Capital Flows to Central and Eastern Europe: The Impact of the EU Accession Process

(1)

Capital Flows to Central and Eastern Europe:

The Impact of the EU Accession Process

Master’s Thesis

MSc Economic Development & Globalization

Ahmet Ihsan Kaya - S4065077

a.i.kaya@student.rug.nl

Supervisor: Prof. Dr. Jakob de Haan

Co-assessor: Dr. Andreas C. Steiner

(2)

i

Capital Flows to Central and Eastern Europe:

The Impact of the EU Accession Process

Ahmet Ihsan Kaya*

Abstract

This study investigates the impact of EU integration on capital flows to the CEE countries. Using annual data between 1992 and 2018, the empirical models regress different types of capital flows on EU candidacy and membership dummies along with several push and pull factors. The results show that although being a candidate country increases net capital inflows, being a member of the EU is negatively associated with net capital flows. The sensitivity analysis demonstrates that EU membership had a strong positive impact on gross flows before the GFC, but this impact has turned to negative for both gross and net flows after the GFC. We also find that net FDI inflows are affected positively by the EU candidacy, while the impact of EU membership on net FDI inflows is negative. Moreover, the positive impact of EU membership on net other flows has been through its interaction with the domestic institutional quality.

Keywords: Capital flows; push and pull factors; European Union integration; fixed effect panel data model

JEL Classification: C33; F21; F36

(3)

ii

Following the dissolution of the Soviet Union, the European Union (EU) supported the Central and Eastern European (CEE1_{) countries’ democratization and transition to a market economy in} line with its aim of bringing peace, stability and prosperity to the continent. These efforts led to successive reforms in the CEE countries during the 1990s and 2000s. As an institutional and legal anchor for these countries, the EU accession process paved the way for political and economic stability, deepening their trade and financial integration to the EU, and improvements in the domestic business environment. These developments, together with increasing financial globalization, triggered a massive amount of capital flows to the region, which supported their convergence to the EU level of income per capita.

The impact of foreign capital flows on domestic economic performance continues to be a topic of discussion in the literature. However, there is a consensus among economists that the costs and benefits of capital flows differ according to the types of flows. Most studies argue that short-term and fickle types of foreign capital flows introduce external risks and global volatility to the domestic economy, thus exposing them to the global financial cycle (Rossi, 2007; Magud et al., 2014). On the other hand, long-term and stable capital flows help host countries in financing their investments, lowering their capital costs, diversifying the funding risks and transferring technology. Additionally, foreign capital flows also provide collateral benefits through their impact on domestic financial developments, institutional quality, corporate governance and macroeconomic policies (Kose et al., 2009). Furthermore, the literature shows that financial development is vital for domestic economic performance through its impact on country-specific characteristics such as law and enforcement mechanisms, legal system adaptabilities and

1_{The CEE term is often used to describe former communist states in Europe, but, it varies in different country}

(7)

2

flexibilities, regulatory and supervisory efficiencies, and better monetary and fiscal policies (Levine, 2005). Arcand et al. (2015) also argue that the relationship between financial development and economic performance may be non-linear and that the positive impact of financial depth on economic growth vanishes after a certain threshold. They show that more finance is associated with lower economic growth when credit to the private sector reaches around 100 percent of GDP because of rising economic volatility, increased probability of banking crisis, and potential misallocation of resources. Because of these direct and indirect impacts on the domestic economy, the drivers of capital flows have been investigated extensively in the literature.

Most studies focus on developing countries, acknowledging that the impact and drivers differ between advanced economies (AE) and emerging markets (EM). However, EMs are also heterogeneous in their economic structures and growth models, thus the impact and drivers of capital flows also differ among them (Kang & Kim, 2019). Considering that the EU integration process serves as an anchor for the economic and political transformation of the CEE countries (Schönfelder & Wagner, 2016) and that their growth models differ from other EMs (Becker et al., 2010), specific analysis is needed for the CEE countries. A couple of studies in the literature attempted to fill this gap by investigating the impact of EU integration and showed that the accession process had a direct and indirect positive influence on foreign direct investment (FDI) in the CEE countries (Bandelj, 2010; Bevan & Estrin, 2004; Clausing & Dorobantu, 2005). However, these studies ignored portfolio and other credit flows, which make up a significant portion of the total flows into the region.

(8)

3

regression models, the Arellano & Bover (1995) and Blundell & Bond (1998) system GMM estimator is also utilized as a sensitivity analysis. Further, we split the sample before and after the crisis to examine whether the behaviour of the flows changed as a result of the global financial crisis (GFC) and we check whether the results are robust to excluding countries which do not have a communist history.

The rest of the study is organized as follows. Section 2 provides a brief background of the EU enlargement to the CEE and summarizes the capital flow developments in recent decades. Section 3 outlines the literature on the determinants of capital flows to developing countries. Section 4 describes the data and methodology. Section 5 presents the estimation results and the last section concludes.

2. EU Enlargement to the CEE and Capital Flow Developments

Since the foundation of the European Economic Community (EEC) between Belgium, France, West Germany, Italy, Luxembourg and the Netherlands with the Treaty of Rome in 1958, there have been different phases in the EU enlargement process. In the first phase, Denmark, Ireland and the United Kingdom joined the EEC in 1973. The second and third phases consisted of the accessions of Southern European countries in the 1980s. In 1981, Greece became the 10th member country of the EU, followed by Spain and Portugal in 1986. In the fourth phase, Austria, Finland and Sweden have been granted full membership status in 1995 and the total number of members increased to 15 countries (Emmert & Petrovi, 2014).

The CEE enlargement was one of the central topics during the Copenhagen Summit in 1993. The EU welcomed the CEE countries’ transition efforts and provided them with an opportunity to become member states on the condition that they would satisfy the Copenhagen Criteria2_{. In line} with the European Commission’s Agenda 2000 report, Hungary, Poland, Estonia, Czechia and

2_{The Presidency Conclusions of the European Council meeting in Copenhagen in 1993 specifies those conditions as}

(9)

4

Slovenia were considered as the first round of enlargement countries among the CEE (European Commission, 1997) and together with Cyprus, Malta, Latvia, Lithuania and Slovakia, they constituted the fifth enlargement of the EU. The sixth and seventh enlargement of the EU also focused on the CEE. With the accessions of Bulgaria and Romania in 2007 and Croatia in 2013, the number of EU member countries became 28. After the withdrawal of the UK in 2020, the EU now covers 4 million km2_{and has more than 445 million inhabitants (Eurostat, 2020).}

During this process, the member countries also focused on deepening the union by constantly reforming its institutional structure and by gradually transforming the economic union into a supra-governmental decision-making body. The prominent cornerstones of the deepening process were the Single European Act signed in 1986 to speed up the single market and the Maastricht Treaty (Treaty on European Union) signed in 1992 to prepare the European Economic and Monetary Union (EMU) and to introduce several elements necessary for forming a political union such as common foreign and internal affairs policies. The deepening of the EU continued with the Treaty of Amsterdam in 1997 and the Treaty of Nice in 2001 to reform the EU institutions in preparation for the enlargement with future members. Finally, the Treaty of Lisbon (Treaty on the Functioning of the European Union) was signed in 2007 to make the EU more democratic, more efficient and better able to address global problems (Eurofound, 2017).

(10)

5

15 more have been under negotiation since then. However, despite the collaborations in the areas related to migration, counterterrorism, investment and trade, the accession negotiations between the EU and Turkey have come to a standstill and the European Commission ruled out Turkey’s full membership for the near future due to the deterioration of bilateral relations in recent years (Juncker, 2017). Apart from these five official candidates, Bosnia and Herzegovina and Kosovo are considered as potential candidates as of 2020 (European Commission, 2020).

The EU enlargement was perceived as one of the most important challenges of the EU because of the massive differences between Western Europe and the CEE in terms of economic and social development. On the other hand, it was an opportunity for the EU to inspire democratic change and economic liberalisation in the region, thus extending stability to Eastern Europe. Baldwin et al. (1997) discuss the costs and benefits of the enlargement for the EU and the CEE. From the EU side, the benefits are summarized as peace and stability in the continent, increasing market size and higher capital gains with favourable investment opportunities. The possible costs are related to extra budget costs and increasing economic divergence within the union, which could limit the ability to make decisions that each member country benefits. From the CEE countries’ perspective, the main benefits of (possible) EU membership are increasing investment opportunities as a result of improvement in the investment climate, the decline in risk premia, exploiting the trade opportunities of the single market and direct transfers from the EU. The associated costs are related to political and bureaucratic adjustment costs. Considering that the gains more than offset the costs, Baldwin et al. (1997) argue that the CEE enlargement is beneficial for both sides. Breuss (2002), on the other hand, by pointing out the enormous heterogeneity between member states, argues that although the average impact would be positive, the CEE enlargement could also have acted as an exogenous shock which leads to asymmetric disturbances in the EU. Overall, the CEE enlargement has been found successful and the European economy became stronger, more dynamic and better equipped to global competition3_.

3_{A summary of the results of the studies measuring the economic impact of enlargement can be found in DG-ECFIN}

(11)

6

The large-scale structural reforms in line with the EU accession process led to a substantial amount of capital inflows to the CEE countries during the 2000s. Figure 1 shows the gradual escalation of the gross capital inflows and outflows to the CEE countries after the dissolution of the Soviet Union. Thanks to the abundance of global liquidity and in parallel with their gradual integration with the EU, cross-border capital flows had accelerated and the gross inflows to the CEE peaked with 20.8 percent of GDP in 2007. After that, gross inflows to GDP ratio plummeted as a result of the GFC and have remained around 6 percent on average until the end of 2018. A similar pattern can be observed in the gross capital outflows from the CEE. After reaching 11 percent in 2007, the gross outflows-to-GDP ratio has been around 1 to 5 percent.

As gross inflows outweighed the outflows, there have been net capital inflows to the region throughout the period. As can be seen from Figure 2, FDI has been the most prominent contributor of net capital flows to the CEE4_{. FDI has also been the most stable type of capital flows: it} continued to be positive even during the GFC although declining afterwards. Net portfolio flows and other credit flows, on the other hand, have been volatile. Credit flows surged just before the GFC, while portfolio flows were the most important driver of the net capital inflows between 2010 and 2014 and turned negative afterwards. Although the region attracted large capital inflows until the GFC, as can be seen from Figure 3, the inflows were unevenly distributed across countries between 1992 and 2018. Montenegro had been the most net foreign capital attracting country with 21.5 percent of GDP, followed by Bosnia and Herzegovina, and Serbia. Although Turkey attracted more foreign capital than all other countries in terms of the total amount, it was among the lowest group of countries when scaled by GDP. It is also striking that except for Cyprus, all countries registered positive net FDI inflows between 1992 and 2018.

4_{FDI is cross-border investments which provide non-residents to control or to have a significant degree of influence}

(12)

7

Figure 1: Gross Inflows and Outflows5 _{Figure 2: Net Capital Flows by Components}

Figure 3: Net Capital Flows by Countries Figure 4: Foreign Assets and Liabilities6

Sources: IMF BOP and World Economic Outlook (WEO)

* Net FDI (65.9 %) and net portfolio (-69.8 %) flows in Malta is not shown in Figure 3 for illustrative purpose.

The surge in capital flows resulted in a massive accumulation of net foreign liabilities in the CEE countries, which can be observed from the international investment position (IIP) in Figure 4. After a period of steady increase from 1992 to 2005, the net IIP position escalated rapidly until the GFC and has remained around 50 percent of GDP since then. The FDI stock constitutes the majority of the foreign liabilities of the CEE countries with 76 percent of GDP in 2018, followed by other liabilities with 42.2 percent and portfolio liabilities with 23.2 percent. On the assets side, the CEE residents hold FDI assets which equal 38.7 percent relative to GDP, followed by other assets with

5_{According to the Balance of Payments (BOP) methodology, the incurrence of domestic liabilities to non-residents}

(foreign activity in the host country) implies gross inflows, while the acquisition of foreign assets from non-residents (domestic activity in the rest of the world) indicates gross outflows from host country’s point of view (Wang, 2009). Gross outflows are shown negative for illustrative purposes.

6_{The assets are shown negative for illustrative purposes. Further information is provided in the previous footnote.}

-15 -10 -5 0 5 10 15 20 25 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P)

Gross Inflows Gross Outflows Net Capital Flows

-4 -2 0 2 4 6 8 10 12 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P)

Net FDI Net Portfolio Investments Net Other Flows Net Capital Flows

-5 0 5 10 15 20 25 M o nt en eg ro B os n ia & H er z. Se rb ia B ul ga ri a C ro at ia A lb an ia M ac ed o ni a R om an ia Sl o va ki a La tv ia C ze ch ia Es to n ia Li th u an ia Ko so vo H u ng ar y C yp ru s Po la nd Tu rk ey M al ta * Sl o ve ni a (% o f G D P)

-150 -100 -50 0 50 100 150 200 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P)

(13)

8

21.3 percent and portfolio assets with 14.4 percent in 2018. Additionally, the reserve assets of all CEE countries are 20.4 percent as of 2018.

Although one of the most important drivers of capital flows was the excess global liquidity especially before the GFC, we do not see the same surge in net capital flows to other developing regions. As depicted in Figure 5, the net-capital-flows-to-GDP ratio in Emerging and Developing Asia ranged from -3 to 4 percent in the 1997-2018 period. Thanks to steady FDI inflows, Latin America and the Caribbean region attracted more foreign capital during the entire period from 1992 to 2018 than other regions except for the CEE. The capital flows to Sub-Saharan Africa have also been positive since the GFC, but the magnitude of the flows relative to GDP remained limited. On the other hand, the net flows to the Middle East and Central Asia region have been volatile. Although they attracted FDI inflows to some extent, portfolio investments and other credit flows have been negative after 2000.

(14)

9

Figure 5: Net Capital Flows and Components in Different Regions

Source: IMF WEO Database

As their growth model largely depends on foreign capital inflows, the CEE countries were affected more severely compared to other developing economies when the cross-border capital flows dried up during the GFC. Thus, the growth model depending on the foreign flows has been questioned. On the one hand, capital flows provided opportunities for the CEE countries to finance their high investment needs, to spur domestic production, to transfer technology; and hence to converge to the EU-15. On the other hand, they caused instability, increased countries’ vulnerabilities and propagated the impact of the GFC. The latter is especially noticeable for portfolio and other flows to the CEE, which are unstable, fickle and unfettered by definition (Rossi, 2007). Therefore, it is important to analyse the drivers of capital flows in the CEE countries and to examine how the EU integration process contributed to different types of flows in these countries, which is the central aim of this study.

-4 -3 -2 -1 0 1 2 3 4 5 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P)

Emerging and Developing Asia

-4 -3 -2 -1 0 1 2 3 4 5 6 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P)

Latin America and the Caribbean

-8 -6 -4 -2 0 2 4 6 8 10 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P)

Middle East and Central Asia

-6 -4 -2 0 2 4 6 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 (% o f G D P) Sub-Saharan Africa

(15)

10 3. Literature on the Determinants of Capital Flows

The financial globalization that accelerated since the 1990s caused a significant modification in the domestic economic policymaking processes. Although most of the international financial flows are within advanced countries (Borio & Disyatat, 2011), the impact of financial globalization on developing economies might be more vital in terms of influencing the real economy. This is because the magnitude and direction of the business cycles are more symmetric in advanced economies (AEs) than in their developing counterparts (Rand & Tarp, 2002), allowing the former to follow similar macroeconomic policies. For instance, as a response to a negative shock to the economy, AEs apply expansionary monetary policies to revive domestic economic activity without worrying too much about inflation. However, loose monetary policy in developing countries might discourage foreign investors that they need to finance current account deficits and to rollover matured foreign debts. Besides, as the CEE countries’ economic structures are more similar to those of developing countries, our focus in this chapter will be on capital flows into developing countries.

Before delving into the literature on the drivers of capital flows, three distinctive tendencies within the literature should be mentioned. Firstly, some studies analyse the factors affecting capital flows by distinguishing them as push (global) and pull (domestic) factors following the seminal contributions of Calvo et al. (1993) and Fernandez-Arias (1996), while others prefer using the growth and interest rate differentials between EMs and AEs such as Herrmann & Mihaljek (2013) and Ahmed & Zlate (2014). The rationale behind the differential approach is that some factors like the contagion effects and investor behaviours may not be classified as a push or pull factor. Additionally, from the perspective of financial stability, it may not be important to take into account whether the change in the growth and interest rate differentials is because of the developments in advanced economies or host countries (Ahmed & Zlate, 2014).

(16)

11

However, the decline in the AE’s growth rates might also lower the supply of financial funds available, thereby reducing the capital flows towards the EMs. Indeed, the literature provides evidence that the AE growth rate moves in the same direction as capital flows to EMs. Furthermore, the push-pull approach provides a clearer picture of the sources of flows, which helps to understand whether the sustainability of these flows is under the control of host country policymakers. If the push factors are more dominant than pull factors, one can argue that the inflows are volatile and subject to sudden stops and reversals (Fernandez-Arias, 1996).

The second divergence in the literature is about which type of flows should be considered. In the Balance of Payments (BOP) definition, gross inflows cover all types of inward fund flows of non-residents, while gross outflows look at outward investments of residents. The difference between gross inflows and outflows gives net capital flows, which is the “below the line” counterpart of the current account balance, excluding reserves and including capital accounts. Gross inflows can bring considerable benefits such as allowing residents to diversify their portfolios, providing more efficient resource allocation and improving liquidity in host countries (Taylor & Sarno, 1997). However, they can also be harmful in terms of financial stability by intensifying credit growth and leading to asset price bubbles (Ahmed & Zlate, 2014). Therefore, proponents of this approach argue that it is important to take into account the gross flows and to differentiate domestic and foreign investor behaviour when analysing international financial transactions.

(17)

12

Settlements (BIS) as a proxy for the BOP definition of capital flows (see e.g. Friedrich & Guérin, 2019; Herrmann & Mihaljek, 2013; Li et al., 2018).

The third major distinctive approach in the literature is that an increasing number of studies first identify the extreme episodes of flows such as “surges” and “sudden stops” and then analyse the impact of push and pull factors on those extreme flows (Reinhart & Reinhart, 2008; Forbes & Warnock, 2012; Ghosh et al., 2014; Li et al., 2018, 2019). Considering that the dynamics of foreign capital differ and the distortive impacts are greater than their benefits when the flows show excessive behaviour, this approach increases our understanding of international financial transactions. On the other hand, this approach ignores the dynamics of flows in normal times which is more common by definition. Additionally, this approach makes it difficult to find out the long-term delong-terminants of capital flows and how they changed over time (Ahmed & Zlate, 2014). Another shortcoming of this approach is that the methodologies to detect the extreme episodes largely depend on arbitrary thresholds and sample-specific assumptions (Friedrich & Guérin, 2019). In the following, we will outline the prominent literature concerning the different type of approaches discussed above.

Among the earlier literature on the determinants of capital flows, Calvo et al. (1993) investigate the role of the external factors on capital inflows in Latin American countries by using monthly data from January 1988 to December 1991. The results from structural VAR analysis and impulse-response functions show that low interest rates, economic recession and balance of payments developments in the US played an important role in explaining capital flows to Latin American countries, which led to the accumulation of foreign reserves and real exchange rate appreciation. Fernandez-Arias (1996) also supports the “push” story with the finding that declining global interest rates have been the most important determinant of private capital inflows to middle-income countries after 1989. He further argues that falling international interest rates exert additional “push” impact on capital flows by improving host country creditworthiness which was commonly considered as “pull” factors.

(18)

13

Asian countries. However, they find that the impact of global factors has been more pronounced in the case of bond flows. Using weekly cross-border fund flows data from the EPFR database, Fratzscher (2012) investigates whether push or pull factors are the main driver of the capital flows to AEs and EMs during the GFC period. The results from his factor model show that push factors such as global liquidity, risk factors and key crisis events explain a large part of the flows during the GFC. On the other hand, pull factors captured by macroeconomic fundamentals, country risk indicator and quality of institutions determined the allocation of these fund flows between countries.

Among the studies which employ the differential approach as discussed above, Ahmed & Zlate (2014) examine the drivers of net private capital inflows to EMs between 2002 and 2013. Their main findings indicate that growth and interest rate differentials between EMs and AEs as well as the risk appetite of global investors are important determinants of net private capital flows. However, the behaviour of portfolio inflows has changed after the GFC because of its higher sensitivity to the interest rate differentials. Their extended models also reveal that unconventional monetary policy in the US has also a positive effect on portfolio inflows, while capital controls against the fickle type of flows are effective to dissuade total and net portfolio inflows. Growth and interest rate differentials between host and mature countries are also found important for cross-border bank flows in the study by Herrmann & Mihaljek (2013). Based on a gravity model for 45 advanced and emerging countries from 1993 to 2008, their results also show that the decline of cross-border bank flows during the crisis is mainly because of increasing global risk aversion and financial market volatility.

(19)

14

(20)

15

Hannan (2017) also investigates the push, pull and structural drivers of different type of flows in EMs after the GFC by using a differential approach. For the total and private net capital flows, he finds that along with the growth and interest rate differentials, international reserves, global risk aversion and the US corporate spread have been important for both gross and net capital inflows. However, these results significantly differ across the type of flows. For instance, while trade openness and global risk aversion exert considerable impact on gross portfolio inflows, capital account openness and financial development are found essential for net FDI and net portfolio equity flows. Additionally, the results show that the sensitivity of certain flows to the push and pull factors significantly increases when the flows exceed one standard deviation above or below the mean.

Although it is widely accepted that the AEs and EMs differ in terms of forces that attract capital flows as they are considered riskier (Bluedorn et al., 2013), the heterogeneity among EMs in size, economic structure and level of development is largely ignored in the literature. Recently, Kang & Kim (2019) investigated this heterogeneity in the impact of push and pull factors on net capital flows by grouping EMs as Asia, Eastern Europe, Latin America and others. Their findings show that while both push and pull factors are significant for AEs, the impact of push factors on EM capital flows are more noticeable. Also, they showed that EMs are indeed heterogeneous. In explaining net capital flows, only domestic GDP growth is significantly positive for Latin American countries, while the US interest rates are the only significant factor for Eastern European countries. In Asian countries, however, various global and domestic factors were in force from 1997 to 2015.

(21)

16

cost differences. They argue that investors from Western countries consider the potential accession to the EU as a signal of creditworthiness of the CEE countries.

Clausing & Dorobantu (2005) also analyse the impact of key EU announcements about the accession process on FDI inflows to the CEE countries by using three different dummies. The first dummy represents the European Council’s Copenhagen commitment in 1993 to grant the CEE access to the EU. The second and third dummies stand to represent the EU’s first and second round enlargement countries in line with Agenda 2000. The results show that along with the market size and cost-minimizing motives, the EU integration announcements had statistically and economically important impact on the candidate CEE countries. The study by Bandelj (2010) shows that the EU integration indirectly influenced FDI inflows to CEE through its impact on state decision-making and country legacies, instead of a direct effect due to reduced investment risks. Two important mechanisms have been pointed out. The first one is through the policy change to provide the liberalisation of strategic sectors to foreigners required by the Copenhagen Criteria and the acquis communautaire. The second mechanism is through the impact of domestic policymakers’ attempts to attract FDI as a policy for their economic development.

(22)

17 4. Data, Methodology and Preliminary Analysis 4.1. Data and Definition of Explanatory Variables

(23)

18

expected that the willingness to invest in the CEE countries increases with increasing global economic activity.

Domestic pull factors refer to host countries’ macroeconomic and institutional determinants that help them to attract foreign capital. Similar to global factors, we include domestic real interest rates and GDP growth to the model to capture domestic investment returns. Although we employed the long-term government bond yields in the US for global financial conditions, since the maturities of government bonds are shorter and adequate government bond data covering the sample period is not available in most of these countries, we mainly rely on deposit interest rates and extend it with benchmark government bond yields or monetary policy interest rates for some countries7_{. Higher rates of real domestic interest rates and GDP growth associate with higher} domestic returns and are likely to increase capital flows to the CEE. Further, we include the inflation environment and real effective exchange rate (REER) deviations from the trend as potential explanatory domestic variables for capital flows (Ghosh et al., 2014; Li et al., 2018)8_. Low and stable inflation is likely to positively affect capital flows, while a real exchange rate depreciation makes domestic assets cheaper for international investors and may increase the opportunities of domestic beneficiaries to reach foreign funds. We also include de jure financial openness measured by Chinn & Ito (2008) for which we expect a positive coefficient.

In addition to domestic macroeconomic variables, we also investigate the impact of domestic institutional quality on capital flows. Because the institutional quality is a broad, multidimensional and qualitative indicator to a large extent, it is very hard to construct a single variable that takes into account the different dimensions of institutions. There are different indices in the literature based on various qualitative and quantitative indicators focusing on different aspects of institutions, yet, efforts to produce a complete and composite index is relatively limited. Kunčič (2014) provides a composite index on legal, political and economic institutions by clustering 30 different institutional indicators in the literature. However, it does not cover the period after 2010, which prevents us to use it as a proxy to the institutional quality. A widely used multidimensional

7_{For Estonia, Latvia, Lithuania, and Malta harmonized Euro Area interest rates; for Cyprus, Slovakia, and Slovenia}

benchmark government bond yields; for Poland, monetary policy interest rates are used.

(24)

19

dataset on institutional quality is Worldwide Governance Indicators (WGI)9_{, provided by} Kaufmann et al. (2011). As the dataset does not provide any single indicator, we constructed a composite institutional quality index by factor analysis using six sub-components of the WGI10_. Considering the role of domestic institutional quality for attracting foreign capital (Papaioannou, 2009), we expect the coefficient on our institutional quality proxy to be significantly positive. Detailed descriptions of the variables and data sources are presented in the Appendix.

4.2. Estimation Methodology

As discussed, capital flows to developing countries have been analysed by considering push and pull factors following the influential works by Calvo et al. (1993) and Fernandez-Arias (1996). In this study, we follow the same approach and estimate the following econometric model for our base estimation:

𝑌 = 𝛼 + 𝑋 𝛽 + 𝑋 𝛽 + 𝑋 𝛽 + 𝜀 (1)

where 𝑌 is capital flows as a percentage of GDP by instrument k (gross flows, net flows, direct investments, portfolio investments and other investments) for country i in year t, 𝑋 is a vector of global push factors, 𝑋 is a vector of lagged domestic pull factors11_{, 𝑋 is a vector of} institutional factors, 𝛼 is the unobserved country-specific effects and 𝛽 , 𝛽 and 𝛽 are estimated coefficients for global, domestic and institutional variables. To investigate the impact of EU integration, we extend the model by utilizing candidacy and membership dummies as follows:

𝑌 = 𝛼 + 𝑋 𝛽 + 𝑋 𝛽 + 𝑋 𝛽 + 𝐷 .𝛿 + 𝜀 (2)

9_{The Worldwide Governance Indicators (WGI) measures the institutional quality for over 200 countries in following}

dimensions: Voice and Accountability, Regulatory Quality, Political Stability and Absence of Violence, Rule of Law, Government Effectiveness, Control of Corruption. Each component of the index takes a value between -2.5 and 2.5 with higher values to indicate better governance. The index covers the period between 1996-2018 and is available on a two-year basis until 2002. To avoid further loss of observation, we filled the missing years by taking the average of the previous and following years’ WGI values as in Schönfelder & Wagner (2016).

10_{The results of the factor analysis and specification tests presented in Table A6 and Table A7 in the Appendix.}

Bartlett and Kaiser-Meyer-Olkin (KMO) tests imply that the data are appropriate for factor analysis (Dziuban & Shirkey, 1974). The eigenvalues of the factor analysis also show that the data can be reduced to a single factor.

11_{Domestic macroeconomic and institutional variables are included in the models one year lagged to control the}

(25)

20

where 𝐷 ._{are main variables of interest, which are dummy variables that take the value of 1 after} a country gains the candidacy or membership status, and 0 otherwise. If the EU integration process increased the type k instrument of capital flows to these countries, 𝛿 coefficient vector for candidacy or membership dummies are expected to be significant and positive.

𝑌 = 𝛼 + 𝑋 𝛽 + 𝑋 𝛽 + 𝑋 𝛽 + 𝐷 .𝛿 + 𝐷 .𝑋 𝛿 + 𝜀 (3)

Along with the level impact of the EU integration, we also take into account the interaction of candidacy and membership dummies with the domestic institutional factors (𝑋 ) to investigate how the EU candidacy and membership statuses affected the role of domestic institutions on attracting capital flows12_{. 𝛿 stands to gauge this possible impact in equation (3).}

Although most of the countries in our sample have a lot in common, i.e. shared background, common objective to be part of the EU and apply similar economic policies, there are striking economic and institutional differences among them. Most importantly, the sample varies in economic size and population. Even though we scale variables with GDP, these differences may create significant heterogeneity in terms of the impact of drivers of capital flows in those countries. Ignoring this heterogeneity among cross-sections may yield inconsistent coefficient estimations (Baltagi, 2013). Pesaran & Smith (1995) and Pesaran (2006) propose the mean group (MG) and the common correlated effects (CCE) estimators which both take into account the cross-sectional heterogeneity and the latter is also robust to cross-sectional dependence. However, as the underlying methodologies of the MG and the CCE estimators require one to apply separate time-series regressions for each cross-section, practically, these methods are only suitable for the data if the time dimension is large enough, which is not the case in our sample. Therefore, in order to control the unobserved cross-country heterogeneity, we stick to the fixed effects estimator independent from the Hausman (1978) test results.

12_{As discussed by Schönfelder & Wagner (2016), EU integration requires a candidate country to undertake a}

(26)

21

We perform the following robustness checks in addition to our main regressions. Firstly, the global financial crisis (GFC) had major implications to the dynamics of capital flows to developing countries and specifically to CEE countries because of their distinctive growth model in the pre-crisis period (Becker et al., 2010). Along with its direct impact on the magnitude of the flows, which we control for by using a dummy variable for the crisis in the main regression models, the GFC and the subsequent unconventional monetary policies by the AE central banks may also have changed the impact of capital flow drivers. Hence, as a first robustness check, we split the sample before and after the crisis to see whether the crisis has changed the impact of EU integration and other determinants.

Secondly, similar to the trade flows, capital flows may show persistence as international investors tend to invest more in countries in which they have previous experiences. This is especially true for FDI flows, which are more stable and may take time to invest in host countries (Becker & Noone, 2008 and Hossain, 2015). For this reason, we also consider the dynamic nature of capital flows by including the lag of the flows into the model as another robustness check. The inclusion of the lagged dependent variable in fixed-effects panel data models yields biased and inconsistent estimations (Nickell, 1981) as the individual-specific effects and the error terms would be correlated (Baltagi, 2013). Therefore, we use Arellano & Bover (1995) and Blundell & Bond (1998) system GMM estimator for the models in which we consider the dynamic nature of the capital flows13_.

Finally, as we have discussed, although most of the sample countries are former Eastern Bloc countries; Cyprus, Malta and Turkey do not have such a history. Even though we use country-specific fixed effects to control for unobserved heterogeneities, these three countries are fundamentally different from others. For this reason, as a final sensitivity analysis, we exclude these countries from the sample and only focus on former Eastern Bloc countries.

13_{In fact, the system GMM estimator is more suitable to panels where the time dimension is short and the}

(27)

22 4.3. Descriptive Statistics and Unit Root Tests

Descriptive statistics of different types of capital flows by EU integration status are presented in Table 1. As can be seen, mean values and standard deviations of gross inflows considerably increased as they become more integrated with the EU by gaining candidacy and membership status. The difference is more visible in gross outflows, which are significantly higher in the candidate and member countries. This explains the third column which shows that net flows were lower in the candidate countries and even declined more when they gained membership status. To put it differently, gaining candidacy and membership statuses only led to an increase in the volume of gross inflows and outflows as a result of the elimination of transaction costs in international fund transfers, but the net flows declined with increasing integration.

Table 1: Descriptive Statistics of Capital Flows by Integration Status

Gross Inflows / GDP (%) Gross Outflows / GDP (%) Net Flows / GDP (%) FDI / GDP (%) Portfolio Inv. / GDP (%) Other Inv. / GDP (%) No Integration: Observation 159 159 159 159 139 159 Mean 10.4 3.7 6.7 4.1 0.1 2.6 St. Dev. 10.6 8.1 7.2 4.5 2.7 5.3 Min -10.6 -11.2 -13.9 0.1 -13.6 -16.4 Max 69.6 61.4 51.1 36.1 9.8 30.2 Candidacy: Observation 314 314 314 313 314 314 Mean 22.5 17.7 4.8 7.8 -4.9 1.9 St. Dev. 65.7 65.4 6.8 26.1 27.3 14.0 Min -42.1 -43.9 -12.6 -77.6 -232.3 -58.2 Max 678.3 666.5 43.8 263.8 72.2 84.4 Membership: Observation 180 180 180 179 180 180 Mean 31.5 28.5 3.0 10.0 -8.6 1.7 St. Dev. 85.6 84.7 7.6 34.2 35.5 18.1 Min -42.1 -43.9 -12.6 -77.6 -232.3 -58.2 Max 678.3 666.5 43.8 263.8 72.2 84.4

Source: IMF Balance of Payment Statistics and author’s calculations.

(28)

23

investments can be the result of the convergence of bond and equity returns in new member countries. Additionally, Table A2 in the Appendix provides average capital flows by sample countries and their EU integration status. As can be seen, there is a huge variability especially in gross flows which is disguised by panel averages in Table 1. Cyprus and Malta stand as outliers with average gross inflows to GDP ratios of 91.4 percent and 142.1 percent in their candidacy period and 120.5 and 173.9 percent in their membership period14_{. Descriptive statistics of other} explanatory variables are also presented in the Appendix.

Before the regression analysis, we check the stationarity of the variables as ignoring the stationarity might result in spurious regression (Baltagi, 2013). Table 2 provides the results of the unit root tests which are performed with and without a trend. For the global variables which are cross-sectionally invariant, the standard Augmented Dickey-Fuller (ADF) test is applied (Dickey & Fuller, 1979) and the results are placed in the first panel of the table. The results show that only the world GDP growth reflects stationary behaviour at the 5 percent significance level, so, we include it into the regression model in levels. VIX, global liquidity and EPU variables are found stationary in their first differences, thus, we take the first difference of these variables as in Li et al. (2018). The result for the US real interest rates variable is a bit tricky. Although the interest rate series fluctuates within a certain corridor and thus the mean, variance and covariances are expected to be constant over time, the ADF test implies that it has a unit root. For robustness, the Phillips-Perron (1988) test that accounts for possible serial-correlation in the series is also utilized. The results of the Phillips-Perron unit root test justify our suspicion and reject the non-stationarity hypothesis of the US real interest rates in the specification with the trend15_{. As for the domestic} variables, we perform the second-generation cross-sectionally augmented panel unit root (CIPS) test developed by Pesaran (2007). As can be seen from the test results presented in the second

14_{Since gross outflows to GDP ratio in these countries also jumped with their integration status along with gross}

inflows, the net flows to GDP ratio remained relatively modest. Therefore, we excluded these two countries from the sample in the gross inflows to GDP regression specification. The average gross inflows to GDP ratios are 9.3 percent for no integration, 8.9 percent for candidacy, and 8.4 percent for membership statuses when we exclude Cyprus and Malta.

15_{The Z(t) statistics of specifications with and without trend are -2.257 and -4.799, and corresponding MacKinnon}

(29)

24

panel of Table 2, all domestic variables are stationary in their levels except for financial openness and institutional quality, for which we use the first difference in our regression analysis.

Table 2: Unit Root Test Results

Level 1st Difference _Stationarity

No Trend Trend No Trend Trend

Global Variables:

US 10 YR Real -1.207 -2.753 -4.293*** -4.207*** I(0)

VIX (Log) -2.721* -2.705 -2.910** -2.969 I(1)

World GDP Growth -3.124** -3.290* -4.234*** -4.118*** I(0)

Global Liquidity (% of GDP) -0.556 -2.602 -2.897** -2.814 I(1)

EPU (Log) -1.089 -2.375 -3.768** -3.773** I(1)

Domestic Variables:

Real Interest Rates -2.399*** -0.626 -4.192*** -1.770** I(0)

Inflation -3.880*** -2.090** -4.486*** -3.432*** I(0)

GDP Growth -2.723*** -1.257 -4.699*** -1.780** I(0)

REER Deviation -6.337*** -4.153*** -7.111*** -3.999*** I(0)

Financial Openness -1.619* -0.434 -3.915*** -3.144*** I(1)

Institutional Quality 2.209 2.205 -3.915*** -4.636*** I(1)

Notes: 1) The first panel shows the adjusted t statistics of the ADF tests for the cross-sectionally invariant (global) variables. The second panel, on the other hand, demonstrates the Zt_bar statistics of the Pesaran (2007) CIPS test for the cross-sectionally variant domestic variables. 2) Only 2 lags included in each test as the data frequency is annual and the time dimension is short. 3) The null hypothesis assumes that all series have a unit root. *p<0.1, **p<0.05, ***p<0.01.

(30)

25

results imply that including all variables together into the model will not result in multicollinearity problems16_.

5. Estimation Results and Discussion 5.1. Gross Capital Inflows

Table 3 presents the results of the estimated model given in equations (1) to (3) for gross (columns 1 to 4) and net (columns 5 to 8) capital inflows17_{. The first column provides the base model in} which we include the global and domestic variables, the second column includes candidacy and membership dummies, and the third and fourth columns further include interaction variables. Starting from the global factors, we see that the change in risk appetite (VIX), world economic growth and the change in global liquidity variables are found positive and significant. Specifically, a one percent increase in the VIX increases the gross capital inflows to GDP by approximately 6.4 percentage points. Although it is only significant at the 10 percent level, this finding is in line with Li et al. (2018). The impact of a one percentage point rise in the world economic growth rate on gross capital flows is around 3.8 percentage point increase which tallies with the results of previous studies (Baek, 2006; De Vita & Kyaw, 2008; Forbes & Warnock, 2012). Similar to the studies by Ghosh et al. (2014), Lim & Mohapatra (2016) and Yang et al. (2019), the change in global liquidity has a positive impact on gross inflows. Although the US long-term interest rate and the change in the European policy uncertainty (EPU) are found insignificant, these results support the findings by Rey (2015) who argues that gross flows are driven by global financial cycle.

Among domestic variables, we find that the impact of real domestic interest rates on gross inflows is negative which is contrary to higher return-seeking investor behaviour. However, this finding may also reflect that higher interest rates might be the result of the higher risks that the CEE countries face, which causes global investors to divert their investments. Although the impact is weak, higher domestic inflation is associated with lower gross inflows as expected. Similar to the

16_{Pearson correlation matrix of explanatory variables and results of the VIF analysis are presented in Table A4 and}

Table A5 in the Appendix.

17_{As there are several outliers in the sample that we discussed before, Cyprus and Malta are excluded from the gross}

(31)

26

world economic growth rate, the impact of domestic output growth on gross flows is significantly positive and the estimated coefficients are around 0.6 in different specifications. Real exchange rate overvaluation is also significantly associated with lower gross inflows, supporting the findings by Ghosh et al. (2014) and Li et al. (2018). As to the domestic institutional variables, we find evidence that financial openness has a strong positive effect on gross capital inflows, but the coefficient on institutional quality is found insignificant.

The second column extends the analysis by including the candidacy and membership dummies into the base model. Inclusion of these dummies does not change the sign and magnitude of the estimated coefficients of global and domestic factors considerably, however, we see that both dummies are insignificant. Third and fourth columns investigate how the EU integration process affected the impact of institutional factors on capital flows. However, once again, we do not find a significant relationship between EU integration and gross capital inflows in the CEE countries. It can be concluded that although gross capital inflows to the CEE countries increased with their EU integration as can be observed from the descriptive statistics in Table 1, this increase is mainly driven by a huge increase in gross capital inflows to Cyprus and Malta18_.

5.2. Net Capital Inflows

The columns 5 to 8 in Table 3 demonstrate the estimation results for net capital inflows. The impact of world output growth and change in global liquidity on net capital flows is still significant and positive. The VIX is also positively associated with net inflows in some specifications, but it is only significant at the 10 percent level. As to the domestic variables, real interest rates and inflation affect net capital flows negatively as in the gross inflow equations. Domestic GDP growth exerts a significantly positive impact on net capital flows. These results are also in line with the established literature (Koepke, 2019). On the other hand, REER deviations do not affect the net flows contrary to gross flows. Additionally, we see that financial openness and institutional quality have significantly and quantitatively strong positive impacts on net capital flows.

18_{When we use the full sample in the gross inflows regressions, most of the explanatory variables become insignificant}

(32)

27 Table 3: Estimation Results for Gross and Net Inflows

Gross Inflows / GDP Net Inflows / GDP

(1) (2) (3) (4) (5) (6) (7) (8)

US 10 YR Real -0.606 -0.463 -0.350 -0.540 0.299 -0.200 0.614* -0.386

(0.365) (0.567) (0.354) (0.510) (0.319) (0.436) (0.345) (0.433)

VIX (Log diff.) 6.424* 6.595* 6.612* 6.374* 3.172 4.400* 3.420 4.182*

(3.407) (3.361) (3.556) (3.209) (1.985) (2.150) (2.002) (2.194)

World GDP Growth 3.750*** 3.814*** 3.898*** 3.669*** 2.179*** 2.495*** 2.334*** 2.378***

(0.911) (0.859) (0.998) (0.861) (0.516) (0.535) (0.564) (0.528)

Global Liquidity (Diff.) 0.683* 0.688 0.679* 0.675 0.478*** 0.352** 0.474*** 0.334**

(0.385) (0.421) (0.360) (0.418) (0.152) (0.149) (0.134) (0.151)

EPU (Log diff.) -0.326 -0.270 0.014 -0.124 0.781 0.867 1.110 0.867

(1.965) (1.948) (2.092) (1.977) (1.319) (1.262) (1.256) (1.264)

Real Interest Rates -0.129* -0.139* -0.165** -0.119 -0.189*** -0.160*** -0.230*** -0.134**

(0.069) (0.073) (0.069) (0.074) (0.049) (0.054) (0.049) (0.052) Inflation -0.010* -0.011* -0.013** -0.012** -0.017*** -0.014*** -0.020*** -0.013*** (0.006) (0.005) (0.005) (0.006) (0.005) (0.004) (0.004) (0.004) GDP Growth 0.673*** 0.670*** 0.682*** 0.689*** 0.502*** 0.473*** 0.509*** 0.484*** (0.168) (0.171) (0.172) (0.182) (0.128) (0.130) (0.126) (0.134) REER Deviation -0.322*** -0.330*** -0.313** -0.308** -0.133 -0.148 -0.130 -0.134 (0.103) (0.106) (0.111) (0.107) (0.091) (0.098) (0.106) (0.098)

Financial Openness (Diff.) 14.737** 14.569** 14.602** 13.496* 10.042** 8.436** 9.831** 7.876*

(6.827) (6.482) (6.782) (6.368) (4.148) (3.809) (4.078) (3.829)

Institutional Quality (Diff.) 4.280 4.427 4.023 3.371 8.582*** 8.240*** 8.258*** 7.615***

(3.463) (3.524) (4.226) (4.162) (1.767) (1.733) (1.703) (1.767)

Candidacy 1.674 2.339 2.078 3.000*

(1.735) (1.968) (1.684) (1.580)

Membership -0.160 -1.906 -2.901** -4.048**

(1.606) (1.722) (1.123) (1.470)

Candidacy x Ins. Qual. 1.823 1.594

(2.764) (2.065)

Membership x Inst. Qual. 3.502 1.756

(3.375) (2.000)

Observations 311 311 311 311 348 348 348 348

R-squared 0.294 0.296 0.299 0.302 0.330 0.357 0.344 0.355

Wald Chi-squared 1167.6*** 1110.8*** 1051.4*** 1357.3*** 175.6*** 197.9*** 122.8*** 220.2***

Wooldridge F-Stat. 2.6 2.4 2.6 2.4 10.9*** 11.6*** 10.7*** 12.9***

(33)

28

The columns 6 to 8 of Table 3, which include EU integration dummies and their interactions, show that becoming an official EU candidate country is positively linked with net capital flows at the 10 percent significance level. Becoming a member country, on the other hand, is negatively associated with net capital flows. In addition, once again, we find no significant relationship between interaction variables and net capital flows.

5.3. Net Capital Inflows by Instruments

Table 4 presents the estimation results for net FDI, net portfolio and net other credit flows. Overall, we observe that the explanatory power and the number of significant variables decline considerably in the models that investigate the determinants of different types of flows. As can be seen from the first two columns, world GDP growth has a positive impact on net FDI inflows at the 1 percent significance level. The change in VIX and global liquidity also affect the net FDI inflows at the 10 percent significance level as in De Vita & Kyaw (2008) and Hannan (2017). However, similar to gross inflows, the US long-term interest rates and change in economic policy uncertainty in the EU variables are not found significant. Nevertheless, as Koepke (2019) pointed out, FDI flows might be less related to the global financial cycle because direct investments are largely driven by multi-national companies’ micro-level strategic decisions in line with the Dunning’s (1981) ownership, location and internalization (OLI) framework20_.

Among domestic variables, while GDP growth has a positive impact on net FDI; real interest rates, inflation environment and REER deviation are negatively correlated with net FDI inflows. Similarly, the change in financial openness and institutional quality variables positively affect net FDI inflows at the 10 percent significance level. Gaining EU candidacy status increases the FDI inflows by 1.7 percentage point at the 5 percent significance level. On the other hand, gaining membership status significantly decreases the net FDI inflows to the CEE countries by 1.9 percentage point. However, candidacy and membership interactions with institutional quality are

20_{According to this approach, the multinational companies’ decision to engage in FDI depends on whether the firm}

(34)

29

not found significant. These results partially support the findings of Clausing & Dorobantu (2005) and Bandelj (2010) in terms of the beneficial impact of EU integration on FDI inflows.

Table 4: Estimation Results by Different Instruments

FDI / GDP Portfolio / GDP Other Flows / GDP

US 10 YR Real 0.283 -0.403 -0.054 -0.395* 0.311 0.278

(0.254) (0.259) (0.129) (0.203) (0.185) (0.327)

VIX (Log diff.) 1.085 1.718* -0.904 -0.388 2.583 2.375

(1.001) (0.947) (1.052) (1.288) (1.665) (1.817)

World GDP Growth 0.778** 0.900*** -0.221 -0.049 1.611** 1.417**

(0.306) (0.280) (0.291) (0.252) (0.624) (0.552)

Global Liquidity (Diff.) 0.161* 0.050 -0.060 -0.132 0.443** 0.458**

(0.088) (0.086) (0.094) (0.092) (0.166) (0.173)

EPU (Log diff.) 0.002 -0.030 -0.877 -0.816 1.269 1.129

(0.603) (0.591) (1.202) (1.186) (2.144) (2.099)

Real Interest Rates -0.076*** -0.031 -0.058* -0.051* -0.104** -0.071

(0.025) (0.024) (0.029) (0.025) (0.040) (0.047) Inflation -0.010*** -0.006** -0.007** -0.005** -0.004 -0.004 (0.002) (0.002) (0.002) (0.002) (0.003) (0.004) GDP Growth 0.098* 0.081 -0.013 -0.026 0.459*** 0.463*** (0.049) (0.047) (0.077) (0.076) (0.133) (0.136) REER Deviation -0.085** -0.090* -0.016 -0.020 -0.042 -0.044 (0.038) (0.042) (0.062) (0.059) (0.121) (0.114)

Financial Openness (Diff.) 9.329* 7.830 -0.799 -1.511 2.211 1.409

(4.705) (4.483) (1.833) (2.003) (2.397) (1.916)

Institutional Quality (Diff.) 1.909* 1.473 3.834** 3.841** 1.952 1.601

(0.913) (0.861) (1.360) (1.386) (1.662) (1.887)

Candidacy 1.725** -0.179 1.143

(0.801) (0.829) (0.933)

Candidacy x Ins. Qual. -0.302 -0.627 1.885

(0.659) (1.517) (1.220)

Membership -1.851*** -0.662 -1.446

(0.485) (0.658) (1.145)

Membership x Inst. Qual. -0.516 -1.294 2.965**

(0.579) (1.341) (1.155)

Observations 310 310 304 304 311 311

R-squared 0.203 0.216 0.036 0.052 0.237 0.245

Wald Chi-squared 1263*** 2525*** 4495*** 3365*** 322*** 385***

Wooldridge F-Stat. 1.6 1.8 4.6* 6.2** 6.7** 7.5**

Notes: 1) Cyprus and Malta are excluded from the sample due to outliers. 2) Domestic factors are lagged one year to control the possible endogeneity. 3) Country-specific fixed effects and a dummy variable representing global financial crisis included in all models. 4) In line with the Modified Wald test for heteroscedasticity results, standard errors are clustered by countries. 5) Standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01

(35)

30

impact on net portfolio investment flows. As to the variable of interest, we find no significant evidence on the relationship between EU integration and portfolio investment flows.

The results in the last two columns of Table 4 indicate that both global economic growth and global liquidity significantly increase the net other credit inflows to the CEE countries. Additionally, while domestic GDP growth has a significantly positive impact on net other flows, domestic real interest rates are negatively correlated with net other flows. Although being a candidate country is not found statistically significant, the interaction of membership and institutional quality is significantly positive. Thus, in the countries which have a higher level of institutional quality, other credit inflows are positively affected by the membership.

5.4. Robustness Checks

(36)

31

As a second robustness check, we consider that capital flows may exhibit a dynamic behaviour because of the market experience, investment lags and persistency in investment decisions. The estimation results from the Arellano & Bover (1995) and Blundell & Bond (1998) system GMM estimator are shown in Table A9. The results indicate that the flows in the previous year have a significantly positive impact on gross inflows and net other flows, while negatively associated with net portfolio investment flows. Among global factors, world output growth is an important determinant explaining gross flows, net flows and net FDI flows. The change in the global risk appetite and global liquidity variables are only significantly positive for the net capital flows. Global liquidity variable is also significantly positive in the net other flows specification. As to domestic factors, real interest rates have a significantly negative impact on gross and net capital inflows, while inflation is negatively correlated with the net capital inflows. Domestic GDP growth positively affects gross, net and FDI flows to the CEE countries as well as the other type of flows. While institutional quality is only significant and positive for net flows, financial openness has a positive impact on gross, net and FDI inflows. Although we found no significant evidence on the relationship between candidacy and any type of flows, the membership dummy is significantly negative for net FDI, net portfolio flows and total net capital flows. The results for global and domestic variables are broadly in line with the base results, except for VIX, global liquidity, real domestic interest rates and REER deviation variables lose their significance for gross, net FDI and net portfolio inflows. As to the EU integration, the positive and significant coefficient estimation on the candidacy dummy in net flows and net FDI inflows becomes insignificant in the dynamic specification. Membership dummy, on the other hand, is still significant and negative for the net total and net FDI inflows and also significantly negative for net other flows in the GMM estimation results. These results can be interpreted as the overall negative impact of membership is robust, while the positive impact of candidacy is not in the dynamic specification.

(37)

32

flows similar to the results of base specifications. As to the domestic variables, real interest rates are negatively associated with gross flows, net flows and other flows similar to the main findings, while inflation is negatively associated with each type of flows except for net other flows. Likewise, domestic GDP growth is also correlated with gross flows, net flows and net other flows. REER deviation is also negatively correlated with gross flows and net FDI flows similar to the main results. Contrary to main results, financial openness is only significantly positive for the gross flows, while the impact of institutional quality is significantly positive for net flows, net FDI flows and net portfolio flows. As for the EU integration, the candidacy dummy only has a positive impact on net FDI inflows at the 10 percent significance level. On the other hand, membership has a significantly negative impact on net flows and net FDI inflows, which are similar to the main findings. Overall, the results for both global and domestic variables as well as EU integration dummies are found robust to excluding non-communist origin countries.

6. Conclusion

The CEE countries have experienced a substantial political and economic transformation in recent decades. The EU accession process has served as an anchor for their transition from centrally planned to market economies. Foreign capital flows have been a key tool for this transformation both by directly providing necessary funds to improve human and physical capital within the region and by indirectly inducing improvements in domestic institutions and transferring technology and know-how. On the other hand, the increasing reliance on foreign capital also gave rise to external vulnerabilities and the CEE region has suffered most from the GFC when cross-border capital flows plunged.

(38)

33

performed different sensitivity analyses including splitting the sample, utilizing the Arellano & Bover (1995) and Blundell & Bond (1998) system GMM estimator and excluding countries which do not have a communist history.

Overall, the base estimation results for the global and domestic control variables are in line with the literature. Although the magnitude of the impact of world output growth on capital flows is higher in gross flows, the impact of global financial conditions captured by the change in global liquidity is significantly positive for net flows. The US long-term interest rates and European policy uncertainty are not found significant in both gross and net flows specifications. The impacts of global risk sentiment on gross and net flows are positive contrary to our expectations, but this result is only weakly significant and not robust to different specifications and sensitivity analyses. Moreover, domestic factors such as real interest rates, inflation and GDP growth are important for both gross and net flows. REER deviation from trend is negatively associated only with gross flows. Additionally, while the financial openness is found essential for attracting foreign capital both in gross and net flows, institutional quality only affects net flows.

(39)

34

Among different type of flows, our findings indicate that net FDI inflows are affected positively by being an EU candidate member, while the impact of actual EU membership on net FDI inflows is negative. Although we do not find any significant relationship between EU integration and net portfolio flows, the positive impact of the EU membership on net other flows has been through its interaction with the domestic institutional quality. In addition to these base findings, we also find that the EU candidacy had a strong positive impact on net FDI inflows in both pre- and post-crisis periods. Similar to gross flows, although the EU membership dummy is significantly positive for net other flows before the GFC, this impact has become negative since the post-crisis period. These results are also robust to the dynamic model specification and to excluding countries which have the non-communist historical background.

(40)

35

the EU may have reversed the direction of capital flows as investors from the CEE headed towards safe havens after the GFC.

Capital Flows to Central and Eastern Europe: The Impact of the EU Accession Process

Capital Flows to Central and Eastern Europe:

The Impact of the EU Accession Process

Master’s Thesis

MSc Economic Development & Globalization

Ahmet Ihsan Kaya - S4065077

a.i.kaya@student.rug.nl

Supervisor: Prof. Dr. Jakob de Haan

Co-assessor: Dr. Andreas C. Steiner

Capital Flows to Central and Eastern Europe:

The Impact of the EU Accession Process

Abstract

Table of Contents