1
U
NIVERSITY OF
A
MSTERDAM
Gross Capital Inflows: Global and National Determinants of aggregate and
disaggregate gross capital inflows in High, Upper-Middle and Low-Middle
income countries
Master Thesis
Puneet Sondh
Student Number – 11144556
Supervisor – Dr. J.E. Ligterink
2
ABSTRACT
This paper assesses empirically the determinants of aggregate and disaggregate gross capital inflows to countries based on their income level. The sample consists of 71 countries categorized as per income level based on the World Bank classification. It analyzes the effect of Global Factors, national macroeconomic factors and structural factors on gross capital inflows. Using Panel method (Fixed effect), the study finds that global factors – VIX, US Long term and short interest rates, growth in G7 countries – National macroeconomic factors – National real interest rate, Real GDP growth rate – and Structural factors – Human Capital – drive gross capital inflows. The effect of these factors differ by each income group, and the factors differ between upper middle income and lower middle income group, though both are clubbed as developing countries as per the IMF classification. We find structural factors playing a role in attracting capital in the high income group, which suggests that high income group also has varying structural factors and they create a pull for capital. The results of the study indicate that in the past decade as economies became more integrated, no group of country can insulate itself from global factors, unless they disintegrate themselves from the global financial system.
3
Table of Contents
1. Introduction ... 4
2. Literature Review ... 7
3. Hypothesis and Framework ... 11
4. Data and Methodology ... 16
5. Empirical Findings ... 24
6. Conclusion ... 27
7. References ... 29
4 1. Introduction
Many countries experienced waves of capital flows in the 1980s and the 1990s. With increased globalization and integration of world markets, the volatilities in capital flows have increased in the past decade. Capital flows surged in the mid-2000s and contracted during the Global financial crisis (2008-09) and rebounded in 2010. Volatility in capital flows can have widespread economic consequences such as increasing financial system vulnerabilities and aggravating macroeconomic instability. Capital flows can also be beneficial, during crises, when domestic investor liquidate foreign investments and repatriate capital, resulting in beneficial capital inflows for a country.
Waves in capital flows have generated an extensive academic literature. Several papers have examined “sudden stops”, “surges” or “bonanzas” or “capital flight”. The prominent papers on stops, surges, bonanzas and capital flight are that of Calvo (1998), Calvo et al (2004), Reinhart and Reinhart (2009) and Rotenberg and Warnock (2011). A strand of literature has focused on explaining contagion and crises in capital flows, while another strand has focussed on
determinants of capital flows to emerging economies. Prominent papers on determinants of capital flows are Calvo et al (1993), Chuhan et al (1998) and Griffin et al (2004). Some papers have focussed on microeconomic impact of crises such as the Asian crisis, Russian crisis at the firm-level. As a consequence, a vast literature has grown, analysing the cyclical behaviour of capital flows, mostly in the emerging markets – Asia, Latin America and Europe.
A major theme that runs through much of the research on capital flows is whether the forces driving capital flows are “push” factors or “pull” factors. Push factors are considered to be driven by a set of favourable global factors, which may be temporary and can reverse themselves in the future. On the other pull factors are country specific and depend on
macroeconomic fundamentals and institutions of a country. While globalization has increased financial integration among countries, understanding the factors which influence capital flows is important for policy makers, as misguided policies can result in exposure to greater volatility in capital flows. The seminal papers in the literature- Calvo et al (1993), Chuhan et al (1998) – find that push factors are more important than domestic fundamentals in driving capital flows. Griffin et al (2004) argue that pull and push factors are important in understanding cross border equity flows. Jevac et al (2010) look at capital flows to new EU member states from Central and Eastern Europe (NMS10) during the last decade, and find that both push and pull factors play an important role in attracting capital flows.
5
The majority of the literature has concentrated on studying the net capital flows, defined as the difference between total capital inflow and outflow. In recent years, the focus has shifted from net capital flows to gross capital flows to examine the effect of capital flows in emerging markets. According to Rothenberg and Warnock (2011) net capital flows may be driven by capital inflows and outflows which may be related to different factors. Capital inflow is defined as net purchases of domestic assets by foreigners, while capital outflow is defined as net
purchases of foreign assets by domestic agents. As gross capital flows reflect behaviour of foreign investors and domestic agents, capital in- and outflows need to be studied separately. Forbes and Warnock (2012) have suggested few papers have studied gross capital inflow data, instead focusing on the more readily available net flow data. They also identify important global variables that drive extreme movements in capital inflows and outflows. Byrne and Fiess (2015) build on the idea of gross capital inflows and investigate the nature and determinants of
aggregate and disaggregate capital inflows to emerging markets, using quarterly inflow data from Euromoney Bondware and Loanware, with a sample period 1993Q1 to 2009 Q1.
Most research on capital flows is based on the development level of a country and follows a developing/developed dichotomy. Nielsen (2011) suggests that a developing/developed
dichotomy is too restrictive and that a classification system with more than two categories could better capture the development outcomes across countries. Nielsen mentions the challenge of classifying countries like Malaysia and Russia based on the developing/developed dichotomy. While, Malaysia and Russia are classified as developing countries, their development indicators may vary significantly from a country like Burkina Faso, which is classified as a developing country as well. Nielsen (2011) further explores the classification system adopted by World Bank, IMF, United Nation Development program. Of the three classification system World Bank and IMF classify countries for operational and analytical purposes. In 1978, World Bank for the first time constructed an analytical country classification system, which was later reformed in 1989, as the 1978 WDI had some high income countries classified as developing economies. The reform in 1989 created income thresholds for country classification which is used by the bank to this date.
Similarly, IMF constructed an analytical country classification system which categorized the countries as (1) Industrialized countries (2) Other high-income countries and (3) less-developed countries. In the 1970s the classification system was revised twice before the fund adopted a significantly simplified classification system in the early 1980, which categorized the countries as (1) Industrial countries and (2) Developing countries. Initially, the industrial group consisted of
6
21 countries, however, Greece and Poland were classified in 1989 as industrial countries. The relevant report – the October 1989 WEO – is silent on the reasons for the re-classification of Poland and Greece. In 1997 industrial country group was renamed the advanced countries group and several countries were added to the group in 1997 (Israel, Singapore, Korea), 2001 (Cyprus), 2007 (Slovenia), 2008 (Malta), 2009 (Czech Republic, Slovak Republic). However, the relevant WEO reports provide no reasons for the re-classifications.
While Nielsen(2011) contends that the present system of classifications is not perfect, he points out that among IMF,UNDP and World Bank, World Bank’s high income group is broader and it encompasses all designated advanced and developed countries of the other two institutions as well as countries which are neither “advanced” or “developed”. In 2010, World Bank 26% of countries classified as “developed” compared to 17% for IMF. In 1990 the share of “developed” countries were 16% and 13% for the World Bank and IMF respectively. Hence, it can be
deduced that the share of countries classified as “developing” has fallen more for World Bank than the IMF.
While most research has focused on the dichotomous classification of developed/developing countries or emerging markets, no study to the best of my knowledge has studied the
determinants of flows based on the income classification of countries. Also, Forbes and Warnock (2012) have suggested few papers have studied gross capital inflow data, instead focusing on the more readily available net flow data. While Byrne and Fiess(2015) have studied commonalities in gross flows for a set of countries, the focus has been on developing countries as per the dichotomous classification, and not on the income based classification.
The capital flows from 2001 to 2014 (Source: IMF, BOP Data) for a group 70 countries show the following trends among income groups –
7
Figure 1 – Gross Capital Inflow trend
We see that the share of high income countries fell drastically from above 90% in 2001 to around 40% in 2009, while the relative share of upper middle income and lower income
countries increased drastically. However, post 2010 the share of lower middle income countries has been relatively stable compared to high income and upper middle income countries. Though Lower middle income countries and upper middle countries are clubbed as developing countries, the above figure illustrates that the share of gross capital inflows varies among income groups and a study based on income levels will help us in understanding the factors affecting gross inflows (investor preferences) based on a country’s income level.
Understanding factors affecting gross flows will also provide an insight on how policy makers of different income group can deal with volatility in capital flows. In case, country specific factors are significant in affecting flows, then policy makers should focus on making their domestic economies more resilient by focusing on institutions and macroeconomic fundamentals. The first aim of the paper will be to determine the global and country specific determinants of aggregate gross capital inflows. The second aim of the paper will be to understand the
determinants of disaggregate gross capital inflows across income groups. The period of study will be 2001-2014.
The paper is structured as follows: Section 2 provides a literature review, Section 3 covers the hypothesis and framework, Section 4 data and methodology, Section 5 provides empirical results and Section 6 concludes with the discussion of results.
2. Literature Review
The area of capital flows is highly researched in Macroeconomics and International financial management. Increased capital flows in the 1980s and 1990s, coupled with Asian and Russian crises have made it an extensively researched field. The Global Financial crisis (GFC) renewed the debate on the nature of capital flows and its volatility to macroeconomic stability and policy responses. In this section we review the literature associated with capital flows and existing empirical evidence on capital flows. Several studies have been done to understand the
determinants of capital flows, episodes of surge and retrenchment and institutional effectiveness on constituents of capital flows. These studies have been done over different points of time and geographic areas, and conclusions have varied based on the variables chosen in the study. Studies have used net capital flows, gross capital flows, portfolio inflow, FDI to measure the macro economic variables affecting the flows. In the context of this thesis, study was done to identify
8
macroeconomic variables which can explain the gross capital inflows to a country, and understand the econometric method used by the researchers in their study.
Capital Flows – Global or Country specific factors
A major theme that runs through the literature on capital flows is whether forces driving capital flows are “push” factors (Global) or “pull” factors (Country – specific. One of the seminal paper on determinants of capital flows – Calvo et al (1993) – argues that although domestic factors were a necessary ingredient for reviving capital flows, they could partially explain Latin America’s forceful re-entry into international capital markets. They argued that some of the renewals resulted from external factors, which could be considered as external shocks common to the region. They find that international interest rates and recession in the United States were external factors that explained inflows into Latin America.
Chuhan et al (1998) examined Portfolio flows into Latin America and Asia and found that external and country specific factors were important in explaining flows in Latin America and Asia. They found that more than half of explained increases in flows to Latin America could be attributed to US interest rates and recession in the US. On the other hand, country specific factors were nearly three to four times more important than the global factors for Asia, with credit rating being a significant variable.
Griffin et al (2004) argue that push and pull factors are important in understanding cross border equity flows. Zlate and Ahmed(2014) find that growth and interest rate differentials between Emerging markets and advanced economies as well as global risk appetite are important
determinants of net private capital inflows. They also find significant changes in the behaviour of net portfolio inflows from the period prior before the GFC to the post crisis period, explained by greater sensitivity to interest rates. Jevack et al (2010) using net capital flows analyse the role of various types of flows over time and across countries (NMS10). They find the importance of global factors – euro area interest rate, business cycle, risk – and national factors – domestic economic conditions and policies – in the ability of NMS10 countries to attract capital flows. Byrne and Fiess (2015) use gross capital inflows to investigate the nature of aggregate and disaggregate capital flows in emerging countries, and find advanced economies long-run bond yield and commodity prices are external factors impacting flows to emerging economies, while financial openness and institution are country specific factors. The impact of US long term interest rates and capital flows to emerging economies was studied by Olaberria(2014) and he
9
found that cycle in capital flows to emerging economies was linked to global risk aversion and long term US interest rates. Lucas (1990) compared production per person between India and US, and estimated India’s marginal product of capital to 58 times that of the US, and as per neo classical theory, capital should flow from US to India. Lucas (1990) proposed Human Capital to be a significant factor in explaining the departure from the neo classical theory, and his theory was termed “The Lucas Paradox”.
The GFC spawned a surge in theoretical research on crises and capital flows. Much of the literature focuses on “push” factors driving capital flows and especially the role of risk and liquidity/credit. Caballero et al (2008) and Mendoza et al (2009) focus on pull factors and highlight the size, depth and fragility of a country’s financial system in attracting capital flows from abroad for developed countries or driving capital flows out of the country for less developed financial markets. Milesi- Ferretti and Tiles (2010) find that across types of flows, banking flows were the hardest hit due to their sensitivity of risk perception and found capital retrenchment to be short lives in developing economies compared to advanced economies. However, the existing literature on capital flows does not favour one determinant playing a major role over others.
Capital flight, stops and bonanzas
Calvo (1998) originated the literature on extreme capital flow episodes in his analysis of “sudden stops” defined as sharp slowdowns in capital flows. His definition was broadened by adding criteria such as 1) the requirement that a stop occurred at the same time as a contraction in output (Calvo et al (2004) or that 2) stop had to occur in conjunction with a sharp rise in interest rate spreads, to capture a global component and qualify as “systemic sudden stop” (Calvo et al 2008). The mirror image of capital sudden stop is the capital surge defined as a sharp increase in net capital flows (Reinhart and Reinhart 2009). Milesi-Ferretti and Tile (2010) examine
retrenchment of capital flows during the Global Financial crisis and find that retrenchment in capital flows is a heterogeneous phenomenon across time and across flows. They also show that magnitude of the retrenchment across countries is linked to the extent of financial integration, its specific nature as well as domestic macroeconomic conditions and their connection to world flows. They also shifted attention to the importance of considering gross capital flows instead of simply net flows.
Building on this Cowan et al (2008) and Rothenberg and Warnock (2011) point out that measures of “sudden stop” constructed from proxies of net capital flow are unable to
10
differentiate between stops due to action of foreigners and those due to domestic agents fleeing the domestic market. Forbes and Warnock (2012) further make use of gross capital flows for identifying episodes of extreme capital flows using data that differentiate activities by foreigners and domestics. Their approach yields fundamentally different results from literature based on net capital flows. They find global risk an important factor in explaining capital inflows, however, they find little association between capital controls and probability of surge or stops driven by foreign capital flows
Institutional Effects and Capital Controls
Another strand of literature focusses on Foreign Direct Investments (FDI) and institutional effects. The literature on institutional effects measures the role played by financial openness- proxy for capital controls- human capital, institutional strength on capital flows and direct investments. In a seminal paper North (1994) argued that neoclassical economic theory focused on technological development and had recently stared to focus on human capital. He further asserted that in the evaluation of economic performance through time it contained two erroneous assumptions: (1) Institutions do not matter and (2) Time does not matter. North states that it is the interaction between institutions and organizations that shape the institutional evolution of an economy, and, if institutions are the rules, then organizations are the players. Shapiro and Globerman(2002) studied the role of governance infrastructure on global foreign direct investment flows and found that government infrastructure to be an important indicator of FDI inflow and outflow. Investments in governance infrastructure not only created capital but also created the conditions under which domestic multinational emerge and invest abroad. Jude and Levieuge(2013) investigate the effect of FDI on economic growth conditional on institutional quality for developing economies and find that in order to benefit from FDI led growth, improvement of institutional framework should precede FDI attraction. Gani(2011) studied the role of governance and growth in developing countries using the World Bank indicators and found political stability and government effectiveness to be positively associated with growth. On the other hand, voice and accountability and corruption to be significantly negatively associated with growth. Surprisingly regulatory quality and rule of law were insignificantly correlated with growth.
Chinn and Ito(2006) focussed on the link between capital account liberalization, legal and institutional development, and financial development especially in the equity markets. In an analysis encompassing 20 years and 108 countries, their results suggest that higher level of
11
financial openness leads to equity market development if a threshold level of general legal system and institutions is attained. They also find that finance related legal/institutional variables do not enhance the effect of capital account opening as strongly as the general legal/institutional variables. In their study Chinn and Ito (2006) developed a measure of financial openness KAOPEN based on the first principal component of the four IMF binary variables from IMF’s Annual report on Exchange Arrangements and Exchange Restrictions (AREAER). The index was developed to measure the extent of openness in capital account transactions and the prevalent measures did not fully capture the intensity of capital controls. The index has been used in other studies notably Forbes and Warnock (2012) and Byrne and Fiess (2015). A summary of key literature is encapsulated in the table below –
Year Researchers Measure Significant Variables
1993 Calvo Capital flows from BOP US interest rates
1998 Chuhan et al Equity and Bond flows
US Interest rates, US Industrial Production, Country Credit Rating, Country Price Earnings
ratios
2002 Globerman and Shapiro FDI Flows (Logarithm) Government Infrastructure 2006 Chinn and Ito Financial Development Financial Openness, Legal System, Institutions 2010 Jevack et al Net Capital flows as percentage of GDP Interest Rate, Risk, National GDP growth 2012 Forbes and Warnock Gross Capital Flows Global Risk, Global interest rates, Global Growth 2012 Fratzscher Net Capital Flow to Country(i) (EPFR fund level Data) Sovereign Ratings, FX Reserves, Global Risk, Short Term Debt
2013 Jude and Levieuge Growth rate of GDP Institutional Quality
2014 Ahmed and Zlate Net Private Inflows as a percentage of Nominal GDP Interest Rate differentials, Global Risk and Capital Controls 2015 Byrne and Fiess Quarterly Capital Inflow Data
Financial Openness, Institutions, Advanced economy bond yields,
commodity price (Non-oil)
3. Hypothesis and Framework
From the literature review we can see that the existing literature does not favour one determinant playing a major role over the other. Past studies have used different data sets and econometric methods to determine pull and push factors, institutional effects and effect of extreme episodes
12
of capital flows. Initially the studies focussed on net capital flows, however, after Forbes and Warnock (2012) used gross capital flows to study episodes of extreme flows, which yielded fundamentally different results from net capital flows, there has been a shift in research towards the use of gross capital flows. Gross capital flows are considered to be a measure of purchase of domestic assets by foreign investors, and, purchase of foreign assets by domestic agents. Hence, gross capital flows are a much more relevant metric to study financial market development and financial integration, especially the effect of capital controls and structural factors like
government effectiveness, rule of law, human capital, etc.
Based on the literature review, this study uses a set of macroeconomic variables and institutional effectiveness measure to determine the pull and push factors of gross capital inflows for
countries grouped by their income level. The choice of variables is modelled on the study done by Byrne and Fiess (2015). While Byrne and Fiess(2015) used gross capital inflow data from bond ware, this study will utilize gross capital inflow data from IMF Balance of Payments statistics. The choice of data source is based on access and availability of data from the source. The dependent variable for the study will be aggregate and disaggregate gross capital inflows divided by nominal GDP, similar to the study done by Byrne and Fiess(2015). One departure from the model used by Byrne and Fiess(2015) is the use of variable to measure Institutional effectiveness. Byrne and Fiess(2015) have used Institutional Quality from the International Country Risk Guide(ICRG). Due to non-accessibility of Institutional quality data from ICRG, this study utilizes World Bank Indicator of Government effectiveness and Rule of Law to measure Institutional quality.
The explanatory variables chosen for the study are Financial Openness, Government
effectiveness, Rule of law, Non-energy commodity price index, US Long term interest rate, US short term interest rate, Growth in G7 countries, VIX Risk (Global Volatility) index, National real Interest rate, Debt to GDP. The frequency of the data is yearly and the time period is 2001-2014. The data is available for majority of the countries for the time period 2001-2001-2014. The data has been sourced from IMF Balance of Payment and International Financial Statistics, World Bank, Gap Minder, CBOE and OECD databases.
Gross capital inflow encompasses Foreign direct investment, Portfolio investment (debt and equity) and other investments (primarily bank loans) and are drawn from the liabilities side of the financial account of the IMF balance of payment statistics.
13 Theoretical Framework
We can see from the literature review that there are no fully accepted theories on the drivers of gross capital flows or net capital flows. Studies such as Byrne and Fiess(2015), Forbes and Warnock(2012), Chuhan (1998), Calvo(1993) have used different econometric models and data sources to identify drivers of capital flows. Bases on previous studies of Chuhan et al (1998) and Byrne and Fiess (2015), this study will employ panel data approach to identify drivers that affect gross capital inflows to high, upper middle and lower middle income countries.
Apart from considerations of global factors, the study is also interested in understanding country specific determinants. The empirical model for global and country specific determinants is given by the following equation –
𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 = 𝛼𝛼0 + f (𝑌𝑌𝑖𝑖𝑖𝑖) + f (𝑋𝑋𝑖𝑖 ) + 𝜀𝜀𝑖𝑖𝑖𝑖 t = 1,2,….T
The capital flows for countries are affected by both a vector of global determinants denoted by the subscript t and country specific determinants denoted by the subscript “it”.
Empirical Model
The study will employ a Panel data approach for the following reasons –
1. Panel data estimation method is among the most efficient techniques to analyse the impact of a common set of global factors across a diverse set of countries.
2. The structure acknowledges that each country can have its own characteristics which can be correlated or uncorrelated with some or all of the explanatory variables.
3. It is an appropriate method to alleviate the effect of omitted time-invariant variables that are correlated with the explanatory variables.
4. Panel data technique resolves some of the econometric problem by increasing the data points and decreasing collinearity among explanatory variables.
The Panel data approach separates a time series of Country specific capital flows into country specific fixed effects for each country, a common factor which varies over time t and is associated with a factor loading and idiosyncratic component.
The model specification for the study is as follows –
14
+ α7 DGDPit +α8RCP.t + α9RSRUS.t + α10RLRUS.t + α11VIX .t
+ α12Y.t G 7 + ξit i = 1, … , N ; t = 1, … , T FO = Financial Openness;
G = Government Effectiveness; HC = Human Capital; ΔY = Economic growth; R = Local Interest Rate; L = Rule of Law; DGDP = Debt to GDP;
RCP = Non –oil commodity Index; RSUS = US Interest rate (short-term); RLUS = US interest rate (long-term); VIX = Risk (Volatility Index);
Y(G7) = Growth in Advanced economies;
The Panel time series 𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖 is tested for non - stationarity by examining individual unit roots of
all variables in the above equation. The Panel unit root tests will be done using the Augmented Dickey fuller test (ADF). Global determinants denoted by the subscript t and country specific determinants denoted by the subscript “it”.
Hypothesis
Literature on capital flows, Foreign Direct investments, institutional effects do not favour one determinants playing a major role over another. Caballero et al (2008) and Mendoza et al (2009) focus on pull factors and highlight the size, depth and fragility of a country’s financial system in attracting capital flows from abroad for developed countries or driving capital flows out of the country for less developed financial markets. Shapiro and Globerman(2002) studied the role of governance infrastructure on global foreign direct investment flows and found that government infrastructure to be an important indicator of FDI inflow and outflow. Similarly, other studies have found different determinants to be significant in explaining capital flows, episodes of capital
15
flight or FDI investments. From the literature we infer that structural factors are considered key for financial development as well as attracting investments. We can also infer that as a country becomes more financially integrated, global factors affect the flow of capital to a country. Based on the World Bank classification, we can hypothesize that high income countries will be
characterized by strong institutions, human capital, financial openness compared to upper middle income and lower middle income countries. As the income of a country is affected by structural factors, it can be hypothesized that upper middle income countries will have better institutions and human capital compared to low income countries. Apart from structural factors, national macroeconomic factors like the growth of GDP, interest rates will affect flows, as investment theory suggests that investors will seek to diversify their risk as markets become more integrated. Hence, to test our assumptions around the impact of structural, national macroeconomic and global factors we will assess the following hypotheses-
Hypothesis 1: Push Factors -US short-term and long term interest rates, Global Risk as measured by VIX and growth of G7 countries – will be significant in explaining gross capital inflows to high income countries, as most of the countries have well-functioning domestic institutions and macroeconomic policies.
Hypothesis 2: For Upper Middle and Lower Middle income countries, both push and pull factors will be significant in explaining gross capital inflows. Pull factors will vary between the two income groups. Domestic economic factors – real interest rate, growth in real national GDP- will be significant for upper middle income countries, while structural factors – human capital, financial openness, government effectiveness, rule of law – will be significant.
Hypothesis 3: Global factors – Risk, US Short term and long term interest rate- and country specific factors – real national interest rates, Debt to GDP, GDP growth – will be significant in explaining gross portfolio and loan inflows for all income groups.
Hypothesis 4: Global factors- Growth in G7 countries, US interest rates- and country specific factor – growth in real GDP- will be significant factors for FDI in high income countries. Price of non-oil commodities, VIX and country specific factors – financial openness, government effectiveness, rule of law – will be significant factors for FDI in upper middle income countries. Risk, US interest rates, and country specific factors – GDP growth, Human Capital, Rule of law, Financial openness- will be significant factors for FDI in lower middle income countries.
16 4. Data and Methodology
Countries for the Study
The countries chosen for the study are based on the income classification of World Bank. For the purpose of this study, Low Income countries and Fragile countries were excluded. The reason for exclusion of Low income countries is that most of the capital flows in those countries are grants and loans by international financial institution. Moreover, according to the current income classification of World Bank, only 5 countries are classified as Low Income Countries. Tax havens like Panama, Luxembourg et al were also excluded from the dataset. Other notable exceptions are Argentina, which is currently non-classified by World Bank, and Greece, which faced a sovereign crisis in 2010-11. The final short list of 71 countries was arrived basis the cumulative nominal GDP share of these countries to the world’ nominal GDP. The 71 countries chosen for the study account for almost 95% of the world’s nominal GDP. The list of the countries for the purpose of the study is provided in Appendix 1. The break-up of the countries by income group is provided below -
Table 2 – No. of Countries by income group The regression equation to test the hypothesis is –
CAPit = α0 + α1FOit + α2Git + α3HCit + α4ΔYit + α5Rit+α6 Lit
+ α7 DGDPit +α8RCP.t + α9RSRUS.t + α10RLRUS.t + α11VIX .t
+ α12Y.t G 7 + ξit i = 1, … , N ; t = 1, … , T
Global determinants denoted by the subscript t and country specific determinants denoted by the subscript “it”. CAPit is the aggregate and disaggregate flows divided by the nominal GDP to account for the relative size of flows across all countries. CAPit also measures the change in capital inflow relative to the change in the economic growth of a country, hence making it a more robust variable for the study than just the change in capital inflows of a country.
Income Classification Number of Countries
High income 33
Lower middle income 19
17 Global Variables
VIX – It is a volatility index calculated by the Chicago Board Option Exchange (CBOE). It measures implied volatility over a wide range of S&P 500 options and captures economic uncertainty or risk and reflects risk aversion in the global markets. A rise in the implied volatility reflects a decline in investors risk appetite. Data Source: CBOE
RSUS – It is the Real US short term interest rate calculated by deflating the 3 month US treasury rate by the annual US Consumer price inflation. A fall in the interest rate could lead to investors investing in countries with higher yields. Date Source: World Bank and OECD.
RLUS – It is the Real US long term interest rate calculated by deflating the 10 year US rate by the annual Consumer price inflation. According to Byrne and Fiess(2015), the long term rate has not been widely used. However, investors may choose to diversify their portfolio by investing in assets of different maturities. Data Source: World Bank and OECD
RCP – It is the non-oil commodity prices Index. Reinhart and Reinhart (2009) state that as most emerging countries are exporters of non-oil commodity, an increase in the prices shall attract investors. Data Source: IMF
Y(G7) – It is the growth in the Real GDP of G7 countries. A slowdown in growth of G7 countries, leads to expansion of capital inflows in other economies, especially growing economies as investors chase higher returns. Data Source: OECD, World Bank and IMF. Country Specific Variables
FO – It is a measure to capture the extent of openness in the capital account and capture the intensity of capital controls. We use the Chinn and Ito Index (2006) which was developed specifically to measure the intensity of capital controls, as the existing measures did not fully capture the metric. Data Source: Chinn and Ito Index
G – Government Effectiveness is a World Bank Indicator, which measures the perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies. This variable is used in lieu of the PRS data used by Byrne and Fiess(2015) to capture the effect of institutions. Gani(2011) had found Government effectiveness to be positively associated with growth. Data Source: World Bank
18
HC – Human capital measures the average number of year of years of schooling for each country. It measures the average number of years for both sexes aged 25 and above. Lucas (1990) explained the role of human capital to explain the paradox of why capital does not flow from rich to poor countries. Low human capital may be a bar to capital inflows. Low average year of schooling for population aged 25 and above suggests low human capital for a country. We use the same data source as used by Byrne and Fiess(2015) to measure Human Capital. Data Source: Institute for health metrics and evaluation data from Gap Minder.
L – Rule of Law is a World Bank Indicator which measures the perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence. The variable is used in combination with Government Effectiveness in lieu of the PRS data used by Byrne and Fiess(2015) to capture the effect of institutions. Chinn and Ito (2006) had found general legal/institution enhance the effect of capital account opening strongly that finance related legal/institutional variables, which implies investors have more confidence in countries with strong rule of law. Data Source: World Bank
ΔY – The variable measures the economic growth in country by growth in real GDP. Data Source: World Bank and IMF
R – The variable measures the National Real Interest Rate of country at time t. Real interest rate is the lending interest rate adjusted for inflation as measured by the GDP deflator. The terms and conditions attached to lending rates differ by country, however, limiting their comparability. Date Source: World Bank and IMF.
DGDP – The variable measures the Gross Government Debt to GDP(nominal) ratio. It measures the financial leverage of an economy, and high debts can have harmful effect on the economy. Forbes and Warnock (2012) is their study of extreme capital flow episodes found flight episodes were associated with countries with low debt and high financial controls. Including the variable will help us understand whether government indebtedness is associated with gross capital inflows, considering one of the Euro convergence criteria was having a government debt to ratio of 60%, and euro countries like Italy, today have a Government Debt to GDP ratio in excess of 100%. Data Source: IMF WEO
Descriptive Statistics
We now look at the descriptive statistics for each income group, for dependant and explanatory variables.
19 High Income Group
1. Aggregated Flows -
Table 2 – High Income Group Descriptive Statistics
Most of the variables appear to be close to a normal distribution, however, two variables – CF/GDP and R (%)- have a high kurtosis which indicate that the distribution may be non-normal. To reduce the kurtosis and make the distribution more normal, we apply the log
transformation on the two variables. The log transformation helps in reducing the kurtosis of the two variables and making the distribution appear closer to normal. The interpretation of the model will be that apart from R, if we change the any other variable in the equation by 1 unit, we would expect our variable CAPit to change by 100* α percent, while for a 1% change in R we would expect our variable CAPit to change by α percent.
Though we are using Panel method to specify our model which decreases multi-collinearity between explanatory variable, we still do a correlation check between the dependant and explanatory variables for each income group, to understand the extent of association between the variables. The result for aggregated gross inflows for high income group is seen below –
Table 3 – High Income Group
We see that RLUS and RSUS are highly correlated with each other indicating that the long-term as well as short term rates tend to be positively associated and increase in parallel. Similarly, we see that Government effectiveness and Rule of Law are strongly correlated and tend to increase
CF/GDP (%) VIX RSUS (%) RLUS(%) FO HC G L DeltaY(%) DGDP(%) R(%) RCP Y(G7)(%)
Mean 13.0% 20.74 1.6% 2.7% 0.92 11.05 1.31 1.26 2.4% 56.1% 3.8% 130.45 2.2% Median 8.4% 19.92 0.6% 2.3% 1.00 11.70 1.50 1.35 2.4% 49.5% 3.5% 133.96 2.4% Standard Deviation 31.2% 6.36 1.9% 1.3% 0.17 2.40 0.65 0.56 3.0% 39.5% 6.4% 37.61 3.8% Kurtosis 37.63 -1.13 0.05 -0.72 4.00 0.49 -0.69 -1.21 2.78 4.74 91.98 -1.38 -0.47 Range 5.31 19.24 0.06 0.04 0.84 10.73 2.82 2.16 0.26 2.49 1.14 114.43 0.14 Minimum -3.04 12.55 0.00 0.01 0.16 3.94 -0.39 -0.04 -0.08 0.00 -0.20 75.97 -0.05 Maximum 2.27 31.79 0.06 0.05 1.00 14.67 2.43 2.12 0.17 2.49 0.94 190.40 0.09 N 461 462 462 462 462 462 462 462 462 462 453 462 462
CF/GDP VIX RSUS RLUS FO HC G L DeltaY DGDP R RCP Y(G7)
CF/GDP 1.00 VIX -0.19 1.00 RSUS 0.24 -0.16 1.00 RLUS 0.18 0.03 0.95 1.00 FO 0.08 -0.03 -0.03 -0.05 1.00 HC -0.02 -0.04 -0.09 -0.11 0.13 1.00 G 0.15 0.00 0.00 0.00 0.44 0.62 1.00 L 0.12 -0.01 -0.03 -0.03 0.46 0.60 0.94 1.00 DeltaY 0.22 -0.35 0.23 0.12 -0.24 -0.22 -0.17 -0.20 1.00 DGDP -0.08 -0.02 -0.11 -0.11 0.29 0.21 0.12 0.05 -0.29 1.00 R -0.11 0.15 0.01 0.08 -0.03 -0.06 -0.07 -0.11 -0.22 0.08 1.00 RCP -0.03 -0.07 -0.26 -0.42 0.10 0.20 -0.04 0.03 -0.04 0.08 -0.17 1.00 Y(G7) 0.18 -0.24 0.11 -0.04 -0.01 -0.07 0.03 0.01 0.44 -0.04 -0.10 -0.19 1.00
20
in parallel, which implies that as rule of law increases, government effectiveness increases as well. Human Capital is also positively associated with Government effectiveness and rule of law, which implies that as government effectiveness and rule of law increase or decrease, Human capital also increases or decreases accordingly. Apart from correlations among these variables, there is no significant correlation (>=0.5) among the other variables. The dependant variable is not strongly correlated with any of the explanatory variables. High correlations between variables can give rise to endogeneity. In the presence of endogeneity, OLS can produce biased results and inconsistent parameter estimates, which can result is misleading hypothesis testing. To account for the same, we will run the regression model with all variables and drop one factor out of the correlated pair to see if the regression results change substantially. If not, we can ignore concerns of endogeneity in the model.
To check for unit roots in the panel, we employ the Augmented Dickey Fuller test for each individual series. For individual series, which show the presence of a unit root, we take the first difference of the variables and check again for unit roots. In case the series is stationary, we take the first difference of the variable in the regression. If, the series still has a unit root, we take the second difference and check for unit roots, till the series is stationary. Variables that have a unit root are summarized in Appendix 1. We use panel regression (Fixed effects) to estimate our model. Generally, in panel regression, if the time component does not excessively exceed the cross-section component, fixed effects should suffice. Autocorrelation and heteroscedasticity are two problems that occur in panel regression. To control for both, we estimate the equation using robust standard error estimates which controls for cross-section heteroscedasticity and
correlation within a cluster.
The descriptive statistics for the explanatory variables of disaggregate flows is the same as that for aggregate flows. The descriptive statistics for the dependent variables is presented below –
Table 4 – Disaggregate gross capital inflows (Dependant Variables)
P_Inflow/GDP L_Inflow/GDP FDI/GDP
Mean 0.038 0.042 0.054 Median 0.031 0.019 0.029 Standard Deviation 0.051 0.274 0.089 Kurtosis 7.010 61.327 21.686 Minimum -0.239 -3.163 -0.161 Maximum 0.358 2.159 0.874 Count 450 457 457
21
Of the three dependent variables, we see that Loan/GDP and FDI/GDP have high kurtosis and may not have a normal distribution. We will apply appropriate transformation technique to make the distribution of both variables closer to normal. We also do a check for correlation between the dependant and explanatory variables and find no significant correlation between the variables.
Upper Middle Income Group
The descriptive statistics of the dependant and explanatory variables is provided below.
Table 5 – Upper Middle Income Countries
Most of the variables have a distribution close to normal. However, there are certain variables which have a very high kurtosis, which indicates that the distribution may be non-normal. To bring the distribution closer to normal we take logs of these variables in our regression equations.
We do a correlation check to understand the extent of association between the explanatory variables, and find that US short term and long term interest rates are highly correlated. We also find high correlation between Government effectiveness and rule of law, human capital and government debt to GDP. We also check for correlation between dependant variables and explanatory variables and find that dependant variables (aggregate and disaggregate flows) have no correlation with the explanatory variables. The correlation table is shown below -
CF/GDP FDI/GDP P_Inflow/GDP Loans/GDP VIX RSUS RLUS FO HC G L DeltaY DGDP R Y(G7) RCP
Mean 0.07 0.04 0.01 0.02 20.74 0.02 0.03 0.45 8.32 -0.22 -0.46 0.05 0.34 0.08 0.02 130.45 Median 0.06 0.03 0.01 0.01 19.92 0.01 0.02 0.41 7.79 -0.22 -0.55 0.05 0.35 0.05 0.02 133.96 Standard Deviation 0.08 0.06 0.02 0.04 6.37 0.02 0.01 0.30 2.30 0.58 0.54 0.05 0.16 0.12 0.04 37.64 Kurtosis 20.44 36.29 5.19 9.78 -1.13 0.07 -0.71 -0.95 -0.42 -0.35 -0.72 8.43 0.72 2.84 -0.46 -1.38 Skewness 3.61 5.27 0.67 1.54 0.36 1.17 0.76 0.46 0.29 0.25 0.27 1.71 0.59 0.66 -0.08 -0.12 Minimum -0.08 -0.06 -0.09 -0.13 12.55 0.00 0.01 0.00 3.10 -1.46 -1.63 -0.08 0.06 -0.34 -0.05 75.97 Maximum 0.65 0.55 0.11 0.27 31.79 0.06 0.05 1.00 13.10 1.25 0.64 0.35 1.02 0.48 0.09 190.40 N 266 266 244 261 266 266 266 266 266 266 266 266 262 230 266 266
22
Table 6 - Upper Middle Income Group (Correlation Matrix)
Considering the high correlation between US short term and long term interest rates, multi collinearity can arise in our model. High correlations between variables can give rise to
endogeneity. In the presence of endogeneity, OLS can produce biased results and inconsistent parameter estimates, which can result is misleading hypothesis testing. To account for the same, we will run the regression model with all variables and drop one factor out of the correlated pair to see if the regression results change substantially. If not, we can ignore concerns of
endogeneity in the model. ADF test will be done on all variables to check for the presence of unit root. In case unit roots are present, we will take the first difference of the variables and do an ADF test till we find that the series is stationary. Variables that have a unit root are
summarized in Appendix 2. We use panel regression (Fixed effects) to estimate our model. Autocorrelation and heteroscedasticity are two problems that occur in panel regression. To control for both, we estimate the equation using robust standard error estimates which controls for cross-section heteroscedasticity and correlation within a cluster.
Lower Middle Income Group
The descriptive statistics of the dependant and explanatory variables is provided below.
Table 7 – Lower Middle Income Group Countries
CF/GDP F/GDP P_Inflow/GDP Loans/GDP VIX RSUS RLUS FO HC G L DeltaY DGDP R RCP Y(G7) Mean 0.05 0.03 0.01 0.01 20.74 0.02 0.03 0.38 5.95 -0.47 -0.60 0.05 0.51 0.06 130.45 0.02 Median 0.04 0.02 0.00 0.01 19.92 0.01 0.02 0.16 5.55 -0.45 -0.60 0.05 0.46 0.06 133.96 0.02 Standard Deviation 0.05 0.02 0.02 0.04 6.37 0.02 0.01 0.32 2.21 0.42 0.47 0.03 0.27 0.07 37.64 0.04 Kurtosis 4.21 1.53 3.12 11.97 -1.13 0.07 -0.71 -0.64 1.41 0.43 -0.94 9.38 5.93 10.00 -1.38 -0.46 Minimum -0.19 -0.05 -0.04 -0.25 12.55 0.00 0.01 0.00 2.42 -1.65 -1.68 -0.15 0.10 -0.42 75.97 -0.05 Maximum 0.27 0.10 0.08 0.16 31.79 0.06 0.05 1.00 13.15 0.63 0.32 0.14 2.16 0.24 190.40 0.09 Count 265 265 237 265 266 266 266 266 266 266 266 266 255 214 266 266
VIX RSUS RLUS FO HC G L DeltaY DGDP R Y(G7) RCP
VIX 1.00 RSUS -0.16 1.00 RLUS 0.03 0.95 1.00 FO -0.01 0.00 -0.03 1.00 HC -0.04 -0.10 -0.13 0.09 1.00 G -0.02 -0.05 -0.06 0.01 -0.14 1.00 L -0.02 -0.06 -0.07 0.12 -0.10 0.88 1.00 DeltaY -0.29 0.31 0.22 -0.25 0.01 -0.23 -0.24 1.00 DGDP 0.03 -0.11 -0.04 -0.10 -0.59 0.33 0.32 -0.26 1.00 R 0.01 -0.03 0.01 0.24 -0.33 -0.02 0.04 -0.13 0.24 1.00 Y(G7) -0.24 0.11 -0.04 0.00 -0.08 -0.02 -0.02 0.34 0.01 -0.12 1.00 RCP -0.07 -0.26 -0.42 0.12 0.22 0.06 0.06 -0.06 -0.24 -0.11 -0.19 1.00
23
The kurtosis of most of the variables indicates that the distributions are close to normal, with one variable exhibiting a high kurtosis. We will use the appropriate transformation method to make the distribution closer to normal.
We do a correlation check to understand the extent of association between the explanatory variables, and find that US short term and long term interest rates are highly correlated. We also find high correlation between Government effectiveness and rule of law. We also check for correlation between dependant variables and explanatory variables and find that dependant variables (aggregate and disaggregate flows) have no correlation with the explanatory variables.
Table 8 – Lower Middle Income Group (Correlation Matrix)
Considering the high correlation between US short term and long term interest rates and
between government effectiveness and rule of law, multi collinearity can arise in our model. High correlations between variables can give rise to endogeneity. In the presence of endogeneity, OLS can produce biased results and inconsistent parameter estimates, which can result is misleading hypothesis testing. To account for the same, we will run the regression model with all variables and drop one factor out of the correlated pair to see if the regression results change substantially. If not, we can ignore concerns of endogeneity in the model. ADF test will be done on all
variables to check for the presence of unit root. In case unit roots are present, we will take the first difference of the variables and do an ADF test till we find that the series is stationary. Variables that have a unit root are summarized in Appendix 2. We use panel regression (Fixed effects) to estimate our model. Autocorrelation and heteroscedasticity are two problems that occur in panel regression. To control for both, we estimate the equation using robust standard error estimates which controls for cross-section heteroscedasticity and correlation within a cluster.
VIX RSUS RLUS FO HC G L DeltaY DGDP R RCP Y(G7)
VIX 1.00 RSUS -0.16 1.00 RLUS 0.03 0.95 1.00 FO 0.04 0.07 0.07 1.00 HC -0.04 -0.09 -0.11 0.01 1.00 G 0.01 0.05 0.06 0.00 0.16 1.00 L -0.01 0.02 0.03 0.01 0.12 0.82 1.00 DeltaY -0.19 0.16 0.09 -0.09 -0.16 -0.15 -0.09 1.00 DGDP -0.02 0.03 0.08 -0.20 -0.10 0.12 0.26 0.12 1.00 R 0.06 -0.02 0.06 0.11 -0.23 -0.08 -0.08 -0.21 -0.16 1.00 RCP -0.07 -0.26 -0.42 -0.08 0.20 -0.05 -0.03 -0.02 -0.34 -0.26 1.00 Y(G7) -0.24 0.11 -0.04 0.04 -0.07 0.03 0.00 0.25 0.10 -0.13 -0.19 1.00
24 5. Empirical Findings
Empirical results and their interpretation is presented for the various Hypothesis described in section 3.
Hypothesis 1: Push Factors -US short-term and long term interest rates, Global Risk as measured by VIX and growth of G7 countries – will be significant in explaining gross capital inflows to high income countries, as most of the countries have well-functioning domestic institutions and macroeconomic policies.
Hypothesis 2: For Upper Middle and Lower Middle income countries, both push and pull factors will be significant in explaining gross capital inflows. Pull factors will vary between the two income groups. Domestic economic factors – real interest rate, growth in real national GDP- will be significant for upper middle income countries, while structural factors – human capital, financial openness, government effectiveness, rule of law – will be significant.
The results for Hypothesis 1 and Hypothesis 2 are provided tables in Appendix 3. Empirical results support the hypothesis that global factors are significant in explaining capital flows to high income countries. US rates – long term and short term are significant in explaining gross capital inflows to High income countries. Significance of US short term rate implies that the US Monetary policy which operates through US short term interest rates (Byrne and Fiess, 2015), significantly impacts flows in high income countries. Growth in G7 countries is also a significant factor at 5% and 10% significance level. A surprising result is the significance of government effectiveness in explaining gross capital inflows to high income countries. High income countries are mostly advanced economies with well-developed financial and governance institutions, so the significance of the variable is surprising. When we consider that High-income countries as defined by World Bank, also include countries termed as developing by IMF, the significance of the variable can be understood, as countries in Euro Area, UK, US have institutions which are significantly different from that of Chile, Uruguay and Saudi Arabia. As the coefficient of the variable is positive, it indicates that countries with better governance will attract more gross capital inflows, which implies that investors will prefer to invest their money in countries with well-developed governance. This finding is in line with the existing literature – Globerman and Shapiro (2002), Chinn and Ito(2006) and Byrne and Fiess (2015).
Results for Upper Middle income countries indicate that country specific factor – growth in the real GDP- is a significant factor in attracting capital along with the national real interest rate and debt to gdp, which are significant at the 10% level. We find no role of global factors in
25
upper middle income countries, along with government backed debt and interest rates make it attractive for investors to invest in these countries. A negative coefficient for real interest rate indicates, that as interest rate fall, asset prices rise, which makes the countries attractive for foreign investors. The findings are in line with our hypothesis that GDP growth and interest rates will pull capital to these countries. This is in line with existing literature on capital flows to emerging markets, most notably the studies done by Jevack et al (2010), which studied flows into Emerging Europe, which consist mostly of upper middle income countries. US short term and US long term interest rates do not play an important role in attracting capital flows. This indicates that the US Monetary policy as measured by the US short term interest rate does not have a bearing on capital inflows for this group. For lower middle income countries, we find growth in G7 countries and national real GDP growth to be significant factors in attracting capital. Both variables have opposite coefficient sign, which suggests that as growth slows down in G7 countries, investors look at diversifying their holdings by investing in lower income countries. This can be a potential problem for these countries as capital that flows will be temporary and can reverse itself if G7 countries experience growth. Our findings for lower middle income countries is in line with literature on episodes of capital flight (Forbes and Warnock 2012). However, we do not find any evidence in support of structural factors being important for attracting capital flows in low income countries, which is contrary to the literature of financial development and FDI flows.
Hypothesis 3: Global factors – Risk, US Short term and long term interest rate- and country specific factors – real national interest rates, Debt to GDP, GDP growth – will be significant in explaining gross portfolio and loan inflows for all income groups.
Hypothesis 4: Global factors- Growth in G7 countries, US interest rates- and country specific factor – growth in real GDP- will be significant factors for FDI in high income countries. Price of non-oil commodities, VIX and country specific factors – financial openness, government effectiveness, rule of law – will be significant factors for FDI in upper middle income countries. Risk, US interest rates, and country specific factors – GDP growth, Human Capital, Rule of law, Financial openness- will be significant factors for FDI in lower middle income countries.
The results for the hypothesis are provided in the tables in Appendix 4 for High Income Group, Appendix 5 for Upper Middle Income Group and Appendix 6 for Lower Middle Income Group. Empirical findings for High Income group support out hypothesis that loan and portfolio flows are more sensitive to global factors especially Risk and US interest rates. Loan flow to high income countries is significantly affected by the US interest rates, while VIX and
26
growth in G7 countries have significant impact on portfolio flows, which are largely volatile compared to FDI flows. Apart from global factors, national real interest rates are also significant at the 10% level in explaining portfolio inflows. Human Capital and Government effectiveness are also significant for portfolio flows for high income countries, which implies that high income countries have significant difference in human capital and governance structures. The coefficient of human capital is negative while that of government effectiveness is positive, which implies that poor human capital deters portfolio flows while improvement in government effectiveness attracts portfolio flows. This implies that apart from global factors, countries in high income groups may have to work on domestic factors especially structural aspects in attracting portfolio flows. The findings for high income countries for portfolio and loan flows are in line with extant literature on role of institutions, (Byrne and Fiess , 2015), (Globerman and Shapiro, 2002) and capital flows (Forbes and Warnock, 2012), (Chuhan etl, 1998), (Byrne and Fiess, 2015). FDI flows to High income countries are significantly affected by US interest rates and debt to GDP. Non-oil commodity price index is also significant at the 10% level; however, the coefficient is quite small. On the other hand, high income countries have government backed debt policies, and increase in the debt to GDP coefficient attracts FDI to these countries. Most of the high income countries have well developed institutions and they do not have any significant effect on FDI inflows.
For Upper middle countries we find Risk, growth in G7 countries and debt to GDP as significant factors in attracting portfolio flows. This is in line with the existing literature on capital flows and supports our hypothesis. For bank loans, non – oil commodity prices are significant. Most of the upper middle income countries comprise of non-oil commodity
exporters hence, increase in commodity prices, results in an increase in bank flows. The finding is in line with findings of Byrne and Fiess (2015). For FDI, we find human capital, growth in GDP and national interest rates to be significant factors in attracting investments. Human capital has a significant positive coefficient, which implies that countries with better human capital will attract significantly higher capital flows compared to countries with low human capital. Similarly, countries with a faster GDP growth will attract more investments, as investors look for higher returns on their investments. Our findings are in line with the extant literature on FDIs and capital flows.
For lower middle income countries, we find the growth in national GDP, Risk and national interest rates to be significant in attracting loan and portfolio flows. Cross border bank loans are affected primarily by national GDP growth, which indicates that loans tend to flow into
27
countries which have a higher growth rate. Apart from GDP growth rate, no other factor appears to be significant. For portfolio flows, Risk, national interest rate and national GDP are significant. As global risk decreases, more investors tend to invest in low income countries, which indicates that portfolio flows are highly volatile, and can reverse if global risk increases. Our finding is in line with existing literature (Jevack et al, 2010), (Forbes and Warnock, 2012), (Fratzcher, 2012) who find global risk to be a significant factor in capital flows. For FDI, global factors – US interest rates, growth in G7 countries and non-oil commodity price – and country effects – human capital – are significant in explaining FDI inflows. FDI flows into low income countries is to take advantage of labour arbitrage and hence we find that human capital has a significant effect on attracting FDI flows. We also see that country specific macroeconomic factor and financial openness do not play a significant role in attracting FDI. These findings are in line with the research done by Walsh and Yu (2010).
Empirical findings highlight the importance of GDP, which is a measure of the economic size of a country. The growth of G7 countries which cumulatively account for almost 1/3rd of the global
GDP, as well as national GDP growth emphasize that capital follows economies that are growing, and decline in the economic output of a country tends to reverse capital inflows. The GDP tends to give an overall picture of the health of an economy, and is an indicator used by investors to invest capital in a country. Decline in economic output will cause investors to pull out money or invest new capital in places where the economic output is increasing. Countries which tend to show an increase in economic output owing to national macroeconomic policies, will tend to pull capital from countries affected by national as well as global slowdown. The findings for all income group highlights this trend, and has a significant effect in attracting capital inflows.
Overall, the empirical findings tend to support our hypothesis for upper middle and lower middle income countries. However, for High income countries we find structural factors play an important role in attracting portfolio inflows, which is a departure from our hypothesis.
6. Conclusion
Empirical results suggest that global factors as well as national factors play a role in attracting capital inflows. Global risk, US interest rates, GDP growth, National real interest rates are significant factors explaining aggregate and disaggregate capital inflows. We also find that capital control aka financial openness is not a significant factor in attracting inflows, which suggests that capital control may not be a huge influencer for inflows, though policy makers in various
28
studied gross capital inflows and not outflows, the efficacy of using capital control as a policy tool cannot be commented upon.
Empirical results show that there is no one specific determinant which can explain gross capital inflows for a heterogeneous group of countries, which is in line with the existing literature. We also find that investor preferences vary among the various income groups, and our empirical findings suggest that investments in various income groups are driven by different factors and are a combination of global, national macroeconomic and structural factors. The empirical findings suggest that Risk and Economic growth are prime drivers for portfolio flows in upper middle income countries, which indicates that portfolio flows are influenced by the investors’ perception of global risk and the economic growth – national as well as of the G7 countries. The results also suggest that national factors play an important role and it will be useful for policy makers to develop policies and institutions to enhance these factors which can enhance the attractiveness of a country and attract more capital inflows. Some of the key factors that were found to be significant in the study and can be influenced by various national policies are the national real interest rate, GDP growth and human capital.
The paper studied various factors which influence gross capital inflows for a country based on its income level, for the time period 2001 – 2014. The time period 2001-2014 was chosen, as data was available for most of the countries for this period. The frequency of the study was annual as data for some variables was available annually. Thus, the number of observations was limited and there was some missing data in a few panels for a very small number of countries. Within the bounds of data limitations, the results of the study are in line with the existing literature, which have studied capital flow based on different measures, factors and time series. In conclusion, it may be implied that gross capital inflow - foreign investor preferences- are driven by the
investment theory of risk-return, and most investors will tend to invest in countries which gives them the requisite returns for the risk that they take.
29 7. References
1. Azmat Gani, 2011. "Governance and Growth in Developing Countries," Journal of Economic Issues, vol. 45(1), 19-40.
2. Byrne, Joseph P, Fiess Norbert, 2015. International capital flows to Emerging Markets: National and Global Determinants. Journal of International Money and Finance 61 (2016), 82-100.
3. Calvo, Guillermo, Leiderman, Leonardo, Reinhart, Carmen, 1993. Capital inflows and real exchange rate appreciation in Latin America: the role of external factors. IMF Staff Papers 40 (1), 108–151.
4. Calvo, Guillermo (1998). Capital flow and market crises: the simple economics of sudden stops. Journal of Applied Economics 35-54(Nov).
5. Calvo, Guillermo, Izquierdo, Alejandro, Mejía, Luis-Fernando, 2004. On the empirics of sudden stops: the relevance of balance-sheet effects. NBER Working Paper 10520. 6. Calvo, Guillermo, Izquierdo, Alejandro, Mejía, Luis-Fernando, 2008. Systemic sudden
stops: the relevance of balance-sheet effects and financial integration. NBER Working Paper 14026.
7. Chuhan, Punam, Claessens, Stijn, Mamingi, Nlandu, 1998. Equity and bond flows to Latin America and Asia: the role of global and country factors. Journal of Development Economics 55 (2), 439–463.
8. Chinn, Menzie D. and Hiro Ito (2006). "What Matters for Financial Development? Capital Controls, Institutions, and Interactions," Journal of Development Economics, Volume 81, Issue 1, Pages 163-192.
9. Cowan, Kevin, De Gregorio, José, Micco, Alejandro, Neilson, Christopher, 2008. Finan- cial diversification, sudden stops and sudden starts. In: Cowan, Kevin, Edwards, Sebastian, Valdés, Rodrigo (Eds.), Current Account and External Finance. Central Bank of Chile.
10. Caballero, Ricardo, Farhi, Emmanuel, Gourinchas, Pierre-Olivier, 2008. An equilibrium model of ‘global imbalances’ and low interest rates. The American Economic Re- view 98 (1), 358–393
11. Cristina Jude, Gr´egory Levieuge. Growth Effect of FDI in Developing Economies: the Role of Institutional Quality. DR LEO 2013-11. 2014.
12. Fratzcher, Marcel, 2011.Capital Flows, Push versus Pull factors and the Global Financial crisis. ECB, Working Paper Series No. 1364.
30
13. Forbes, Kristin J, Warnock, Francis E, 2012. Capital Flow waves: Surges, stop, flight and retrenchment. Journal of International Economics 88 (2012), 235-251.
14. Griffin, John, Nardari, Federico, Stulz, Rene, 2004. Daily cross-border flows: pushed or pulled? The Review of Economics and Statistics 86 (3),641–657.
15. Globerman, Steven, Shapiro, Daniel, 2002. Global Foreign Direct Investment Flows: The role of Governance Infrastructure. World Development 30(11), 1899-1919 16. Jevack, Anton, Setzer, Ralph, Suardi, Massimo, 2010. Determinants of Capital flows to
new EU Member states before and during the Financial Crisis. European Economy, Economic Papers 425.
17. Lucas, Robert, E. Jr, 1990. Why doesn’t Capital flow from Rich to Poor countries? The American Economic Review 80(2), 92-96.
18. Mendoza, Enrique, Quadrini, Vincenzo, Rios-Rull, Jose Victor, 2009. Financial integra- tion, financial development and global imbalances. Journal of Political Economy 117 (3), 371–416.
19. Milesi-Ferretti, Gian Maria and Cedric Tille, 2010. The Great Retrenchment: International Capital flows during the Global Financial Crisis. Mimeo.
20. Nielsen, Lynge,2011. Classification of countries based on their development: How is it done and how it could be done. IMF Working Paper, WP/11/31.
21. North, Douglas C, (1994). Economic performance through time. The American Economic Review, 84(3), 359-368.
22. Olaberría, E. (2014), “US Long Term Interest Rates and Capital Flows to Emerging Economies”, OECD Economics Department Working Papers, No. 1155, OECD Publishing, Paris.
23. Reinhart, Carmen, Reinhart, Vincent, 2009. Capital flow bonanzas: an encompassing view of the past and present. In: Frankel, Jeffrey, Giavazzi, Francesco (Eds.), NBER International Seminar in Macroeconomics 2008. Chicago University Press, Chicago. 24. Rothenberg, Alex, Warnock, Francis E., 2011. Sudden flight and true sudden stops. Re-
view of International Economics 19 (3), 509–524.
25. Walsh, James,P, Yu Jiangyan, 2010. Determinants of Foreign Direct Investment:A sectoral and institutional approach. IMF Working paper WP/10/187.
26. Zlate Andrei, Ahmed Shaghil, 2014. Capital Flow to Emerging market economies: A brave new world? Journal of International Money and Finance 48(2014) 221-248
