2: Average gross capital inflows Eurozone per category 1975-2012, mil USD

Sources: Broner et al., (2013), IMF, and author’s calculations.

-50,000.00 0.00 50,000.00 100,000.00 150,000.00 200,000.00 250,000.00 300,000.00 350,000.00 400,000.00

(6)

6 Figure 3: Average gross capital inflows Non-Eurozone per category 1975-2012, mil USD

Sources: Broner et al., (2013), IMF, and author’s calculations.

The increase of gross capital inflows has been accompanied by rising government debt levels across advanced economies. Figures 4 and 5 show the path of the central government debt-to-GDP ratio over the last 38 years for the group of Eurozone and non-Eurozone countries whose average gross capital inflows are illustrated above. Whereas in 1975 almost all Eurozone countries have an under 50 percent government debt-to-GDP ratio, in 2012 nearly all of them are above 50 percent, with the European periphery countries even above 100 percent. The non-Eurozone countries have been able to keep their government debt-to-GDP ratio much lower, with Japan as an exception. Nearly all of them are under the 50 percent level in 1975, and still are in 2012.

Figure 4: Eurozone countries central government debt-to-GDP ratio 1975-2012

Source: IMF Global Debt Database.

-150,000.00 -100,000.00 -50,000.00 0.00 50,000.00 100,000.00 150,000.00 200,000.00

Average Direct Investment Inflows Non Eurozone Average Portfolio Investment Inflows Non Eurozone Average Other Investment Inflows Non Eurozone

0% 50% 100% 150% 200%

Austria Belgium Finland France

Germany Greece Ireland Italy

(7)

7 0% 50% 100% 150% 200%

Australia Canada Chili

Czech Republic Denmark Iceland

Korea, Republic of New Zealand Norway

Singapore Sweden Switzerland

Turkey United Kingdom United States

Uruguay Japan

Figure 5: Non-Eurozone countries central government debt-to-GDP ratio 1975-2012, in %

Source: IMF Global Debt Database.

An abundance of papers on capital flows describe the effects on the economy and their probability of causing a financial crisis. This thesis extends the effects of capital inflows from the economy to the central government debt-to-GDP ratio, for which empirical evidence is lacking in the literature. For example, Reinhart and Rogoff (2008) do show a correlation between surge episodes and sovereign defaults when the surge comes to an end, but there is no empirical study. Only an illustration of the possible relation between capital inflows and an unsustainable rising government debt level is provided.

This leads us to the following research question: What is the effect of gross capital inflows on the central government debt-to-GDP ratio?

The aim of this thesis is to empirically establish whether any significant relation can be found between the two variables. Secondly, since its periods with extreme capital inflow volatility that are causing macroeconomic instability and financial vulnerabilities, the research question will be tested on different capital inflow episodes. The identified episodes are stable, surge, and stop episodes, and are based on the standard deviation above or below the countries’ gross capital inflows trend. Furthermore, a large difference in size and volatility between the types of gross capital inflows is shown in figures 2 and 3. Additionally, the riskiness to the economy is different for each type of inflow. This leads to the question whether the effect of gross capital inflows on the central government debt-to-GDP ratio differs between capital inflow types.

(8)

8 Figure 6: Variables for the empirical analysis

The results are as follows. During stable episodes of gross capital inflows, an increase in total gross capital inflows significantly reduces the central government debt-to-GDP ratio. After disaggregating the gross capital inflows, it is shown that only additional portfolio and other investment inflows contribute to this reduction, while no significant effect is found for direct investment inflows. During surge episodes, additional increases in total gross capital inflows do not significantly reduce debt ratios, and after disaggregating, only portfolio investment inflows show a significant effect at the 10 percent level. Stop episodes do have significantly negative effect on the central government debt-to-GDP ratio. The decreases in total gross capital inflows increase government debt. This effect is also found significantly for additional portfolio and other investment inflow reductions. Therefore, the government must be aware of the level of gross capital inflows, not only because of the consequences to the economy, but also for the consequences on their own finances. Specifically the levels of gross portfolio and other investment inflows require careful attention.

The structure of the thesis will be as following: section 2 includes the related literature and hypotheses. Section 3 introduces the methodology. Section 4 reveals the empirical results and provides limitations to these results. Section 5 concludes the thesis.

2. Literature Review

Firstly, the explanation for using gross capital inflows instead of net capital inflows will be given. Secondly, the justification for not including gross capital outflows in this thesis is

Independent variable:

Gross capital inflows Disaggregated into: 1. Direct investment 2. Portfolio investment 3. Other investment Measured as a % of GDP Control variables: - Real GDP growth (%)

- Long term real interest rates (%)

- Government primary balance (% of GDP) - Fiscal costs of banking crises (% of GDP)

- Real effective exchange rate

Dependent variable:

Central government debt-to-GDP ratio (%)

(9)

9

explained. Thirdly, determinants of government debt are explained. Lastly, we arrive at the hypotheses and their motivation through related literature.

2.1 Gross capital outflows and government debt

Besides gross capital inflows, we need to consider a potential effect of gross capital outflows on government debt. A massive outflow of capital by domestic investors is defined as capital flight. Capital flight occurs whenever investors have a fear of losing. One of the causes of capital flight is an unstable government in terms of its rising government debt and government deficit. An example is given by Kaminsky (1998), where an increasing government deficit caused Mexican investors to fear about the ability of their government to repay their debt, and capital outflows by domestic residents sharply increased. That capital flight happens when uncertainties regarding the government deficit are high is also found by Hermes and Lensink (2001).

Secondly, a large foreign debt stock can cause capital outflows when doubts about debt unsustainability are raised. Therefore, besides a high government deficit, a high debt level can stimulate capital flight by worsening the macroeconomic environment and increasing the probability of a debt crisis. A cross-sectional study by Collier et al., (2001) also reported a statistically significant positive impact of the total debt stock on capital flight.

Consequently, domestic investors would want to make overseas investments as government debt-to-GDP is increasing. Hence, an increase in the central government debt-debt-to-GDP ratio is suggested to have a positive effect on capital outflows, leading to reverse causality problems, where an increase in government debt explains an increase in gross capital outflows. Therefore it is decided to not include gross capital outflows in this thesis and solely focus on gross capital inflows.

2.2 Gross versus net capital inflows

(10)

10 Figure 7: Net versus gross capital inflows, 1990-1995 versus 2005-2009, in mil USD

Sources: IMF and author’s calculations.

Furthermore, using net capital flows compared to gross capital flows can lead to a wrong identification of a capital flow episode (Schmidt and Zwick, 2015). For example, net capital inflows can display a surge, while using gross capital flows could show a retrenchment (a sharp decrease in capital outflows). Figure 1 has an example of such a case in the year 2008. Relative to 2007, there is a large increase in net capital inflows, which would be identified as a surge, but gross capital inflows actually barely increase. The increase in net inflows is caused by a retrenchment in gross capital outflows. From 2005 to 2006, one would identify a stop episode when looking a net capital inflows, while gross capital inflows only declined slightly, and it should be identified that there is a capital flight (sharp increase in capital outflows). The use of gross capital flows instead of net capital flows is therefore important, to identify the correct episode and showing the larger size and volatility.

Besides correctly identifying the capital flow episodes, it is important to be able to differentiate between the behavior of domestic and foreign investors. In the case of a sudden stop episode, most important is the loss of access of countries to international financial markets. The sudden stop should therefore be measured only through the behavior of foreign investors, and not include domestic investors. (Crystallin, et al., 2015). Only by using gross capital flows, it is possible to differentiate between foreign and domestic investors

These arguments provide the motivation to use gross capital inflows instead of net capital inflows in this thesis.

2.3 Determinants of government debt

The main determinants of the central government debt-to-GDP ratio will now be described. The most obvious determinant is GDP growth. By using a panel data regression with the effect of real

(11)

11

GDP growth on the government debt-to-GDP ratio, Sinha et al., (2011) find a significant negative effect. When the real GDP growth rate increases, the total GDP level and the government’s tax revenues increase, and social welfare expenses decrease. This implies that the total debt gets divided by a larger number, which decreases the debt-to-GDP ratio. Secondly, depending on the government’s fiscal and tax policy, increased revenues and decreased expenditures can lead to a lower government deficit or even a surplus. Middle/high income countries are found to often have an acyclical fiscal policy (Kaminsky et al., 2004). This implies that spending and tax rates remain relatively constant during good and bad times (based on the rate of GDP growth). In good times, an increase in tax revenue therefore will not be used to increase spending, but to reduce government debts. (Holtfrerich, et al., 2016). The difference between the government’s revenues and spending is shown in the government primary balance. The World Bank, (2005) concludes that changes in the government's primary balance affect the government debt-to-GDP ratio. When the primary deficit becomes larger, the government has to accumulate more debt. Either by the issuance of government bonds or by lending directly from financial institutions. When the primary surplus increases, debt can be reduced by a larger amount.

Another variable affecting the government debt-to-GDP ratio is the inflation rate. An increase in the inflation rate reduces the real debt burden of the government (Calvo and Guidotti, 1992). GDP and tax revenues increase, while interest payments stay the same. The interest payments of the government therefore require a smaller portion of total tax revenues. A similar story as with GDP growth arises: the budget deficit decreases or the surplus increase. Therefore, governments have the incentive to increase inflation to decrease their debt burden. Although the effect of inflation on the reduction of the debt value does depend on the average maturity of its debt and the amount of debt that is indexed to inflation. Worst case scenario, when the maturity of the debt is zero, and all debt is indexed to inflation, the government will not be able to use an increase in inflation to reduce its debt burden (Aizenman and Marion, 2011).

Furthermore, the long-term real interest rate has a significant impact on the government debt-to-GDP ratio. This is found in a 15-country case study by The World Bank (2005). Higher interest rates will increase the interest payments on the current government debt and increases the borrowing costs for the government. The opposite occurs for a reduction in interest rates, which is favorable for the government debt level.

Lastly, depending on the share of public debt in a foreign currency, a change in the exchange rate can have a positive or negative effect on the debt-to-GDP ratio. An appreciation of the currency leads to a decrease in the real debt burden, while a devaluation of the currency leads to an increase in the real debt burden, which in turn has a negative effect on the debt to GDP ratio.

(12)

12

debt. The surge in government debt ratios during a banking crisis is explained by the fiscal costs of bank bailouts. For Iceland and Ireland for example, the cost of government intervention in the banking sector during the global financial crisis in 2008 amounted to nearly 40 percent of GDP, and public debt increased by more than 60 and 80 percent of GDP respectively over the next five years (Laeven and Valencia, 2018). Besides the fiscal costs that result from direct governmental interventions in the banking sector, there are indirect fiscal costs of banking crises. These costs are incurred through the impact of banking crises on the real economy, in particular on interest rates, GDP growth, and asset prices. Leading to reduced government revenue and an increase in the cost of borrowing. Consequently, government debt ratios increase (Amaglobeli et al., 2015).

2.4 Hypotheses and motivation

Below are the hypotheses regarding the effect of total gross capital inflows, and the disaggregated gross capital inflows, on the central government debt-to-GDP ratio. The focus is on stable, surge, and stop episodes.

Hypothesis 1: During stable episodes, an increase in gross capital inflows has a significant

negative effect on the central government debt-to-GDP ratio. Meaning an increase in total gross capital inflows reduces the central government debt-to-GDP ratio.

Overall, capital inflows are found to stimulate long-term growth (OECD, 2011). This works through several channels. Firstly, capital inflows can provide the additional finance that is needed to make investments. Without capital inflows these additional investments would be unable to take place (Bosworth and Collins, 1999). Furthermore, capital inflows can increase productivity through the inflow of new technologies and management techniques (IMF, 2018).

The amount of stimulation to investment and productivity by capital inflows does depend on several factors, such as the level of human capital and the amount of corruption. Furthermore, countries with a higher developed financial sector and better capable to reap the benefits of capital inflows (Bailliu, 2000). Advanced economies will have the most favorable outcomes regarding these factors. Therefore an increase in the GDP growth caused by direct investment is expected. In terms of the effects of capital inflows on other macroeconomic variables, Ghosh et al., (2016) find an increase of capital inflows by 1 percent to be associated with an increase of 1.4 percent of the real effective exchange rate and a 0.8 percent increase in the output gap, which is associated with inflationary pressures. Therefore the impact of gross capital inflows on the growth of GDP, exchange rate, and inflation during stable episodes lead to believe that additional inflows reduce the central government debt-to-GDP.

Hypothesis 1.1: During stable episodes, increases in gross direct investment, portfolio investment

(13)

debt-13

to-GDP ratio. Meaning an increase in either of these inflows reduces the central government debt-to-GDP ratio.

Definitions for each gross capital inflow category comes from the IMF’s Balance of Payments and International Investment Position Manual Sixth Edition (BPM 6).

Direct investment inflows are defined as investments where a foreign investor obtains control with a significant degree of influence on the management of a company. The motivation behind direct investment inflows, having control or influence on the management, leads to different behavior compared to the other categories of capital inflows. Direct investment inflows are associated with a longer term investment.

The literature has found a positive effect of direct investment inflows on GDP growth. Choong et al., (2010) finds a positive effect on economic growth by using data from both advanced and developing countries over the years 1988 to 2002. Using data on 100 countries over the period 1990 to 2010, Aizenman et al., (2013) also find a strong positive relation between FDI inflows and economic growth. Furthermore, Ghosh et el., (2016) find that a 1 percent increase in direct invest inflows leads to a 0.2 percent increase in real GDP growth, and a 0.25 percent increase in the output gap. The channels through which economic growth occurs are the availability of more technologies and an increase in productivity of capital and labor, and the encouragement of competition which increases performance and levels of human capital (Lipsey, 2006). An increase in productivity caused by direct invest inflows has also been found by Barrell and Pain (1997). In their study on direct investment inflows and productivity growth in the manufacturing sector, a 1 percent increase in direct investment inflows was found to increase productivity by 0.27 percent in Germany and 0.26 percent in the United Kingdom. In total, direct investment inflows contributed to near a third of the total increase in productivity of the United Kingdom manufacturing sector since 1985. The positive contributions of direct investment inflows on economic growth do depend on the development of the financial system, but this is not problematic for advanced economies, who have the highest developed financial systems, with the existence of well-developed banking institutions, insurance companies, and pension, insurance, and hedge funds (Giovannini et al., 2013)

Not all literature on the contributions of direct investment inflows on growth, productivity, and investment come to positive conclusions. Bosworth and Collins (1999), find that a dollar of direct investment inflows only leads to $0.33 additional investment, suggesting that direct investment inflows simply substitute for domestic capital for the most part. Concerning increases in productivity, Haddad and Harrison (1993), find little evidence that direct investment itself increases productivity. Furthermore, a large portion of direct investment inflows in advanced economies consists of mergers and acquisitions (World Bank Group, 2002). There is no evidence of M&A’s in OECD countries leading to economic growth (Doytch and Cakan, 2011).

(14)

14

Portfolio investment inflows are defined as cross-border transactions and positions involving debt or equity securities, other than those included in direct investment. These include stocks, bonds, mutual funds, and exchange-traded funds. Together with direct investment inflows, portfolio investment inflows is a way for foreign investors to partake in a foreign countries’ economy. The difference with direct investment inflows is that portfolio investment inflows are not associated with obtaining any control. It rather consists of acquiring a passive ownership.

Benefits of portfolio investment inflows to the economy are found by Evans (2002). These are an increase of liquidity, discipline and expertise being brought into the domestic capital markets. Furthermore, by using annual data on portfolio investment inflows for 13 central European countries, it is found that the equity component of portfolio investment inflows positively impacts GDP growth, but no significant effect is found for the debt instruments (Albulescu, 2015). Ghosh et al., (2016) also find a positive effect of portfolio investment inflows on GDP growth at the 10 percent significance level. Specifically, a 1 percent increase in portfolio investment inflows contributes to a 0.07 percent increase in real GDP growth. Additionally, they find a 0.29 percent increase in real effective exchange rate appreciation and a 0.2 percent increase in the output gap. Moreover, portfolio investment inflows contribute to the development of the financial sector. Ahmed and Ansari (1998) find, by using two measures of financial sector development, a positive relation between the amount of portfolio inflows and the development of the financial sector, which stimulates economic growth.

Altogether, positive effects on the GDP growth, exchange rate, and inflation give reason to expect that increases in portfolio investment inflows reduce the central government debt-to-GDP ratio during stable episodes.

(15)

15

Based on these findings, expected is that an increase in other investment inflows reduce the central government debt-to-GDP ratio.

Hypothesis 2: During surge episodes, an increase in gross capital inflows has a significant

negative effect on the central government debt-to-GDP ratio. Although this effect is much lower than during stable episodes. Meaning an increase in gross capital inflows during surge episodes decrease the central government debt-to-GDP ratio. While a decrease worsens the ratio.

Several studies have examined the characteristics, determinants, and consequences of capital inflow surges. Ghosh et al., (2014) find that capital inflow surges can have large macroeconomic and financial consequences to an economy.

A high level of gross capital inflows means there is a very high demand by foreign investors for the domestic currency, leading to an appreciation of the exchange rate. This can create competitiveness issues for the export sector and can hurt economic growth, depending on the importance of this sector. Furthermore, a large increase in investment, consumption, and cheap loans by banks lead to an overheating of the economy, which results in high inflation (Ghosh et al., 2014). Calvo et al., (1996) and Ghosh et al., (2016), find that surge episodes are associated with a higher GDP growth rate. For a 10 percent increase in gross capital inflows, the GDP increases by approximately 2.5-4 percent. Furthermore, overheating of the economy results in a 2 percent increase in the output gap and a 2 percent increase of the real exchange rate. The question is however, what if gross capital inflows are already at a level of 50 percent of GDP, does an additional 10 percent increase of gross capital inflows to GDP result in another 2.5 to 4 percent increase in GDP? This aspect is not touched upon by Ghosh et al., (2016), but the answer would most definitely be negative. Otherwise the data would have shown extremely high growth rates during surge episodes. Furthermore, gross capital inflows do need to contribute to the financing of productive investments in order to contribute to an increase in economic growth (Bailli, 2000). Regarding long-term interest rates, in both the US and the Euro area, surges in gross capital inflows have contributed to very low levels of long-term interest rates since the mid-2000s (Carvalho & Fidora, 2015). Lower interest rates also stimulate investment, output and consumption and decreases savings during a surge period (Gourinchas et al., 2017).

(16)

16 Hypothesis 2.1: During a surge episode, an increase in direct investment inflows does not have a

significant effect on the central government debt-to-GDP ratio. For portfolio and other investment inflows, an increase does have a significant negative effect on the debt ratio. Although those effect are lower than during stable episodes. Meaning an increase in portfolio and other investment inflows reduce the central government debt-to-GDP ratio.

Similar to the argumentation for total gross capital inflows, the growth effects of additional direct investment are expected to stagnate during surge episodes. Cheap direct investment inflows can simply become substitutes for domestic capital and can finance investment that would have taken place anyway. Furthermore, the increasing amount of inflows leads to a high amount flowing towards low productivity projects (Cecchetti and Kharroubi, 2015).

This leads to the expectation that the effect of an increase of direct investment inflows on the central government debt-to-GDP ratio becomes insignificant during surge episodes.

Episodes of high portfolio investment inflows result in an increase in asset prices, credit growth, and a higher exchange rate. Furthermore, the positive impact on output growth can lead to inflationary pressures (ECB, 2011). Additionally, Ghosh et al., (2016), find that portfolio investment inflows, at the 10 percent level, associated with exchange rate appreciation, inflation, and real GDP growth. The effect on GDP growth is diminishing, but through exchange rate appreciation and growth of the inflation rate it is expected that an increase in portfolio investment inflows contribute to a reduction in the central government debt-to-GDP ratio.

(17)

17

stop episode which occurs after the surge episode. Therefore the increased vulnerabilities of the financial system do not lead to any costs for the government during surge episodes. These costs are incurred afterwards, during the stop episode.

While fragility in the banking sector increases, the effect of increased amounts of extended credit to the economy do not provide economic growth during surge episodes. The reason is that credit is increasingly going to real estate assets (Bezemer et al., 2016). While the effect on economic growth might disappear, other investment inflows have an extremely high volatility in the short term and do affect inflation, exchange rate, and interest rate volatility during surge episodes. Therefore an increase in other investment inflows during surge episodes is still expected to decrease the central government debt-to-GDP ratio.

Hypothesis 3: During a sudden stop episode, the large decrease in gross capital inflows has a

positive effect on the central government debt-to-GDP ratio. Meaning reductions in gross capital inflows increase the central government debt-to-GDP ratio.

If the economy cannot adjust to a higher level of capital inflows, a slow down or stop will follow. Hence, capital inflow surges are almost always followed by a reversal or sudden stop of capital inflows. Empirically established by Sula (2010), is that the probability of a sudden stop is positively correlated with the amount of capital inflows in previous years. This is also illustrated in figure 1, where gross capital inflows often fall abruptly after they reach their peak. The fall of gross capital inflows during a sudden stop episode has become larger due to the continuing globalization of financial markets, and the effect on the economy have become more disruptive. There are several papers that have explored the macroeconomic consequences of a sudden reversal in foreign capital inflows, such as Gourinchas et al., (2017), Eichengreen and Gupta (2016) and Neagu and Mihai, (2013). In contrast with a surge episode, stop episodes go hand in hand with a large decline in aggregate demand, also by foreign investors, resulting in a depreciation of the currency, deflation, a reduction in output and a reduction in employment (Calvo and Reinhart). The reduction of foreign capital available increases the cost of funding for banks, which is transmitted to the economy through higher interest rates. Higher interest rates increase the rate at which consumers and firms can borrow, causing investment, consumption and output to decline, and the economy to contract (Gourinchas et al., 2017). The contraction of the economy is not offset by the stimulation of exports due to the currency devaluation (Neagu and Mihai, 2013). Eichengreen and Gupta, (2016), find that GDP growth falls by roughly 4 percent within the first year after a sudden stop of capital inflows.

(18)

18

there is a reduction in the access of banks to foreign funds, which leads to an increase in funding costs of banks (Gourinchas et al., 2017). Demand for loans decrease, resulting in a lower demand for assets such as real estate. The collapse in the real estate market hits the banks, since people borrowed money with the expectance that real estate prices kept rising. Banks often used assets such as equity or real estate as a collateral against loans. This collateral is not worth enough anymore for the bank to recoup all of the losses when the borrower defaults. This worsens the problems in the banking sector (Reinhart and Calvo, 2000). When investors become aware of this, the credit risk of the bank increases because their risk of defaulting increases. Resulting in more difficulty for the bank to obtain funds on the interbank market, as other banks will also have trust issues. Simultaneously, there is an increase in withdrawals of deposits and external funds. Liquidity problems are quickly present due to the high leverage of banks, with little amount of their own equity to withstand these shocks. In a stop episode, the problems of banks are transferred to the government. Reinhart and Rogoff (2008), find that when capital inflows crash after a surge episode, the government will end up having to incur the fiscal costs of a banking crises. These costs can be so large that the government has to default on its debt. Appendices 1 and 2 with the banking crisis costs for the government, and the surge and stop episodes, indeed show that banking crises occur in the same years as when stop episodes are identified.

This implies the following for the central government debt-to-GDP ratio during stop episodes: The economy contracts, GDP growth stagnates and could even become negative. In this case, the GDP level decreases, tax revenues decline, and its deficit increases or surplus decreases. Higher interest rates increase payments on its debt, devaluation of the currency and deflation increase debt burden of the central government. The banking crises that occur during sudden stop episodes can lead to massive bailout costs for the government. Therefore the expectation is that a decrease in gross capital inflows during sudden stop episodes increase the central government debt-to-GDP ratio.

Hypothesis 3.1: During a sudden stop episode, declines in direct investment inflows do not have

a significant impact on the central government debt-to-GDP ratio. Declines in portfolio investment and other investment inflows do have a significant positive effect on the central government debt-to-GDP ratio. Meaning the decline in portfolio and other investment inflows increases the central government debt-to-GDP ratio.

(19)

19

episodes. Furthermore, the type of investments categorized as direct investment inflows can involve large assets which are not easily sold and cannot leave the country so quickly compared to the other capital inflow categories. Due to these characteristics, direct investment inflows are not associated with risks of causing financial crises during stop episodes (Frankel and Rose, 1996) (Fernández-Arias and Hausmann 2001).

The inflows that do have high volatility in the short run are portfolio and other investment inflows. Which means there is a possibility that foreign investors may suddenly decide to leave the country they are investing in. This is dangerous in terms of exchange rate and interest rate volatility (Agarwal, 1997). Furthermore the stop of foreign financing decreases asset prices and output. The effect of the decline in other investment inflows on the central government debt-to-GDP ratio has been described in the previous hypothesis. Other investment inflows have a very high volatility. This makes them being related to financial crises (Ghosh and Qureshi, 2016). Especially the debt-driven banking inflows significantly increase the probability of a banking crisis (Furceri et al., 2012). The increased financial vulnerability during the surge episode bursts during the sudden stop episode. Leading to banking crises which the government needs to solve by using public debt. Furthermore the reduction of bank lending and rising interest rates increase the central government debt-to-GDP ratio. When the decrease in other investment inflows cause a banking crises, there are many adverse effects to the economy as well, the described indirect effects by (Laeven and Valencia, 2018), are an increase in interest rates, lower GDP growth, and falling asset prices. Therefore, a decrease in other investment inflows and portfolio investment inflows during stop episodes are expected to significantly increase the central government debt-to-GDP level.

3. Methodology

3.1 Country sample and timeline

The country sample choice and timeline depend on data availability. This criterion drives the thesis in the direction of using high-income economies, as classified by The World Bank as countries having a GNI per capita of at least 12.055 US$. High-income economies have a much wider data availability. The selection of countries chosen for this thesis consists of 11 Eurozone countries and 17 non-Eurozone high-income economies. The chosen timeline is 1975-2012. To be able to capture enough surge and stop episodes, but also stable episodes with only moderate levels and volatility of gross capital inflows.

3.2 Data description

3.2.1: Government debt

(20)

20

a wider time span, since countries can measure their government debt in several different ways. Some countries only report the central government debt, while other countries measure general government debt (includes debt from states and municipalities). The most widely available database on government debt-to-GDP ratios has been found to be the IMF Global Debt Database. Four indicators for government debt have been produced: Central Government Debt, General Government Debt, Nonfinancial Public Sector Debt, and Public Sector Debt. The chosen indicator for government debt in this thesis is Central Government Debt, which is defined as the amount of debt liabilities issued by the central government as a share of GDP. Most data is available on this specific indicator of government debt. Taking the government debt as a ratio to GDP, compared to the absolute level of debt, makes it easier to compare debt levels between countries, and shows the ability of these countries to pay off their debt.

3.2.2: Calculating gross capital inflows

There are two methods of calculating gross capital flows in the literature. Most widely used in the literature (Ito, 1999), (Prasad and Wei, 2005), (Rothenberg and Warnock, 2006), (Tille and van Wincoop, 2008), (Broner et al., 2013), is to accumulate the changes in net incurrence of liabilities in direct investment, portfolio investment, and other investment by foreign investors (CIF) of a country. This will give the amount of gross capital inflows:

Gross capital inflows = CIF (capital inflows by foreign agents) = change in net incurrence of liabilities of direct investment + change in net incurrence of liabilities of portfolio investment + change in net incurrence of liabilities of other investment.

One will immediately notice that this way of calculating gross capital flows does involve netting. The IMF records types of financial flows as a gross measure, such that assets are recorded under assets and liabilities are recorded under liabilities. The net recording of the IMF refers to the netting of debit entries of certain asset or liability types against the credit entries of the same asset or liability type. For example a bank can acquire an amount of foreign currency, meanwhile another bank sells an equal amount of foreign currency. According to the IMF, no gross inflow of foreign currency would then be registered.

Even though net recording is used in the financial account, the literature has implemented the above equations of gross capital inflows and outflows as their standard measure of calculating gross capital flows. The data simply does not yet allow for a true recording of gross capital flows.

(21)

21

changes are not subtracted from capital outflows but are added to capital inflows. This leads to a higher number of gross capital flows and removes any negative flows, but does not yet solve the problem of calculating gross capital flows with netted capital flows. Figure 8 shows the implementation of the new measure from Janus and Riera-Crichton compared to the standard measure of calculating gross capital inflows and outflows flows from the IMF, for Greece from the years 2009 to 2012.

Figure 8: Standard measure versus new measure of calculating gross capital inflows 2009-2012, Greece, millions of USD

Source: IMF and author’s calculations.

-40,000.0 -20,000.0 0.0 20,000.0 40,000.0 60,000.0 80,000.0 100,000.0 120,000.0 140,000.0 160,000.0 2009 2010 2011 2012

Greece inflows (standard measure) Greece inflows (new measure)

-40,000.0 -20,000.0 0.0 20,000.0 40,000.0 60,000.0 80,000.0 100,000.0 120,000.0 140,000.0 160,000.0 2009 2010 2011 2012

(22)

22

This thesis follows the standard approach of calculating gross capital flows and extends a database created by Broner et al., (2013) to the year 2012. Their database is created by using data from the IMF. The IMF has a large and extensive database on international capital flows called the IMF Balance of Payments and International Investments Position Statistics. This database provides the net incurrence of liabilities and net acquisition of financial assets for direct investment, portfolio flows and other investment. Therefore the data provided allows perfect use of the standard approach of calculating gross capital flows.

3.2.3: Control variables

To complete the data selection for the empirical research, variables are added which are affected by gross capital inflows, and have an impact on the government debt-to-GDP ratio. These are based on findings in the literature on the determinants of the government debt-to-GDP ratio (see literature review). The following variables are selected:

- Real GDP growth rate (%) - Long-term real interest rate (%)

- Government primary balance (% of GDP) - Fiscal costs of banking crisis (% of GDP - Real effective exchange rate

Data on the macroeconomic variables real GDP growth rate, long-term interest rate and the government primary balance are made available by the IMF. Data on the real effective exchange rate is collected from a database by Darvas (2012). It measures the value of a country’s currency against a basket of its trading partners. Their data covers the whole time period and is available for all countries. This is advantageous since the exchange rate for all countries will be calculated with the same methods.

One can see that inflation is not included in this set of variables, while the literature did find that inflation affects the government debt-to-GDP ratio. The reason is that the real GDP growth rate, long-term real interest rate and real effective exchange rate are already adjusted for inflation. Adding inflation as a control variable leads to a multicollinearity problem between these variables. Implying that the independent variables are correlated with each other. This is a problem because the independent variables should be independent of each other. A VIF test on multicollinearity indeed shows that this problem arises after including inflation as a control variable. Removing inflation removes the multicollinearity problem.

(23)

23

among the years in which the banking crisis took place. To give an example, if a country experiences a banking crisis in 2008 and 2009, in which the government had to incur a total fiscal costs of 5% of GDP, these costs will be assigned as 2.5% in 2008 and 2.5% in 2009.

3.3 Identifying stable, surge, and stop episodes

After concluding that gross capital inflows can best be used to correctly identify the extreme capital flow episodes of our interest (surge and stop), the correct identification method needs to be found. A review of recent studies on capital flow surges by Crystallin, et al., (2015) shows that there are several ways to identify surge and stop episodes. The most widely used technique is using the Hodrick-Prescott filter to obtain the trend of gross capital inflows. What differs a lot between the literature is the chosen threshold to qualify capital flows as a surge or stop, and the chosen value of the smoothing parameter. For example, to identify capital inflows as a surge, the inflows must be higher than 4% of GDP and be at least one standard deviation above the trend (Sula, 2010). For Forbes and Warnock (2012), gross capital inflows must be at least one standard deviation above the 5-year rolling average. Similarly, a sudden stop can be calculated by a number of standard deviations below the trend of gross capital inflows, such as in Eichengreen and Gupta (2016), who classify their sudden stop episodes as periods where the inflow of total capital is one standard deviation below the average inflows of the last 20 quarters.

A series of tests is done by Crystallin, et al., (2015) with regards to the ability to detect capital flow surges with different threshold levels of the standard deviation. When a standard deviation of one above the trend is chosen as a threshold, the capital flow surge episodes of the 1990s are completely missed. When reducing the standard deviation threshold to 0.67, these surge episodes do get captured. Since this thesis has data starting from 1975, the threshold chosen to identify surge and stop episodes is 0.67 standard deviations above or below the HP trend, to be able to capture surge and stop episodes before the 2000s.

The smoothing parameter λ for annual data is recommended by Hodrick & Prescott (1997), to be 100. In a later research, Ravn and Uhlig (2002) determine that that value is incorrect. They argue that the best value of the smoothing parameter for annual data is equal to 6.25. A paper regarding capital flow surges by Cardarelli et al., (2010) uses a value of 1000 for the smoothing parameter. As this thesis also involves capital flows, the chosen method is to follow the paper by Cardarelli, et al., (2010) and use a value of 1000 for the smoothing parameter.

To summarize: The Hodrick-Prescott filter with a smoothness parameter of 1000 will be applied to estimate the trend of the gross capital inflows for all countries from 1975 to 2012. Stable episodes and high volatile episodes are identified as following:

Stable: Gross capital inflows > HP trend - 0.67*

σ

and < HP trend + 0.67*

σ

Surge: Gross capital inflows >HP trend + 0.67*

σ

(24)

24

Figure 9 is an illustration of the identification method for Australia in the upper figure and Sweden in the bottom figure. Setting the threshold at 0.67 standard deviations indeed makes it possible to capture relatively smaller sized surge and stop episodes, compared to those after the 2000s, leading to a higher amount of observations.

Figure 9: Illustration of surge and stop episode identification

Source: Author’s calculations

With this measure, 149 Surge years and 131 stop years have been found. The remaining 713 years are assigned to a ‘stable’ episode. Appendix 2 illustrates all the years for all countries in which surge and stop episodes took place.

-5.00% 0.00% 5.00% 10.00% 15.00% 20.00%

Gross Capital inflows (% of GDP) HP trend

0.67 Std above trend 0.67 Std below trend

-5.00% 0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00%

Gross Capital inflows (% of GDP) HP trend

(25)

25

Since extreme episodes in capital flows often occur simultaneously for the disaggregated flows as well (Furceri et al., 2012), the assumption is made in this thesis that when total gross capital flows identify a surge, the surge consist of all three gross capital flows, direct investment inflows, portfolio investment inflows, other investment inflows. The same is assumed for sudden stops.

3.4 Regression technique

The relation between gross capital inflows and the central government debt-to-GDP ratio that is tried to be established during stable, surge, and stop episodes represents an indirect effect in which gross capital inflows affects macroeconomic variables, which affect the central government debt-to-GDP ratio. This has the following structure:

Control

variables (C)

Independent Real GDP growth

variable (X) Long-term interest rate Dependent variable (Y)

Gross capital inflows Real effective exchange rate Debt-to-GDP ratio (and its components) Government primary balance

Banking crises cost

A method for calculating the indirect effect with regression analysis is proposed by Judd and Kenny (1981). To do so, two regressions are required:

Model 1: 𝛾 = 𝛽0 + 𝛽𝑋 + 𝜀

Model 2: 𝛾 = 𝛽0 + 𝛽1𝑋 + 𝛽2𝐶 + 𝜀

The first model regresses only the independent variable on the dependent variable. In the second model the control variables are added. The indirect effect of gross capital inflows on the central government debt-to-GDP ratio is found by subtracting the coefficients β and β1:

βindirect = β – β1 .

The coefficient represents the change in the central government debt-to-GDP ratio for every one percentage change in gross capital inflows through its effects on real GDP growth, government primary balance, long-term real interest rates, real effective exchange rate, and banking crises costs. For every episode (stable, stop, and surge), this model is calculated two times: for the indirect effect of an increase in total gross capital inflows on the central government debt-to-GDP ratio, and for the indirect effects of increases in direct investment, portfolio investment, and other investment inflows on the central government debt-to-GDP ratio.

(26)

26

A Hausman test provided the answer whether it is more appropriate the use a random effects or a fixed effects model. The null hypothesis was rejected at the ten percent level which indicates that the fixed effects model is more appropriate. All regressions will include country and year fixed effects. This allows for unobserved heterogeneity between countries and over time to be controlled for, and allows different slopes and intercepts between countries. To make it easier to interpret the regression outcomes, the gross capital inflows and its disaggregated components have been calculated as a percentage of GDP, since the rest of the data is also shown as percentages. Lastly, heteroskedasticity was found to be present and will be corrected for by adding robust standard errors.

4. Empirical results and limitations

4.1 Empirical results

Firstly, the summary statistics can be found in table 1 below. Of main interest are the variables ‘Central Government Debt-to-GDP ratio’, ‘ Total Gross Capital Inflows’, ‘Direct Investment Inflows’, ‘Portfolio Investment Inflows’, and ‘Other Investment Inflows’. All capital inflows are measured as a percentage of GDP. These data are nearly complete for every country from 1975-2012, resulting in over a 1000 observations. The only variable that has been difficult to acquire is the long-term real interest rate, which leads to a reduction of nearly 300 observations when being included in the regression analysis.

Table 1: Summary statistics

Table 3 shows the first results. During a stable episode, the expectation is a negative effect of an increase in gross capital inflows and all disaggregated components on the central government debt-to-GDP ratio. Columns one and two show the effect of total gross capital inflows. The regression output indeed shows a significant negative effect of a one percent increase in gross capital inflows

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

VARIABLES N mean sd min max

Central government debt-to-GDP ratio 1,031 46.11 29.12 2.741 187.1 Total gross capital inflows 1,007 11.20 20.31 -38.92 192.1 Direct investment inflows 1,008 2.911 5.604 -5.671 87.44 Portfolio investment inflows 1,008 3.690 10.05 -23.86 108.3 Other investment inflows 1,014 4.557 11.72 -48.88 157.0 Real GDP growth rate 1,041 2.817 3.108 -13.38 14.78 Government primary balance 1,026 1.007 4.192 -28.17 20.57 Long-term real interest rate

Real effective exchange rate

(27)

27

on the central government debt-to-GDP ratio by 0.215 percent. This is the indirect effect after subtracting the coefficients of gross capital inflows in columns one and two.

After disaggregating the gross capital inflows, surprisingly, the effect of direct investment inflows on the central government debt-to-GDP ratio is negative, but not significant. During stable episodes, with relatively low amounts of direct investment inflows, increases in direct investment inflows should find profitable investments and contribute to economic growth. By this way a reduction of the central government debt-to-GDP ratio is expected. Perhaps the literature by the World Bank Group (2002), and Haddad and Harrison (1993) were correct about their findings that direct investment inflows do not contribute to productivity and domestic investment as much, and too much of the direct investment inflows in advanced economies includes mergers and acquisitions, which does not lead to additional economy growth. The argument of inflows of new technology would also be less growth enhancing in advanced economies, as these countries already possess high levels of technology. Therefore, there are several arguments that can justify the insignificant effect of direct investment inflows.

Other investment inflows and portfolio investment inflows do contribute significantly to a lower government debt ratio, as expected. Either through their effect on increasing GDP growth, exchange rate appreciation, or higher inflation.

Table 3: Effect of total and disaggregated gross capital inflows on central government debt-to-GDP - stable episode

(1) (2) (3) (4)

VARIABLES Overall Overall with controls

Disaggregated Disaggregated with controls Gross Capital inflows -0.534*** -0.319***

(0.124) (0.101)

Real GDP growth rate -1.259*** -1.267*** (0.375) (0.377) Long-term real interest rate 1.844*** 1.825***

(0.280) (0.270) Government primary balance -1.031 -1.035

(0.801) (0.805) Real Effective Exchange Rate -0.328 -0.333

(0.195) (0.201) Direct investment inflows -0.610 -0.561

(0.572) (0.475) Portfolio investment inflows -0.585*** -0.393***

(0.170) (0.0709) Other investment inflows -0.467* -0.391**

(0.249) (0.142) Constant 26.49*** 78.02*** 27.39*** 78.44***

(28)

28

Observations 713 436 714 436

R-squared 0.360 0.448 0.358 0.448 Number of countries 28 25 28 25

Country FE YES YES YES YES

Year FE YES YES YES YES

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

During a surge episode, the effects of an increase in gross capital inflows on the central government debt-to-GDP ratio are indeed a lot smaller compared to in stable episodes, seen by the low coefficients. Furthermore, the coefficients become insignificant. Even if capital inflows do not promote economic growth anymore, it is expected to appreciate the currency, increase inflation, and lower interest rates, which stimulate government debt reductions. Perhaps the government itself is intervening in these factors through its own monetary policy. It could stop the exchange rate from appreciating by using its foreign reserves. Furthermore, by tightening monetary policy, which means increasing interest rates, the government can reduce the overheating of the economy and the increase in inflation rates.

After disaggregating gross capital inflows, the expected outcomes are an insignificant effect of an increase in direct investment inflows, and a significant effect by an increase in portfolio and other investment inflows. Furthermore, lower coefficients are expected compared to stable episodes. Similar to stable episodes, direct investment inflows once again provide an insignificant coefficient. This can be explained by the crowding-out effect of direct investment inflows, which substitute domestic capital for making investments, and the amount of direct investment inflows contributing to unproductive investments. Therefore, this does not lead to additional economic growth and a reduction in the central government debt-to-GDP ratio.

For portfolio investment inflows, an increase does have a significant negative effect on the debt ratio. Although those effect are lower than during stable episodes. While other investment inflows do not have a significant effect. Even though both these types of capital inflows are very volatile and cause exchange rate, interest rate, and inflation volatility. What makes the difference in their significance could be the impact on GDP growth. For portfolio investment inflows an increase in GDP growth was find, while other investment inflows stimulate the extension of credit by banks to the real estate sector and the growth effect can disappear.

Table 4: Effect of total and disaggregated gross capital inflows on central government debt-to-GDP - surge episode

(1) (2) (3) (4)

(29)

29

(0.0909) (0.0774)

Real GDP growth rate 0.416 0.182 (0.481) (0.551) Long-term real interest rate 1.550 1.479

(1.298) (1.310) Government primary balance -0.444 -0.413

(0.661) (0.655) Real effective exchange rate -0.267 -0.283

(0.179) (0.181) Direct investment inflows -0.149 -0.0766

(0.145) (0.139) Portfolio investment inflows -0.216* -0.125*

(0.107) (0.0792) Other investment inflows -0.106 -0.0510

(0.0783) (0.0627) Constant 48.82*** 80.17*** 48.78*** 82.65*** (8.448) (15.28) (8.596) (15.63) Observations 149 120 149 120 R-squared 0.431 0.675 0.438 0.680 Number of countries 28 24 28 24

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

During a stop episode, we expect reductions in gross capital inflows. The results show a decrease in gross capital inflows will significantly increase the central government debt-to-GDP ratio, which is as expected by the literature. Every one percent reduction of gross capital inflows indirectly increases the central government debt-to-GDP ratio by 0.167 percent. This percentage seems small, but in some countries, gross capital inflows dropped by over a hundred percent of GDP during the sudden stop episode, which would increase the central government debt-to-GDP ratio by 16.7 percent, and therefore becoming a substantial increase. Add up the fiscal costs of banking crises for governments and stop episodes could lead to government defaults.

Results of the regressions on the disaggregated gross capital inflows during sudden stop episodes are shown in column three and four. A reduction in direct investment inflows does not significantly increase the central government debt-to-GDP ratio. This would make sense considering the argument that direct investment inflows during surge episodes increasingly finance unproductive investments. The decrease in direct investment inflows are most likely coming from the stop in financing those kind of projects.

(30)

30

indeed show a significant negative effect on the central government debt-to-GDP ratio. By affecting the economy and causing banking crises. Overall, it are the other investment and portfolio investment inflows that can overheat the economy during surge episodes, and are the ones that causes the most disruption during stop episodes.

Table 5: Effect of total and disaggregated gross capital inflows on central government debt-to-GDP - stop episode

(1) (2) (3) (4)

Disaggregated Disaggregated with controls Gross capital inflows -0.301*** -0.134***

(0.0790) (0.118)

Real GDP growth rate -1.488 -0.372

(1.104) (1.067)

Long-term real interest rate 3.223*** 3.274***

(0.935) (0.899)

Government primary balance -1.546 -2.083*

(1.078) (1.210)

Real effective exchange rate -0.924* -0.918*

(0.510) (0.466)

Banking crisis costs -2.284 -1.796

(2.023) (1.658)

Direct investment inflows -0.670 -0.191 (0.347) (0.556) Portfolio investment inflows -0.906* -0.693*

(0.490) (0.490) Other investment inflows -0.310*** -0.150**

(0.0828) (0.108) Constant 59.14*** 167.9*** 57.73*** 170.5*** (3.904) (54.96) (4.098) (51.22) Observations 131 100 131 100 R-squared 0.516 0.606 0.528 0.642 Number of countries 28 24 28 24

(31)

31

4.2 Limitations

This thesis is dealing with several limitations. In terms of studying the effects during surge and stop episodes, most papers use quarterly data. Especially portfolio and other investment inflows can have extremely high volatility. Therefore surge and stop episodes can occur in a certain year and already recover to a stable level of gross capital inflows within that year. These surge and stop episodes are not identified when using annual data. Using quarterly data would have provided a more accurate representation of all the surge and stop episodes of capital flows. However, data limitations have made it difficult to collect quarterly data on all variables. Besides this, the episodes have been calculated based on total gross capital inflows only, and the assumptions has been made that all components would surge and stop during the same time. Although a correlation can be seen in figures 1 to 3, there are also some situations where one of the three components does not experience a surge while the others do.

Furthermore, the method of calculation the gross capital inflows still involves netting leading some negative values for gross capital inflows. While the method is currently standard in the literature, it does not represent true gross capital inflows. A step in the right direction is the method by Janus and Chrichton (2013). By adding negative gross inflows to outflows, and negative gross outflows to inflows, negative inflows are removed and the level of flows is increased. While this method could have been used, calculating all these changes by hand is too time consuming.

5. Conclusion

(32)

32

(33)

33

References

Agarwal, R. (1997). Foreign Portfolio Investment In Some Developing Countries: A Study of Determinants and Macroeconomic Impact . Indian Economic Review, 217-229. Ahmed, S., & Ansari, M. (1998). Financial sector development and economic growth: The

South-Asian experience. Journal of Asian Economics, 503-517.

Aizenman, J., & Marion, N. (2011). Using inflation to erode the US public debt. Journal of

Macroeconomics, 524-541.

Aizenman, J., Jinjarak, Y., & Park, D. (2013). Capital flows and economic growth in the era of financial integration and crisis. Open Economies Review, 371-396.

Albulescu, C. T. (2015). Do Foreign Direct and Portfolio Investments Affect Long-Term Economic Growth in Central and Eastern Europe? Procedia Economics and Finance , 507-512.

Amaglobeli, D., End, N., Jarmuzek, M., & Palomba, G. (2015). From Systemic Banking Crises

to Fiscal Costs: Risk Factors. IMF Working Paper No. 15/166.

Bailliu, J. N. (2000). Private Capital Flows, Financial Development, and Economic Growth in

Developing Countries. Bank of Canada Working Paper 2000-15.

Banu, I. M. (2013). The Impact of Credit on Economic Growth in the Global Crisis Context.

Procedia Economics and Finance, 25-30.

Barrell, R., & Pain, N. (1997). Foreign Direct Investment, Technological Change, and Economic Growth within Europe. Economic Journal, 1770–1786.

Bezemer, D., Grydaki, M., & Zhang, L. (2016). More mortgages, lower growth? . Economic

Inquiry, 652-674.

Bosworth, B., & Collins, S. (1999). Capital inflows, Investment, and Growth.

Broner, F., Didier, T., Erce, A., & Schmukler, S. L. (2013). Gross capital flows: Dynamics and crises. Journal of Monetary Economics, 113-133.

Bruno, V., & Shin, H. S. (2015). Cross-Border Banking and Global Liquidity . The Review of

Economic Studies, 535-564.

Calderon, C., & Kubota, M. (2012). Gross inflows gone wild : gross capital inflows, credit

booms and crises. Policy Research Working Paper Series 6270.

Calvo, G. A., Leiderman, L., & Reinhart, C. M. (1996). Inflows of Capital to Developing Countries in the 1990s. Journal of Economic Perspectives , 123-139.

Calvo, G., & Guidotti, P. (1992). Optimal Maturity of Nominal Government Debt: An Infinite-Horizon Model. International Economic Review, 895-919.

Cardarelli, R., Elekdag, S., & Kose, A. (2010). Capital inflows: Macroeconomic implications and policy responses. Economic Systems, 333-356.

Carstens, A., & Schwartz, M. J. (1998). Capital flows and the financial crisis in Mexico. Journal

of Asian Economics, 207-226.

Carvalho, D., & Fidora, M. (2015). Capital inflows and euro area long-term interest rates.

(34)

34

Cecchetti, S. G., & Kharroubi, E. (2015). Why does financial sector growth crowd out real

economic growth?

Choong, C.-K., Baharumshah, A. Z., Yusop, Z., & Habibullah, M. S. (2010). Private capital flows, stock market and economic growth in developed and developing countries: A comparative analysis. Japan and the World Economy, 107-117.

Collier, P., Hoeffler, A., & Pattillo, C. (2001). Flight Capital as a Portfolio Choice . World Bank

Economic Review, 55-79.

Crystallin, M., Efremidze, L., Kim, S., Nugroho, W., Sula, O., & Willett, T. (2015). How

Common Are Capital Flows Surges? How They Are Measured Matters -A Lot. CIEPS

Working Paper.

Doytch, N., & Cakan, E. (2011). Growth Effects of Mergers and Acquisitions: A Sector-level Study of OECD countries. Journal of Applied Economics and Business Research, 120-129.

ECB. (2011). Financial Stability Review.

Eichengreen, B., & Gupta, P. (2016). Managing Sudden Stops. World Bank Policy Research Working Paper No. 7639.

Fernández-Arias, E., & Hausmann, R. (2001). Is foreign direct investment a safer form of financing? . Emerging Markets Review, 34-49.

Forbes, K. L., & Warnock, F. E. (2012). Capital flow waves: Surges, stops, flight, and retrenchment. Journal of International Economics, 235-251.

Forbes, K., & Warnock, F. (2012). Capital flow waves: surges, stops, flight, and retrenchment.

Journal of International Economics 88, 235-251.

Forster, K., Vasardani, M., & Ca'Zorzi, M. (2011). Euro Area Cross Border Financial Flows and

the Global Financial Crisis. ECB Occasional Paper Series no. 126.

Frankel, J., & Rose, A. (1996).

Currency Crashes in Emerging Markets: An Empirical Treatment.

International Finance Discussion Paper 534.

Furceri, D., Guichard, S., & Rusticelli, E. (2012). Episodes of Large Capital Inflows, Banking and Currency Crises, and Sudden Stops. International Finance, 1-35.

Ghosh, A. R., Qureshi, M. S., Kim, J. i., & Zalduendo, J. (2014). Surges . Journal of

International Economics, 266-285.

Ghosh, A., & Qureshi, M. (2016). Capital Inflow Surges and Consequences. ABDI Working Paper 585.

Giovannini, A., Iacopetta, M., & Minetti, R. (2013). Financial Markets Banks and Growth:

Disentangling the Links. Sciences Po.

Gourinchas, P.-O., & Obstfeld, M. (2012). Stories of the Twentieth Century for the Twenty-First.

American Economic Journal: Macroeconomics, 226-265.

Gourinchas, P.-O., Philippon, T., & Vayanos, D. (2017). The Analytics of the Greek Crisis.

The effect of gross capital inflows on the central government debt-to-GDP ratio