International capital flows and sudden tops : political and institutional determinants

(1)

UNIVERSITY OF AMSTERDAM

International capital flows and sudden tops

Political and institutional determinants

Jiewei (Derek) Geng

Supervisor: Prof. dr. S.J.G. (Sweder) van Wijnbergen Date: 15/07/2016

Abstract: This paper examines political and institutional determinants of capital

flows on the equity fund level and sudden stops of equity fund flows. Employing fixed effects model and probit model for monthly data of 70 countries from January 2000 to July 2015, the paper finds that together with economic factors, political factors also contributes to determining capital flows and sudden stops. Lower political risk drives more capital inflows. In the meantime, countries with low political risk are more easily exposed to sudden stops. As a range of indicators for political risk, government stability, investment profile, socioeconomic condition, corruption, and democracy have heterogeneous effects on capital flows and sudden stops across countries.

JEL classifications: C33, F32, F33, G23

(2)

Statement of Originality

This document is written by Student [Jiewei Geng] who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1. Introduction

International capital flows have been growing substantially across the countries since the 21st century, especially the portfolio capital flows on the fund level. According to EPFR Global, the total assets under management of funds increase enormously from less than 0.15 trillion US dollars in 2000 to more than 25 trillion US dollars in 2015. Unlike foreign direct investments (FDI) flows, capital flows on the fund level is considered as short-term investments in bond and equity markets such as mutual funds, exchange-traded funds (ETFs) and hedge funds (Li et al., 2015). It is the net of all cross-border cash inflows and outflows in and out of various financial assets. In comparison to FDI flows, fund flows are widely used by analysts and institutions to gauge investor sentiment within the markets. It is more volatile and easily signals broad-based optimism or pessimism over the whole markets.

Due to the increasing amount of capital flows across countries, the volatility of capital flows has also become much higher. The high volatility of capital flows can lead to extreme scenarios such as sudden stops. A sudden stop is defined as “an abrupt and major reduction in capital inflows to a country that has been receiving large volumes of foreign capital” Edwards (2004, p. 59). Sudden stops are associated with several financial and economic disruptions, such as currency depreciation, higher costs of external finance (Calvo et al., 2004), and even banking and currency crises (Calderón and Kubota, 2013). Consequently, the economic growth of the country will drop. United Kingdom withdrawal from the European Union (Brexit) can be considered as the most recent sudden stops scenario caused by political risk. On Friday 24th, 2016, the British pound fell to $1.3228 - the lowest level since 1985. More than £100 billion wiped off London's FTSE 100.

Some studies (cf. Forbes and Warnock, 2012; Fratzscher, 2012; Eichengreen and Gupta, 2016) investigate the relationship between capital flows and economic factors. Both global economic factors such as global liquidity and risk, as well as domestic economic factors such as industrial production growth and equity return are found to have significant impacts on capital flows and sudden stops of capital flows.

Theoretically, political and institutional determinants also affect capital flows because low-quality institutions can have a negative effect on economic growth and lower

(5)

political stability can lead to a higher growth volatility. Moreover, higher political risk can lower the expected return of investors which might result in withdrawing capital investments from the country while a well-regulated and stable political environment would strengthen investors’ belief and attract more capital inflows. However, countries with stable and promising political environment could also be more exposed to sudden stops of capital flows because these countries usually have sound regulations and high level of trade freedom. Institutions and individuals can invest or withdraw capital more easily. So far, several empirical studies have examined the influences of capital flows on institutional quality. For instance, Papaioannou (2009) concludes that well-functioning institutions are a key driving force for international bank flows. Alonso (2015) finds that countries with better governance and public sector credibility tend to attract more FDI flows.

Most of the studies have examined either political or economic factors on FDI flows. This paper focus on the linkage between political determinants and net equity fund flows. In addition, a wider range of indicators for political risk, i.e. government stability, socioeconomic condition, investment profile, corruption, and democracy are used to identify the relative importance of these indicators for fund flows. Lastly, political factors are also examined whether they drive the sudden stops of fund flows and the impacts differences across income groups or regions.

By comparison with FDI flows data on the quarterly or yearly basis, fund flows data is reported on the monthly basis which could give better insights of which factors drive fund flows and sudden stops of fund flows. Using equity fund flows data from EPFR Global database for 70 countries, including 31 developed and 39 developing countries, from January 2000 to July 2015, the main results in this paper suggest that besides economic factors, political and institutional determinants, especially government stability is a key driving force for equity fund inflows. This finding is relatively significant in developing countries. Moreover, a lower political risk is associated with higher likelihood of sudden stops. More specifically, most of the political factor components are significantly related to the sudden stops. A stable government with low corruption and satisfactory investment profile increase the probability of sudden stops while a strong socioeconomic condition decreases the probability of sudden stops. These findings are more significant in developed countries than in developing ones. To summarize, a stable political environment with

(6)

high quality of institutions drives more capital inflows in the country. In the meantime, it also increases the probability of sudden stops.

The contribution of this paper is twofold. First, it finds that political risk is a key force for driving equity fund flows. A lower political risk is associated with more equity fund inflows. Second, the paper provides new evidence that countries with low political risk are also more exposed to sudden stops. Additionally, in terms of policy implications, a country can implement monetary policies such as forward guidance to increase consumer confidence or fiscal policies to reduce unemployment. With a stronger socioeconomic condition, a domestic country can reduce their exposure to sudden stops.

This paper is organized as follows. The next section reviews some relevant literature regarding the role of institutions and why it is relevant for capital flows. In addition, the literature in determining sudden stops are also discussed. Section III presents data description and methodologies for sudden stop identifications and regression models used in this paper. Section IV reports the empirical results for determining capital flows and sudden stops. The final section concludes.

2. Literature

2.1 Why institutions matter

The role of institutions has received considerable attention in terms of economic research in recent decades. Numerous studies have provided evidence that the differences in institutional quality could have a significant effect on output per capita, such as Knack S, Keefer P (1995), Mauro, P. (1995) and Rodrik et al. (2004). Mauro, P. (1995) concludes that corruption lowers private investment, thereby reducing economic growth, even in countries in which bureaucratic regulations are cumbersome. Moreover, the role of institutions in financial development has also been assessed, especially on the effects of regulatory and legal environment on the functioning of financial markets. A legal and well-regulated system including property rights protection and contracts enforcement is highly essential for financial development. La Porta et al. (1997, 1998), one of the most notable studies, argue that the origins of the legal code substantially influence the treatment of shareholders,

(7)

creditors and the efficiency of contract enforcement. The conclusion is that poor shareholder rights are associated with weakly developed equity markets. In contrast, common law countries with high levels of shareholder rights correspond with stronger equity market development. In addition, they also find that greater creditor rights are positively related to financial intermediary development. Rajan and Zingales (2003) emphasize the role of politics (protectionism, lobbying) in financial development. They question the link between legal origins and cross-country differences in financial development and instead pointed out the crucial role of political forces in implementing policies toward financial markets and their development. Law et al. (2012) examine the effect of institutional quality on financial development in developed and developing countries by using banking sector and stock market development indicators. Based on dynamic system generalized method of moments (GMM) estimations, the results demonstrate that a high-quality institutional environment is essential in explaining financial development, specifically for the banking sector.

2.2 Why institutions are relevant for capital flows

Capital flows of a country can be strongly affected by both domestic and foreign investors. The behavior of investors is not only dependent on economic determinants but also on political determinants of the country. First, low-quality institutions are associated with poor economic performance. Not only do they have a negative effect on economic growth, but also lead to a higher growth volatility (Acemoglu et al., 2003). Second, greater political instability could lower investors’ expected return. Once the expected return is lower, investors will tend to withdraw their investments and shift to other alternative countries. As a result, the net capital flows will drop and even become significantly negative. Perotti and van Oijen (2001) use ICRG political risk data as a proxy to measure political risk and institutional quality. The results demonstrate that political instability is followed by lower stock returns in emerging economies. Third, a transparent and well-regulated system mitigates information and monitoring costs. The friction of the asset trade activities can be alleviated to a large extent (Papaioannou, 2009).

Over the last few decades, despite the surge in capital mobility across countries, capital flows from rich to poor countries have been at much lower levels than predicted by the standard neoclassical models. In the neoclassical model, countries

(8)

produce the same goods with the same constant returns to scale production function, same technology and same factors of production. Developing countries generally have a lower level of capital per worker. The scarcity of capital relative to labor indicates that the return to capital is higher than in developing countries. As a response, investors in developed countries will invest more capital in developing countries due to the effect of diminishing return of capital and this will continue until the returns of capital in all the countries are equalized. Given by the implications of the neoclassical theory, Lucas (1990) examines this question empirically and concludes that capital flows from rich to poor countries are much lower than predicted which constitutes the “Lucas Paradox”.1_{The main theoretical explanations for the “Lucas Paradox” can be}

grouped into two categories: i) differences in fundamentals that affect the production structure of the economy, i.e. institutions, government policies, and technology differences. ii) international capital market imperfections, mainly sovereign risk and asymmetric information.2

However, little theory directly links the capital flows with political and institutional determinants. The most related theoretical framework is constructed by Shleifer and Wolfenzon (2002). They build an agency model in which an entrepreneur tries raise equity finance for a project. The entrepreneur maximizes her personal wealth, which is a function of the fraction of the how much equity she decides to sell and how big the project to undertake. Assuming that the entrepreneur has the full control of the project after the initial share offering. This entrepreneur operates in an environment with a limited legal protection of outside shareholders, and also has an opportunity to divert some of the profits of the firm. The profit diversion depends on the efficacy of the legal system and becomes costly with well-defined and protected investor’s rights. Both domestic and foreign investors can anticipate the probability of diversion and are thus unwilling to invest in the low quality institutional environment. As a consequence, capital will not flow to countries with low levels of investor protection. Similarly, capital will flow out of the country once the political risk increases. This study has shown a strong causal effect of legal system effectiveness indicators on the capital flows activities.

1_{See Obstfeld and Rogoff (2000) for an overview of the major puzzles in international economies}

including “Lucas paradox”.

(9)

Several empirical studies so far have confirmed the results of Shleifer and Wolfenzon (2002) and explain a significant part of Lucas (1990) paradox of why capital does not flow from rich to poor countries. Alfaro et al. (2007) find that institutional quality is an important determinant of capital flows. The evidence shows that the historical legal origin of a country has a direct impact on capital inflows. Also, both low institutional quality and bad monetary policies, play a role in explaining the long-run volatility of capital flows during 1970–2000. (Papaioannou, 2009) examines the institutional quality on international bank flows by using a gravity model and IV estimates to control for endogeneity. The results show that well-functioning institutions are a key driving force for international bank flows. Specifically, foreign banks invest substantially more in countries with i) uncorrupt bureaucracies, ii) high-quality legal system, and iii) a non-government controlled banking system. Alonso (2015) tests institutional quality relevance for gross capital flows using a panel of 56 countries, differentiating between high-income and low-income economies, over the period 1996-2012. The results demonstrate that institutional quality is a significant factor influencing the behavior of both foreign and domestic investors. Countries with better governance and public sector credibility tend to attract more flows.

All the aforementioned literature mainly focus on whether institutional quality affect the level of capital flows but none of them have looked into on how strongly political risk and institutional determinants influence the net capital flows, in other words, whether political and institutional determinants have an influence on the sudden stops in capital flows.

2.3 Identification and determinants of sudden stops

There are two major methods defining sudden stops. The first mainly focuses on the change of capital flows. Calvo et al. (2004) define sudden stop as a phase that meets the following conditions: i) it contains at least one observation where the year-on-year fall in capital flows lies at least two standard deviations below its sample mean. ii) the phase ends once the annual change in capital flows exceeds one standard deviation below its sample mean. iii) the phase of the sudden stop-start is identified by the first time the annual change in capital flows falls one standard deviation below the sample mean. The second method is a based on both the change of capital flows and its relative size to GDP. Guidotti et al. (2004) identify a sudden stop period when the

(10)

capital account contraction is larger than one standard deviation below its sample mean and the capital account contraction exceeds 5% of GDP.

Some studies have examined the determinants of sudden stops. (cf. Calvo et al., 2004; Bordo et al., 2010, Forbes and Warnock, 2012; Calderón and Kubota, 2013). Calvo et al. (2004) demonstrate that large real exchange rate (RER) fluctuations are related to sudden stops are basically an emerging economy phenomenon because they can be forced to large adjustments in the absorption of tradable goods. Edwards (2004) finds no empirical evidence that countries with higher capital mobility have a higher incidence of crises or tend to face a higher probability of having a crisis, than countries with lower mobility. Forbes and Warnock (2012) find that global macroeconomic factors, especially global risk are significantly associated with sudden stops of capital flows while domestic macroeconomic characteristics are generally less important. Contagion factors through geography or trade are also associated with extreme capital flow episodes. Eichengreen and Gupta (2016) analyze the sudden stops in portfolio flows to emerging markets since 1991. Although the results show that frequency and duration of sudden stops have remained largely unchanged, the relative importance of different factors in their incidence has changed. In particular, global factors appear to have become more important relative to domestic factors, Fratzscher (2012) employs a factor model together with a dataset of high-frequency portfolio capital flows on the fund level which is the same as this paper uses. The results suggest that key crisis events, as well as changes to global liquidity and risk, exert a large effect on capital flows both in the most recent financial crisis and its recovery.

3.Data and Methodology

3.1 Capital flows

Some papers use FDI as capital flows and some focus on portfolio flows. Eichengreen and Gupta (2016) compare FDI and portfolio flows and demonstrate that FDI flows are relatively stable. This paper uses portfolio capital flows at the fund level from EPFR Global which tracks assets allocations of a large number of international funds.

(11)

3

The EPFR data contains weekly and monthly flows with more than 16,000 equity funds and more than 8,000 bond funds and equity funds account for the biggest share of total assets under management. Due to the availability of data, this paper will focus on equity fund flows only. The time horizon is from January 2000 to July 2015 in monthly frequency and the country sample is 70, including 31 developed countries and 39 developing countries. 4 According to Jotikasthira et al. (2010), although EPFR data only covers about 10 to 20% of the market capitalization for most countries, it is a fairly representative sample. In their paper, the point is convincible by showing a highly correlated match between EPFR portfolios and portfolios flows from total balance-of-payments data. Moreover, EPFR data provides both net capital flows and change in total assets under management each period. The data is in high frequency which can provide better insights of how net capital flows change for each period. Lastly, comparing with other data source for funds, it does not only cover most advanced countries but also includes a large number of developing countries.

Figure 1 provides an overview of equity fund flows in magnitude from January 2000 to July 2015 between developed countries and developing countries. Clearly, during the period of 2000 to 2005, the magnitudes of capital flows are relatively small across countries. Starting from the year 2005, possibly due to the trend of globalization, the size of capital flows becomes larger and the volatility of capital flows become much higher than before. Furthermore, there are several plunges in the equity fund flows in the year 2007 and 2008 for developed countries during the period of financial crisis. In the year of 2011, some large decreases are also observed for developed countries possibly due to the sovereign debt crisis. In comparison to developing countries, the magnitudes of capital flows in developed countries are much larger.

Figure 2 demonstrates the equity fund flows scaled by total assets under management(AUM). By comparison with figure 1, figure 2 focuses on the scaled percentage change. In Figure 1, from the year 2000 to 2005, the magnitude of equity fund flows is small while figure 2 still shows the significant fluctuation of equity fund flows during this period. Especially in March 2001, owing to the internet bubble burst, the net equity fund flows in percentage both for developing and developed

3_{These papers also use the EPFR data(cf. Jotikasthira et al., 2010, Fratzscher, 2012, Li et al., 2015 and}

Puy, 2016)

(12)

countries experienced plummets. For developed countries, the net equity fund flows for each dropped more than 4% on average, and in developing countries, it went down by more than 3% for each country. In general, both developed and developing countries illustrate strong co-movement in terms of equity fund flows. Furthermore,

Figure 1

(13)

starting from 2003, possibly due to the trend of globalization and more financial openness in developing countries, the equity fund flows percentage movement in developing countries has become more volatile than in developed countries. This bigger fluctuation can be explained by higher returns and more investment opportunities in developing countries while there also existing more economic and political uncertainties. When negative economic shocks occur, even though the magnitude of capital outflows in developed countries is larger, the percentage changes, in other words, the volatility of capital flows in developing countries are higher.

Figure 3 describes equity fund flows moves with the classification of three geographic regions; Americas, Europe, the middle East and Africa (EMEA) and Asia-Pacific. From the year 2000 to 2005, the capital flows in these three regions seem to follow the same trend and relatively more stable. Starting from 2005, the magnitude of equity fund flows become more volatile in the region of Americas, especially in the year 2008 and 2013, possibly due to the financial crisis and Fed’s quantitative easing (QE) respectively. Moreover, the equity fund flows in EMEA and Asia-Pacific areas seems to be positively correlated, especially after the financial crisis when the trend of Americas diverges from these two.

(14)

Figure 4 depicts the equity fund flows scaled by total assets under management(AUM) in these three regions. Among all of them, EMEA area is the most volatile in terms of equity fund flows by scaling. During the biggest three shock periods, internet bubble, financial crisis, and sovereign debt crisis, EMEA area experienced the most dramatic decreases in equity fund flows by scaling comparing with the other regions, Americas, and Asia-Pacific. In September 2001, Americas also experiences a plunge in equity fund flows under scaling. This phenomenon can be explained by the 9/11 attacks.

Figure 4

3.2 Identifications of sudden stops

In this paper, in order to observe how strongly political risk and institutional determinants influence fund flows, two methods are used to identify the sudden stops of fund flows. First, based on the method of Calvo et al. (2004), the start of one sudden stop period is defined when the annual change of equity fund flows is one standard deviation below the sample mean and eventually reaches two standard deviations below the sample mean. The sudden stop period ends when the year-on-year change reaches one standard deviation below the mean. Second, as an alternative and robustness check, the method of Guidotti et al. (2004) is also implemented. The sudden stop period is identified based on two criteria: i) the annual changes of equity

(15)

fund flows are one standard deviation below the mean. ii) the fund flows scaled by asset under management (AUM) is less than 0%.

The equity fund flows scaled by asset under management (AUM) is labeled as 𝑠𝑡 and

the monthly equity fund flows are labeled as 𝑞_𝑡. Following Forbes and Warnock (2012), the paper also defines 𝐶𝑡 as the 12-months moving sum of lagged values in

equity fund flows so that the seasonal influences can be adjusted. Moreover, the year– over-year changes in 𝐶𝑡, labeled as ∆(𝐶𝑡) is computed as follows:

𝐶𝑡= ∑ 𝑞𝑡−𝑖 12

𝑖=0

, 𝑡 = 1,2, … , 𝑛. ∆(𝐶𝑡) = 𝐶𝑡− 𝐶𝑡−12, 𝑡 = 13,14, … , 𝑛.

The rolling means µ𝑡 and standard deviations σ𝑡 of year-over-year changes ∆(𝐶𝑡) are

calculated over the previous 24 months fund flows data. If ∆(𝐶_𝑡) is one standard deviation σ𝑡 below the rolling mean µ𝑡 in time period t and it eventually reaches two

standard deviations below the rolling mean after time t, this period is identified as sudden stop period and the period ends when ∆(𝐶𝑡) reaches one standard deviation

below the mean. Quantitatively,

If ∆(𝐶_𝑡) ˂ µ_𝑡 - σ_𝑡

If ∆(𝐶_𝑡+𝑛1) ˂ µ_𝑡+𝑛1 - 2 ∗ σ_𝑡+𝑛1 Then sudden stop period ends when

∆(𝐶_𝑡+𝑛2) > µ_𝑡+𝑛2 - σ_𝑡+𝑛2

Then the period from t to t+ n2 is identified as a sudden stop episode. The number of

sudden stops during this period is n2 . The second method is similar only that the

second condition become whether equity fund flows scaled by AUM as 𝑠𝑡 is below

0%. The first method focuses on the decline of net capital flows within a certain sudden stop episode and the second method capture the sudden stops on the specific month t.

Based on the sample of 70 countries from February 2000 to July 2015, Figure 5 illustrates the sudden stops by applying both methods and tells that the results of both

(16)

methods are highly similar. Under the total sample of 13020 panel observations, 2518 sudden stops are identified following the method of Calvo et al. (2004) and 1813 sudden stops are identified applying the method of Guidotti et al. (2004) , 19% and 14% respectively. For the latter empirical analysis, method 1 stemming from Calvo et al. (2004) will be used.

In total, five waves of sudden stops are observed in Figure 5. The first wave is from 2000 to 2001, possibly due to the reason of internet bubble crash and 9/11 attacks. The second wave at the end of 2004 and beginning of 2005 is relatively shorter and fewer countries are involved. The third wave is from 2007 to 2009 which is the longest because of the financial crisis. The fourth wave from 2011-2012 which might result from the sovereign debt crisis and the fifth wave is 2014-2015.

Figure 5

Figure 6 shows that there are high co-movements both across developed countries and developing countries for the sudden stops of capital flows. In the wave of 2004-2005 and 2011-2012, developing countries experienced more sudden stops than developed countries. Moreover, in the waves of 2001-2002 and 2014-2015, developed countries

(17)

experienced more sudden stops. In the wave of the financial crisis, all countries experienced a large number of sudden stops of capital flows. 5

Figure 6

3.3 Political risk and institutional quality

The proxies of political risk and institutional determinants are constructed by Political Risk Services (PRS), namely the International Country Risk Guide (ICRG) "political risk" rating. It is considered by institutional investors such as banks and insurance when they make asset allocation decisions. One of the biggest advantages is that unlike most institutional measures which only contains cross-sectional data or with limited time variability, the ICRG monthly data exhibits substantial “within” variation which controls for unobserved country heterogeneity and time-invariant omitted variables (Papaioannou, 2009).

The “Political risk” rating has long been used by empirical macro literature and it has been employed recently to analyze investment patterns. Gelos and Wei (2002) use the data to explain the portfolio allocation choice of emerging market funds and Papaioannou (2009) employs the data to investigate the impact on inter-bank flows. The Political Risk index is based on 100 points, ranging from 0 denoting minimum

5_{In appendix 1, it provides summary statistics of fund flows sudden stops for both methods under 70}

(18)

level of institutional quality to 100 indicating a total absence of political risk. In other words, the higher the points in political risk index, the less political risk (more political stability) the country has. The PRS suggest that "...the aim of the political

risk rating is to provide a means of assessing the political stability of the countries covered by ICRG on a comparable basis." The “Political risk” rating composites 12

weighted variables covering both political and social attributes. (1) Government Stability (16%), (2) Socioeconomic Conditions (16%), (3) Investment Profile (16%), (4) Internal Conflict (16%), (5) External Conflict (16%), (6) Corruption (8%), (7) Military in Politics (8%), (8) Religion in politics (8%), (9) Law and Order (8%), (10) Ethnic Tensions (8%), (11) Democratic Accountability (8%), (12) Bureaucracy Quality (4%).

In this paper, first, the political risk index will be used to represent political and institutional determinants. Second, as shown below, five subcomponents with crucial weights in the “Political risk” rating index will be examined in order to observe which specific factor has a significant impact on the capital flows and their sudden stops. Government stability (ranges from 0 to 12): this is an assessment both of the government’s ability to carry out its declared programs and its ability to stay in office. government unity, legislative strength, and popular support are considered for the evaluation of government stability. On the one hand, a stable government can strengthen the faith of foreign investors while an unstable government poses a threat to foreign investments which might result in sudden stops. On the other hand, a less stable government could lead to more policy and legislation changes which could be considered as opportunities for investors. A score of 0 equates to very low stability and a score of 12 equates to very high stability.

Socioeconomic conditions (ranges from 0 to 12): this component assesses the socioeconomic pressures at work in the society that could constrain government action or fuel social dissatisfaction. Unemployment, consumer confidence and poverty are taken into account for evaluation of the socioeconomic condition. Strong consumer confidence and low unemployment can boost the economic growth and reduce the probability of sudden stops of capital flows. A score of 0 equates to very low socioeconomic condition and a score of 12 equates to the very high condition.

(19)

Investment Profile (ranges from 0 to 12): this variable reflects the risk to investment that is not covered by other economic or financial risk components. Contract viability/expropriation, profits repatriation and payment delays are included in this factor. A country with good investment profile has a better guarantee of investment returns and give stronger confidence to foreign investors. A score of 0 equates to very high risk and a score of 12 equates to very low risk.

Corruption (ranges from 0 to 6): the most common form of corruption in the business process is the demands for special payments and bribes connected with relevant licenses, tax assessments, and protection. On the one hand, high level of corruption reduces the effectiveness of business and even may force the withdrawal or withholding of an investment. It is considered as a threat to foreign investment for several reasons: firstly, it distorts the economic and financial environment. Secondly, it increases the friction costs and thus reduces the efficiency of governments and businesses. On the other hand, corruption can attract investment because, by bribing, investors can have greater opportunities for taking projects with a higher return. A score of 0 equates to a high level of corruption and a score of 6 equates to a low level of corruption.

Democracy (ranges from 0 to 6): this is a measure how responsive government is to its people. On the one hand, democracy may have a higher tax burden which distorts investment returns due to their propensity toward compulsory redistribution. On the other hand, democracy can ensure political stability and protection of property right which leads to a greater predictability of investment returns. A low score of 0 equates autocracy and a high score of 6 equates democracy.

(20)

Figure 7 provides an overview of political risk index and its subcomponents movement from 2000 to 2015 under the sample of 70 countries. First, political risk index increases sharply in the beginning of 2001 and it dropped down considerably in the following months possibly owing to the global political tension caused by the 9/11 attacks. From 2003 onwards, the political risk index decreases gradually which indicates that political risk has become more critical during the last decade. Second, government stability experiences a marginally decreasing trend with fluctuation for both developed and developing economies. Compared with developed economies, the governments are relatively more stable in developing countries. Third, for both socioeconomic condition and investment profile, developed countries have a higher rating. Moreover, the rating for both indices become lower since the financial crisis in 2007/2008, especially for developed countries. Fourth, corruption and democracy indices are relatively stable, by comparison with the other political factors.

3.4 Economic factors (control variables)

According to the relevant literature of capital flows and their sudden stops (Calvo et al., 2008; Forbes and Warnock, 2012; Fratzscher, 2012; Calderón and Kubota, 2013; Eichengreen and Gupta, 2016), all the economic factors are considered and they are divided into three categories: global, domestic and contagion factors.

(21)

Global macroeconomic factors capture the external conditions which potentially have effects over the whole world. In other papers such as Fratzscher (2012), global factors are also referred as push factors. The global factors used in this paper are global short-term interest rates, global industrial production growth, global equity return,TED spread, Chicago Board Options Exchange volatility index (VIX) and commodity price shock.

Following Puy(2016), global short-term interests, global industrial production growth and global equity return are calculated by the non-weighted average of these three indices in the US, UK, and Japan. These three major macroeconomic factors can describe how the global economy performs in the short term. Moreover, in line with Fratzscher (2012), the VIX is used as a proxy of global risk aversion and the TED spread,which is the difference between the month LIBOR rate and the three-month treasury bill as a measure of global liquidity risk. In order to capture the impact of the indices’ changes on capital flows, the first difference of VIX and TED spread are used in this paper. If there is a sharp increase in volatility or a liquidity squeeze, capital will flow into safer assets to avoid potential losses. According to Ghosh et al. (2014), commodity prices also have effects on capital flows and this paper uses the same proxy Standard & Poor Goldman Sachs Commodity Index (GSCI) to capture the global commodity price movements.

Domestics macroeconomic variables are considered as follows: domestic industrial production growth, domestic equity return, short-term interest rate, CPI inflation, international foreign reserves and real effective exchange rate.

Country-specific macroeconomic characteristics are captured by domestic factors which are also referred as pull factors in other literature. This paper includes domestic industrial production growth and domestic equity return as main economic performance for each country following Puy(2016). High industrial production growth and high equity return will attract capital inflows which reduce the likelihood of sudden stops. Furthermore, the short-term interest rate is also used because capital flows are sensitive to interest differentials between countries. International foreign reserves (as % of GDP) are also included because enough liquidity can prevent sudden stops and reduce the drop of output growth (Fratzscher, 2012). In addition, as a proxy of monetary stability, CPI inflation is added as another control variable in

(22)

domestic factors (Calderón and Kubota, 2013). Moreover, this paper also uses the real exchange rate as a control variable. If domestic currency depreciates, foreign investors will buy more assets in this country because these assets become cheaper for them (Ghosh et al., 2014).

Contagion variables could also influence the sudden stops of capital flows. According to Forbes and Warnock (2012), linkages to certain subsets of countries through geography also impact the probability of experiencing an extreme capital flow episode. In line with their paper, geographical contagion is defined as a dummy variable, which equals one if there is at least one country in the same region that experiences a sudden stop in the previous month. The regions in this papers have been categorized into North America, Latin-America, Europe, Asia-Pacific and Africa, Middle-East countries.

3.5 Empirical model

Appendix 2 provides a detailed description and the sources for all the variables used in this paper. Appendix 3 demonstrates the summary statistics table for all variables. Due to missing data in some variables, particularly in domestic factors, 60 countries (28 developed countries and 32 developing countries) are included in the regression analysis. The time series is from January 2000 to July 2015 monthly.

First, this paper examines the impact of political risk on equity fund flows percentage scaled under AUM. 12 variables are added as controls with 6 global economic variables: global industrial production growth, global equity return, TED spread, Chicago Board Options Exchange volatility index (VIX) and commodity price shock; and 6 domestic economic variables: domestic industrial production growth, domestic equity return, short-term interest rate, CPI inflation, international foreign reserves and real exchange rate. The fixed effects model is constructed as follows:

𝑌𝑖,𝑡= 𝛽𝑃′𝑋𝑖,𝑡−1𝑃𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙 𝑟𝑖𝑠𝑘+ {𝛽𝐺 ′_𝑋

𝑡−1𝐺𝑙𝑜𝑏𝑎𝑙+ 𝛽𝐷′𝑋𝑖,𝑡−1𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐

𝐶𝑜𝑛𝑡𝑟𝑜𝑙 } + 𝜀𝑖,𝑡

𝑌_𝑖,𝑡 represents the equity fund flows (scaled by AUM) into country i at time t. 𝑋_{𝑖,𝑡−1}𝑃𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙 𝑟𝑖𝑠𝑘 is the political risk variable. 𝑋_𝑡−1𝐺𝑙𝑜𝑏𝑎𝑙 and 𝑋_{𝑖,𝑡−1}𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐 are vectors containing global and domestic variables. In order to avoid the reverse causality, all

(23)

explanatory variables are lagged one period. 𝛽_𝑃′, 𝛽_𝐺′ and 𝛽_𝐷′ are the vectors of coefficients. 𝜀𝑖,𝑡 is the random error term.

Second, five political risk measures are used to observe which specific factors have significant effects on capital flows.

𝑌_𝑖,𝑡= 𝛽_𝐺′_𝑋 𝑖,𝑡−1𝐺𝑜𝑣𝑒𝑟𝑛𝑚𝑒𝑛𝑡 𝑠𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦+ 𝛽𝑆′𝑋𝑖,𝑡−1𝑆𝑜𝑐𝑖𝑜𝑒𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛+ 𝛽𝐼′𝑋𝑖,𝑡−1𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 𝑝𝑟𝑜𝑓𝑖𝑙𝑒 + 𝛽_𝐶′_𝑋 𝑖,𝑡−1𝐶𝑜𝑟𝑟𝑢𝑝𝑡𝑖𝑜𝑛+ 𝛽𝐷′𝑋𝑖,𝑡−1𝐷𝑒𝑚𝑜𝑐𝑟𝑎𝑐𝑦+ {𝛽𝐺 ′_𝑋 𝑡−1𝐺𝑙𝑜𝑏𝑎𝑙+ 𝛽𝐷′𝑋𝑖,𝑡−1𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 } + 𝜀𝑖,𝑡

Third, in line with Calvo et al. (2008), Calderón and Kubota (2013), a probit model is used to research whether political risk and institutional quality can increase or decrease the probability of occurrence of sudden stops, in other words, whether political risk and institutional quality can cause extreme scenarios of capital flows. 𝑆_𝑖,𝑡 is a dummy variable which takes value one when there is a sudden stop in country i at time t. F(.) is the probit function and it is assumed to be a cumulative function of the standard normal distribution. In the vectors of contagion factors, equity fund flows scaled by AUM in the previous period is also added as a control variable because the probability of sudden stops is also determined by how much capital flows in or out of the country in the previous period.

Pr (𝑆_𝑖,𝑡) = 𝐹(𝛽_𝑃′_𝑋 𝑖,𝑡−1𝑃𝑜𝑙𝑖𝑡𝑖𝑐𝑎𝑙+ {𝛽𝐺 ′_𝑋 𝑡−1𝐺𝑙𝑜𝑏𝑎𝑙+ 𝛽𝐷′𝑋𝑖,𝑡−1𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐+ 𝛽𝐶′𝑋𝑖,𝑡−1 𝐶ontagion 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 })

Fourth, the same political indicators for political risk will be used in probit model regression analysis. Pr (𝑆_𝑖,𝑡) = 𝐹(𝛽_𝑃′_𝑋 𝑖,𝑡−1𝐺𝑜𝑣𝑒𝑟𝑛𝑚𝑒𝑛𝑡 𝑠𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦+ 𝛽𝑃′𝑋𝑖,𝑡−1𝑆𝑜𝑐𝑖𝑜𝑒𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛+ 𝛽𝑃′𝑋𝑖,𝑡−1𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 𝑝𝑟𝑜𝑓𝑖𝑙𝑒 + 𝛽𝑃′𝑋𝑖,𝑡−1𝐶𝑜𝑟𝑟𝑢𝑝𝑡𝑖𝑜𝑛+ 𝛽𝑃′𝑋𝑖,𝑡−1𝐷𝑒𝑚𝑜𝑐𝑟𝑎𝑐𝑦 + {𝛽𝐺′𝑋𝑡−1𝐺𝑙𝑜𝑏𝑎𝑙+ 𝛽𝐷′𝑋𝑖,𝑡−1𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐+ 𝛽𝐶′𝑋𝑖,𝑡−1𝐶ontagion 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 } + 𝜀𝑖,𝑡

Following Forbes and Warnock (2012), the complementary log-log (cloglog) model is also used as a robustness check because cloglog model is appropriate for asymmetrical data. It is frequently used when the probability of an event is small or large. In the data, sudden stops account for 19 % of the total sample.

(24)

Before estimating the model, the correlations of all the variables are examined. Appendix 4 provides a correlation matrix for political and economic (global and domestic) variables. As we can see, the highest correlation is between corruption and socioeconomic condition which is 0.65. Therefore, we can conclude that no multicollinearity is detected in the regression model.

4.Results

4.1 Determinants of capital flows

Table 1 summarizes the main results of determinants of capital flows using fixed effects model.6 Table 2 and Table 3 present the results of differentiating in income level and regions, respectively.

In Table 1, column (1) is considered as a baseline model. Global economic factors and domestic economic factors are included in the regression. Most of the control variables have significant effects on capital flows. Countries are more likely to have more net capital flows when the global interest rate is low, the global growth is high and the global equity return is high. Higher liquidity risk is associated with fewer capital flows. When there is a higher markets’ expectation of stock market volatility, investors tend to invest more in equity funds to diversify their portfolios. Higher commodity price also increases the capital flows. Comparing with global factors, the impact of domestic factors on capital flows is less dominant. If the recipient country has lower inflation, higher equity return, higher foreign reserves and lower exchange rate, the net capital flows will be higher. This result indicates that more capital will fly into the country when its economy is good.

Column (2) adds political risk rating as the proxy of political stability and institutional quality determining capital flows. The results show that lower political risk is associated with more capital flows in the recipient country. Furthermore, column (3) applies five political indicators for political risk. Among these components, only government stability exerts a significant effect in determining capital flows. Higher government stability drives more capital flows into a country. A stable government mitigates the uncertainties of the political environment and strengthen the belief of

(25)

investors.

Table 1: Determinants of capital flows

Dependent variable: Equity fund flows %

Sample of all countries (60 countries), 01/2001-07/2015 (monthly)

(1) (2) (3)

Sample All Countries All Countries All Countries

Political factor

Political risk rating 0.0223** (0.00938) Political factor components

Government stability 0.0381** (0.0176) Socioeconomic condition -0.0155 (0.0275) Investment profile 0.0313 (0.0229) Corruption 0.0251 (0.0519) Democratic 0.0388 (0.0509) Global factors (control)

Interest rate_global -0.103*** -0.124*** -0.122*** (0.0321) (0.0327) (0.0239) ∆ Industrial production_global 0.135*** 0.134*** 0.134*** (0.0293) (0.0292) (0.0228) Equity return_global 0.0885*** 0.0888*** 0.0885*** (0.00827) (0.00827) (0.00648) ∆Ted spread -0.204** -0.196** -0.196* (0.0927) (0.0934) (0.108) ∆VIX 0.0632*** 0.0637*** 0.0637*** (0.00910) (0.00908) (0.00752) Commodity Price 0.731*** 0.775*** 0.789*** (0.273) (0.274) (0.226)

Domestic factors (control)

Interest rate 0.000786 0.00249 0.00163 (0.00767) (0.00762) (0.00592) ∆ Industrial production 0.00728 0.00719 0.00716 (0.00539) (0.00541) (0.00590) Equity return 0.0374*** 0.0373*** 0.0375*** (0.00467) (0.00464) (0.00346) Foreign reserve % 0.0372*** 0.0358*** 0.0351*** (0.0102) (0.0111) (0.00563) Inflation -0.0124*** -0.0116*** -0.0113*** (0.00240) (0.00235) (0.00183) Real exchange rate -0.264** -0.297*** -0.295*** (0.107) (0.105) (0.0857) Constant -1.128 -2.825** -2.136** (0.946) (1.141) (0.831) Observations 9,458 9,458 9,458 R-squared 0.07 0.07 0.07 Number of countries 60 60 60

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 denote significance at the 1%, 5% and 10% respectively

(26)

Sample of developed (28) countries and developing (32) countries, 01/2001-07/2015 (monthly)

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

Sample Developed Developing Developed Developing

Political risk rating -0.00809 0.0362***

(0.0138) (0.0128)

Political factor components

Government stability -0.00388 0.0628* (0.0387) (0.0358) Socioeconomic condition -0.0118 -0.0369 (0.0413) (0.0747) Investment profile -0.0185 0.0567 (0.0292) (0.0458) Corruption -0.175 0.167** (0.112) (0.0800) Democratic -0.0283 0.0262 (0.145) (0.0728)

Global factors (control)

Interest rate_global -0.170** -0.124** -0.182** -0.130*** (0.0746) (0.0497) (0.0745) (0.0443) ∆ Industrial production_global 0.0883*** 0.175*** 0.0865*** 0.172*** (0.0130) (0.0553) (0.0130) (0.0550) Equity return_global 0.0677*** 0.110*** 0.0678*** 0.110*** (0.00762) (0.0140) (0.00754) (0.0144) ∆Ted spread -0.259*** -0.117 -0.258*** -0.121 (0.0834) (0.167) (0.0867) (0.169) ∆VIX 0.0699*** 0.0573*** 0.0702*** 0.0574*** (0.00472) (0.0170) (0.00474) (0.0170) Commodity Price 0.842** 1.046** 0.904** 1.213*** (0.362) (0.437) (0.365) (0.432)

Interest rate -0.00320 0.00510 -0.00211 0.00300 (0.0348) (0.00818) (0.0336) (0.00733) ∆ Industrial production 0.00106 0.0154** 0.00145 0.0151** (0.00687) (0.00706) (0.00713) (0.00707) Equity return 0.0338*** 0.0397*** 0.0337*** 0.0402*** (0.00529) (0.00640) (0.00539) (0.00643) Foreign reserve 0.0244 0.0365*** 0.0243 0.0341*** (0.0239) (0.0115) (0.0236) (0.0107) Inflation -0.0286*** -0.00891*** -0.0300*** -0.00884*** (0.00938) (0.00291) (0.00847) (0.00321) Real exchange rate -0.206 -0.328** -0.217 -0.346** (0.155) (0.133) (0.157) (0.133) Constant 0.843 -4.920*** 1.356 -4.458*** (1.635) (1.571) (1.133) (1.549) Observations 4,873 4,770 4,873 4,770 R-squared 0.09 0.08 0.09 0.08 Number of countries 28 32 28 32

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1denote significance at the 1%, 5% and 10% respectively

(27)

Table 2 describes the regression results by comparing developed countries and developing countries. In column (1), political risk rating is not significantly related to capital flows in developed countries. In column (2), political risk rating is positively related to capital flows in developing countries possibly owning to little difference and high stability in political risk across developed countries. In column (3) and column (4), similar results are found as in table 1, government stability has an important role in determining capital flows into developed countries while none of the other political factors have a significant impact on capital flows. In the developing countries, besides government stability, lower level of corruption also contributes to more capital inflows. In determining capital flows, all the global factors drive the capital flows both in developing and developed countries. The capital flows in developed countries are more likely to be driven by global economic factors while in developing countries the capital flows are more likely to be driven by domestic economic factors. The domestic political factors generally play a minor role.

Following Ghosh et al. (2014), the factors determining capital flows also differ in regions. Thus, in this paper, five regions are defined and results are presented in table 3.

First, in appendix 5, in line with the results in table 2, political risk rating is not significantly associated with capital flows in North America and Europe where most developed countries are located. Moreover, in the regions of Latin America and Asia- Pacific, political and institutional factors are significantly related to capital flows. In Africa and Middle East countries, political risk rating is insignificant determining capital flows. Global economic factors keep being a key force driving capital flows, especially in Europe and Asia-Pacific areas while the impact of domestic economic factors varies across regions.

Second, in Table 3, government stability has significant effects on capital flows in North America, Latin America, and Asia-Pacific. In Latin America and Asia-Pacific, higher government stability is related to more capital flows. However, in North America, only USA and Canada are considered. Government stability is negatively related to capital flows. Investment profile which measures aspects such as profits repatriation and payment delays in a country has an ambiguous effect across regions. In Asia-Pacific, it is positively related to capital flows while in North America and

(28)

Sample of countries in five regions, 01/2001-07/2015 (monthly)

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

Regions North America Latin America Europe Africa &

Middle East

Asia-Pacific Political factor components

Government stability -0.0652* 0.0976** -0.0378 0.0488 0.0564*** (0.00520) (0.0436) (0.0362) (0.0723) (0.0173) Socioeconomic condition -0.0527 0.0752 0.00420 -0.0525 -0.0200 (0.0660) (0.0680) (0.0660) (0.0950) (0.0317) Investment profile -0.183 -0.156** 0.0357 0.122 0.109*** (0.126) (0.0674) (0.0372) (0.106) (0.0341) Corruption 0.00518 -0.0536 -0.102 0.514** 0.0305 (0.0617) (0.113) (0.136) (0.242) (0.0578) Democratic -0.178 0.524*** -0.196*** -0.0624 0.00975 (0.255) (0.114) (0.0678) (0.194) (0.0545)

Interest rate_global -0.0147* -0.0145 -0.282*** 0.0340 -0.0674*** (0.00205) (0.0545) (0.0485) (0.0964) (0.0257) ∆ Industrial production_global 0.0444 0.0262 0.124** 0.326*** 0.0430* (0.0135) (0.0564) (0.0511) (0.0746) (0.0253) Equity return_global 0.0353* 0.105*** 0.0934*** 0.0861*** 0.0798*** (0.00502) (0.0159) (0.0177) (0.0211) (0.00697) ∆Ted spread -0.267 -0.307 -0.387*** 0.173 0.0188 (0.0678) (0.270) (0.121) (0.346) (0.120) ∆VIX 0.0371 0.0575*** 0.0931*** -0.00221 0.0597*** (0.0145) (0.0191) (0.0149) (0.0242) (0.00823) Commodity Price 0.307 -0.184 1.183*** 1.689** 0.157 (0.261) (0.563) (0.383) (0.793) (0.252)

Interest rate -0.0306 0.000733 0.0261 -0.0129 -0.0270** (0.0475) (0.00708) (0.0160) (0.0316) (0.0129) ∆ Industrial production 0.0402 0.0238* 0.00635 0.0267 -0.000909 (0.0228) (0.0141) (0.0116) (0.0216) (0.00468) Equity return 0.0205 0.0331*** 0.0339*** 0.0511*** 0.0395*** (0.0103) (0.00708) (0.00501) (0.0120) (0.00429) Foreign reserve 0.166 -0.000167 0.0660* 0.0412*** 0.0137* (0.117) (0.0138) (0.0328) (0.0112) (0.00747) Inflation -0.0206** -0.00993*** -0.0273*** -0.00201 -0.00678*** (0.000348) (0.00308) (0.00509) (0.00624) (0.00206) Real exchange rate -0.0409 -0.245 -0.365* -0.626** -0.187** (0.0668) (0.168) (0.196) (0.284) (0.0908) Constant 5.012 -0.165 0.228 -7.841*** -0.822 (0.991) (2.102) (1.362) (2.983) (0.895) Observations 349 1,177 4,380 1,630 2,107 R-squared 0.21 0.13 0.08 0.10 0.17 Number of countries 2 7 26 12 13

Robust Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 denote significance at the 1%, 5% and 10% respectively

Latin America it is negatively associated with capital flows. Democracy is also relatively important in Latin America and Europe. Higher democracy would increase

(29)

net capital flows in the recipient country while in Europe it is the opposite. On the one hand, more democracy promotes capital inflows because it protects property rights and reduces transaction costs for foreign investors. On the other hand, a higher level of democracy could hinder capital inflows by limiting the monopolistic or oligopolistic behaviors of multinational enterprises and constraining host governments' ability to offer generous financial and fiscal regulation to foreign investors.

4.2 Determinants of sudden stops in capital flows

This section presents the results of factors determining the extreme scenarios of capital flows, in other words, sudden stops. In addition to political factors, global economic and domestic economic factors, as well as contagion factors, are included. Table 4 follows the same set-up as Table 1, column (1) describes the baseline results without considering political factors. The occurrences of sudden stops of equity fund flows are significantly related to global, domestic and contagion factors. A country is more likely to experience sudden stops with a high global interest rate, low global industrial production growth, and low global equity return. The probability of a sudden stop also increases when global liquidity risk is high and implied volatility of the stock market is low. Lower commodity price increases the likelihood of sudden stops. Moreover, sudden stops are driven by domestic variables such as high inflation, low equity return and high financial exposure of the recipient country. A lower domestic real exchange rate also drives the occurrences of sudden stops. In addition, contagion factors play an important role in determining sudden stops. When there is a country in the same region that experiences a sudden stop in the previous period, the likelihood of sudden stops for the other country in the same region also increases. With fewer capital flows into the country in the previous period, the country is more likely to have sudden stops in the subsequent periods.

In column (2), political risk rating is added to investigate whether political and institutional determinants have significant effects on sudden stops. The results demonstrate that with less political risk, a sudden stop is more likely to occur although the coefficient is small. The reason could be with lower political risk, the probability of surges in capital flows goes up and it increases the likelihood of sudden stops in the future. In column (3), four out of five political factors have significant

(30)

Table 4: Determinants of sudden stops

Dependent variable: Sudden stops

Sample of all countries (60 countries), 01/2001-07/2015 (monthly)

(1) (2) (3)

Sample All Countries All Countries All Countries

Political risk rating 0.00967*** (0.00346) Political factor components

Government stability 0.0283** (0.0138) Socioeconomic condition -0.0621*** (0.0181) Investment profile 0.0431** (0.0171) Corruption 0.162*** (0.0316) Democratic 0.00222 (0.0251) Global factors (control)

Interest rate_global 0.186*** 0.176*** 0.168*** (0.0182) (0.0185) (0.0189) ∆ Industrial production_global -0.0596*** -0.0597*** -0.0595*** (0.0184) (0.0184) (0.0184) Equity return_global -0.0157*** -0.0155*** -0.0152*** (0.00524) (0.00524) (0.00526) ∆Ted spread 0.183** 0.186** 0.191** (0.0761) (0.0762) (0.0765) ∆VIX -0.0228*** -0.0226*** -0.0223*** (0.00567) (0.00567) (0.00569) Commodity Price -0.436*** -0.414** -0.316* (0.163) (0.163) (0.170)

Interest rate 0.00241 0.00486 0.00551 (0.00419) (0.00422) (0.00420) ∆ Industrial production -0.00260 -0.00240 -0.00276 (0.00476) (0.00477) (0.00477) Equity return -0.00633** -0.00634** -0.00588** (0.00285) (0.00285) (0.00286) Foreign reserve 0.0164*** 0.0167*** 0.0178*** (0.00394) (0.00389) (0.00394) Inflation 0.00342** 0.00383*** 0.00437*** (0.00140) (0.00141) (0.00147) Real exchange rate -0.261*** -0.263*** -0.277*** (0.0698) (0.0697) (0.0708)

Contagion factors (control)

Geographic contagion 1.344*** 1.346*** 1.364*** (0.0472) (0.0472) (0.0475) Capital flows -0.0702*** -0.0701*** -0.0705*** (0.00832) (0.00831) (0.00832) Constant -0.759 -1.536** -2.007*** (0.560) (0.627) (0.632) Observations 9,603 9,603 9,603 Number of countries 60 60 60

Standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1 denote significance at the 1%, 5% and 10% respectively

(31)

Dependent variable: sudden stops

Sample of developed (28) countries and developing (32) countries, 01/2001-07/2015 (monthly)

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

Sample Developed Developing Developed Developing

Political risk rating 0.0243*** 0.00340

(0.00629) (0.00449)

Political factor components

Government stability 0.0512** 0.00890 (0.0203) (0.0196) Socioeconomic condition -0.0558* -0.0778*** (0.0291) (0.0250) Investment profile 0.105*** -0.0109 (0.0266) (0.0243) Corruption 0.190*** 0.295*** (0.0425) (0.0570) Democratic 0.148*** -0.0483* (0.0519) (0.0285)

Interest rate_global 0.338*** 0.0153 0.350*** -0.00192 (0.0313) (0.0280) (0.0312) (0.0287) ∆ Industrial production_global -0.0868*** -0.0410 -0.0814*** -0.0434* (0.0262) (0.0261) (0.0262) (0.0262) Equity return_global -0.0312*** -0.00198 -0.0324*** -2.38e-05

(0.00788) (0.00725) (0.00790) (0.00729) ∆Ted spread 0.297*** 0.0610 0.312*** 0.0617 (0.106) (0.112) (0.106) (0.113) ∆VIX -0.0221*** -0.0238*** -0.0225*** -0.0230*** (0.00840) (0.00796) (0.00844) (0.00799) Commodity Price -1.669*** 0.789*** -1.915*** 1.140*** (0.238) (0.257) (0.254) (0.274)

Interest rate 0.0380** 0.00245 0.0325* 0.000459 (0.0172) (0.00429) (0.0170) (0.00441) ∆ Industrial production -0.00369 -0.00218 -0.00426 -0.00269 (0.00646) (0.00725) (0.00650) (0.00729) Equity return -0.000974 -0.00754** -0.000681 -0.00680* (0.00522) (0.00348) (0.00525) (0.00351) Foreign reserve 0.0151 0.0148*** 0.0162 0.0144*** (0.0109) (0.00412) (0.0106) (0.00419) Inflation 0.0268*** -0.00365** 0.0302*** -0.00416** (0.00364) (0.00173) (0.00379) (0.00190) Real exchange rate -0.467*** -0.154* -0.436*** -0.214** (0.120) (0.0880) (0.121) (0.0901)

Contagion factors (control)

Geographic contagion 1.202*** 1.453*** 1.215*** 1.495*** (0.0680) (0.0685) (0.0682) (0.0700) Capital flows -0.0602*** -0.0784*** -0.0573*** -0.0816*** (0.0135) (0.0108) (0.0135) (0.0109) Constant -0.362 -4.695*** -0.703 -5.652*** (0.957) (0.937) (0.915) (0.986) Observations 4,847 4,756 4,847 4,756 Number of countries 28 32 28 32

(32)

impacts on sudden stops. High government stability, satisfactory investment profile, and lower corruption are positively related to the probability of sudden stops. A strong socioeconomic condition in the recipient country decreases the likelihood of sudden stops.

In table 5, political risk rating is significantly associated with sudden stops in the sample of developed countries. In column (3), the results for developed countries are in line with table 1. A stable, democratic environment with good investment profile and low corruption will drive the probability of sudden stops in developed countries. In addition, corruption also decreases the likelihood of sudden stops in developing countries because sometimes higher corruption can facilitate the process and provide investors greater opportunities through bribing the bureaucracy thus reducing the sudden stop probability.

In appendix 6, the composite political risk rating index is only found to be significantly related to sudden stops in Europe. When political risk is broken down into each political instrument in table 6 below, most of the political factor components, i.e. government stability, socioeconomic condition, corruption, and democracy have significant effects on sudden stops in Latin America and Asia-Pacific areas. Political factors are not significantly associated with sudden stops in Africa and the Middle East countries. Corruption plays an important role in determining sudden stops in all the regions except Africa and Middle East. A socioeconomic condition is also significant in determining sudden stops in Latin America and Asia-Pacific where are most emerging markets economies (EME) located. A higher level of socioeconomic condition implies strong consumer confidence which reduces the likelihood of sudden stops.

4.3 Robustness check

Serial correlation within the panel data is detected in both fixed effects model and probit model. For fixed effects model in determining the magnitudes of capital flows, robust standard errors are used to correct heteroscedasticity and serial correlation. For probit model in determining sudden stops, capital flows are included as a lagged dependent variable to reduce the bias caused by serial correlation. The results in all probit models demonstrate that fewer capital flows in the previous period increase the probability of sudden stops for the next period.

(33)

Dependent variable: sudden stops

Sample of countries in five regions, 01/2001-07/2015 (monthly)

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

Sample North America Latin America Europe Africa &

Middle East

Asia-Pacific Political factor components

Government stability -0.131 0.139*** 0.0327 -0.0652* 0.0717*** (0.103) (0.0444) (0.0234) (0.0374) (0.0269) Socioeconomic condition 0.250 -0.182*** -0.0263 -0.00327 -0.227*** (0.184) (0.0520) (0.0305) (0.0341) (0.0533) Investment profile -0.443 0.0450 0.0764*** -0.00747 -0.0563 (0.286) (0.0556) (0.0263) (0.0427) (0.0500) Corruption 0.874*** 0.236** 0.133*** -0.00640 0.337*** (0.217) (0.0969) (0.0397) (0.104) (0.0889) Democratic 1.543*** 0.323** 0.0407 -0.0351 0.114 (0.464) (0.126) (0.0598) (0.0451) (0.0742)

Interest rate_global 0.324 -0.173** 0.324*** -0.00242 0.0854** (0.299) (0.0810) (0.0322) (0.0518) (0.0391) ∆ Industrial production_global -0.0987 -0.209*** -0.0405 -0.0460 -0.0754** (0.0992) (0.0629) (0.0302) (0.0456) (0.0367) Equity return_global -0.0256 0.00400 -0.0301*** 0.0174 -0.0210* (0.0256) (0.0182) (0.00884) (0.0117) (0.0111) ∆Ted spread 0.527 0.149 0.296*** 0.00290 0.117 (0.413) (0.305) (0.114) (0.202) (0.161) ∆VIX -0.0569* -0.000562 -0.0198** -0.0241* -0.0325*** (0.0308) (0.0201) (0.00904) (0.0135) (0.0118) Commodity Price -2.462** 1.281* -1.571*** 0.0869 1.653*** (1.105) (0.713) (0.275) (0.484) (0.380)

Interest rate -0.426** 0.00422 0.00134 0.0155 0.0293 (0.195) (0.00495) (0.00910) (0.0159) (0.0190) ∆ Industrial production -0.0248 -0.00114 0.00677 -0.0102 -0.00817 (0.106) (0.0175) (0.00879) (0.0124) (0.00750) Equity return -0.00286 -0.00159 0.00350 -0.00734 -0.0175** (0.0275) (0.00886) (0.00469) (0.00647) (0.00698) Foreign reserves -0.866*** 0.0171 0.0153 0.0192*** 0.00162 (0.294) (0.0135) (0.0106) (0.00548) (0.0118) Inflation -0.0540** -0.0179*** 0.0183*** -0.00718** 0.00433 (0.0270) (0.00643) (0.00293) (0.00351) (0.00315) Real exchange rate -0.222 0.0439 -0.688*** -0.322** 0.229

(0.386) (0.175) (0.149) (0.161) (0.143) Contagion factors (control)

Geographic contagions 0.409** 2.168*** 1.502*** 1.291*** 1.077*** (0.184) (0.173) (0.0908) (0.107) (0.103) Capital flows -0.765*** -0.0710** -0.0658*** -0.0678*** -0.185*** (0.217) (0.0361) (0.0116) (0.0152) (0.0417) Constant 6.054 -7.937*** 0.389 -0.426 -8.583*** (6.155) (2.723) (0.977) (1.815) (1.400) Observations 347 1,172 4,360 1,626 2,098 Number of countries 2 7 26 12 13

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 denote significance at the 1%, 5% and 10% respectively

International capital flows and sudden tops : political and institutional determinants