U
NIVERSITY VAN
A
MSTERDAM
Amsterdam Business School
The difference in influence of foreign direct investments on
systemic risk between developed and transition countries in
Europe
Student: Kseniia Vasileva Student number: 11088893
MSC Business Economics, Finance track Master Thesis
2
Abstract
Nowadays there is no common opinion about the influence of FDI flows on countries. Large capital flows can lead to monetary expansion, excessive real exchange rate appreciation and the increase of a current account deficit, the increase in demand for nation’s currency and because of the pressure on interest rate in the investment-recipient country (Calvo, Leiderman, Reinhart (1996), Kaminsky and Reinhart (1999), Reinhart and Rogoff (2009), Rey (2013)). While some authors claim that sharp surge in FDI can have negative influence and lead to the crisis in developing (transition) countries, the others argue that FDI can cause more damage in developed ones. The aim of current thesis is to explore this influence in both groups of countries in Europe.
Statement of Originality
This document is written by Kseniia Vasileva 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
Content
Introduction ... 4
Literature review ... 5
FDI as a source of risk for developing countries ... 6
FDI as a source of risk for developed countries ... 8
Methodology ... 10 Systemic risk ... 10 Data ... 12 Results ... 15 Robustness check ... 18 Conclusion ... 19 Appendix ... 21 References ... 22
4
Introduction
Nowadays there is no common opinion about the influence of Foreign Direct Investments on the countries. While some economists claim that FDI can lead to currency appreciation and increase in exchange rate for developing countries (Ba (2015), Calvo et al (1993,1996)) and lead to a crisis, the others suggest FDI can cause more damage for developed countries since they invest in developing countriesproviding risky investments (Dooley et al (2004), Gourinchas et Rey (2007)).
In this thesis, I would like to check these versions and come to the conclusion what is the influence of the FDI on different countries. Systemic risk is taken as the measure to evaluate the impact of FDI on a particular country. Systemic risk is a propensity of a bank/firm to be undercapitalized while the whole economic system is also undercapitalized. Taking this variable as a dependent is the novelty of the thesis. In the literature mentioned above (Ba (2015), Calvo et al (1993,1996)) crisis is usually taken as the dependent variable. It is a dummy variable with 1 if there is a crisis in a particular year and 0 otherwise. Taking the systemic risk instead of a crisis could allow taking a non-discrete variable as the dependent, to widen a sample (because not in all countries there was crisis, but systemic risk can show that there were problems). Also, there is a possibility that dependence between the FDI and systemic risk is not linear (Calvo, 1993). Perhaps, till some point the influence of the FDI is positive (FDI is a source of capital and can stimulate economic growth because it provides such an externalities as technology and business know-how transfers (Romer (1993), Makki (2004). Herrero and Simon(2003), Carkovic and Levine (2002), Borensztein et al (1998)). However when the abnormal amount of FDI is accumulated the consequences turn into negative ones.
The research question of this thesis is the difference in influence of foreign direct investments on systemic risk between developed and developing countries in Europe. The empirical analysis is made by running regressions on panel data consisted of 28 European countries through the period 2002-2011.
Systemic risk includes the bank’s market capitalization, the sensitivity of its equity return to market shocks, and its financial leverage (see methodology).1
FDI inflows and outflows comprise of the equity capital, reinvested earnings and intra-company loans2.
1
The definition by HEC Lausanne and the NYU Stern Volatility Institute: http://www.crml.ch/index.php?id=44
5
After the increase in FDI inflows a country-recipient of the investments experiences the expansion of the money supply and high levels of debt-based assets purchased by foreigners (Ba, 2016) that causes increase in asset prices and lower interest rates. According to macroeconomic theory, if a country has the market interest rate lower than natural rate provided by real sector productivity, then credit and assets bubble appear (Borio and Disyatat, 2011). That might lead to the drop in the stock market. Then, the market capitalization of a particular bank becomes extremely low (the regulators’ requirement could not be satisfied), the bank could be undercapitalized and need to raise additional equity through financial institutions or from the market. However, the market is in a decline (especially the credit market, if there was a credit bubble which busted), so, the bank actually cannot refinance itself and could default. This is systemic risk. Without any appropriate regulator’s intervention the chain of banks default can occur.
According to R. Aliber (2014) the crisis is triggered when one or several lenders become more suspicious and cautious in extending and providing loans; then prices of the currency’s decline.
The rest of the thesis has the following structure: part 1 analyzes current literature on the topic of the research, part 2 suggests the methodology used in the research, part 3 describes the dataset and shows descriptive statistics with preliminary correlation analysis allowing to build up hypotheses, part 4 shows results and robustness tests, part 5 discusses results and opportunity for future research and applications.
Literature review
According to Charles Kindleberger’s Hegemonic Stability Theory (1973) the world economic system should be stable if there would be one hegemonic country in the center (the country disproportionally well-connected with others and having the strongest economic system). In such hierarchical system the central agent is capable of absorbing shocks and negative events from the other agents (countries) keeping the whole system stable. However, this system did not show such results throughout history. Moreover, the negative events in central (hegemonic) economy spread out through capital channels to the less developed economies causing crises there (Oatley et al, 2013). When the hegemonic country experiences upward period in financial cycle, it increases credit and debts and raise market liquidity. These circumstances cause investment go abroad and influence market interest rate of recipient countries by pushing it down. If this decrease of market interest rate reaches the level below one provided by the productivity of economy, it leads to credit or asset bubbles which in its turn can lead to crises (Ba, 2016).
2 From UNCTAD
6
FDI as a source of risk for developing countries
Ba (2016) explores the influence of one country on others through the capital flows. She advances Kindelbeger’s theory about hegemonic world financial system with one or several central countries (the strongest economies like the US and the UK) and the rest – peripherals (like developing countries, the countries with weaker economic system). According to one of her hypothesis the increase in capital flows should raise the probability of crises in peripheral countries.
Before the crisis of 2008 US decreased interest rates significantly (from 6% in January, 2001 to 1% in June 2003). This reduction provided strong capital inflows to developing countries with higher interest rates and also with better opportunities to invest (FDI inflows almost doubled for Turkey from $38565mln in 2004 up to $71302mln in 2005). In the second half of 2013, the Federal Reserve began the reduction of its quantitative easing program and as a consequence there was a negative effect for emerging countries because of the immediate FDI outflow. For example, there was a shot up in the cost of Turkey government bonds (around 370 basis points) and a drop in Indonesia’s Jakarta Composite Index (approximately 20%). H. Ba analyzes this event by making a parallel with the position of Great Britain in the world during the 19th century. The UK had the most significant and interconnected economy. The periods when the interest rates were low and credits were easy to obtain coincides with an increase in foreign investments into developing countries. During the period from 1881 to 1913 in the years with low interest rates of the Bank of England (from 0,99 till 2,68%) there were 41 surges3 in capital outflows from the UK comparing with 17 in the years with higher interest rates. These rises were followed by financial crises (1890, 1900-1901, 1907-1909). In the United States the three largest crises which took place in 1873, 1890 and 1907 were also preceded by low interest rate and surges in capital flows from the UK. Thus, the economic situation in Britain influenced the US economy through the capital flows.
The research is concentrated on Britain and its 25 economic partners during the XIXth century. To test the hypothesis about influence of capital surges on economies, H. Ba runs logit models with dependent variable – crisis (1 if there is a crisis in a particular country during a given year, 0 – otherwise). The main independent variable – capital flow surge from Britain with a lag of two years (because a bubble is developing over long periods of time). As the result, the probability to have a crisis in two years after the capital surge is two times higher than without the capital surge. H. Ba concludes that economic circumstances in one country can spillover to many other though interdependences provided by cross-border capital flows.
3
Surge is defined as a change in capital flows which is greater than one standard deviation above the average change of capital flows (Ba, 2015)
7
The work of Calvo et al (1993) provides analysis of capital inflow consequences in Latin America (Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Mexico, Peru, Uruguay and Venezuela) in 1980-s and 1990-s,and its influence on exchange rate appreciation. According to the authors the surge in FDI inflows lead to the increase in consumption in country-recipient of the investments. When the increase in spending relates to non-traded goods, their relative price will increase and the real exchange rate appreciates. Thus FDI cause appreciation in real exchange rates, and also lead to boom in the stock and real estate markets. The most remarkable capital inflows to Latin America were during the 1920s and 1978-81. They were followed by economic crises and capital outflows in 1930s and in the mid-1980s. Calvo et al claim that significant capital inflows in Latin American countries (like $850bln to Brazil stock market and $600mln in Argentine market in 1991) occur before stock market booms and cause them. The results are confirmed by the Granger causality test.
Later, in 1996 Calvo et al continued to investigate this problem by taking into consideration not only Latin America but also Asian countries in 1990-s (Indonesia, Malaysia, Philippines, Thailand; Argentina, Brazil, Chile, Colombia, Mexico). They have noticed that surge in capital flows to these countries was followed by acute rise in stock and real estate prices. In 1991 Chile and Mexico offered returns of about 100% while Argentina declared an annual dollar return of 400%.
Kaminsky and Reinhart (1999) use worldwide data4 from 1970s-1995. They have noticed that the increase in credits and capital inflows leads to overvalued currency causing the boom in economic activity followed by recession and crisis. Currency problems deepen banking crises.
Edwards (1990) describes the reasons why developed countries invest in emerging rather than in developed countries. The regression analysis (with FDI/GDP as the dependent variable) in his work reveals that countries with lower income per capita (-0,48E-6), larger internal markets, and higher domestic investment ratios (0,00017), will tend to be more attractive to FDI. Also the government variable (government consumption in GDP) is significant with negative coefficient. To check the common belief about the influence of political factors on FDI, Edwards takes the degree of political instability (Cukierman-Edwards-Tabellini index measures probability of government change in a particular year) and the degree of political polarization and violence. The political instability index has a negative sign and is significant, while the second index of political situation in a country is insignificant in the regression. Also political variables have
4Denmark, Finland, Norway, Spain and Sweden (developed); Argentina, Bolivia, Brazil, Chile, Colombia, Indonesia, Israel, Malaysia, Mexico, Peru, the Philippines, Thailand, Turkey,
8
coefficients much smaller than economic ones, so, FDI volume depends less on political factors than on economic.
FDI as a source of risk for developed countries
et al (Dooley, Landau and Garber, 2004) consider the world as an entire economic system, with the US as its center. After the begining of the Communist system collapse in 1989-1991 the new center was formed: U.S.-Europe-Japan. These countries were able to open their economies for trading and their capital markets for foreign agents. Other countries, primarily in Asia, chose the strategy of post-war Europe and Japan with undervalued exchange rate, the accumulation of reserves, control over the size of foreign intervention and promotion of the export growth to the countries of the “center”. The deficit in the U.S.(around $500bn, 4.7% of the GDP) was financed by the trade in the region and private flows from the region capital. Accumulation of foreign exchange reserves in Japan, China and Taiwan financed 42% of the $489bln US current account deficit in 2002. The inflows have raised the dollar rate and allowed the country to improve its competitive position. In countries of the trading region, the reserves accumulation was a common feature, regardless of the exchange rate type. Thus, according to this article, capital inflows should have a positive effect on the “central” countries, so, on developed ones.
Gourinchas and Rey in their theoretical work analyze the net foreign asset position of the US (Gourinchas, Rey, 2007). The US takes a long position by investing in equity or making a direct investments and lending at a short-term for developing countries. As Dooley and Landau (2004) they argue that the direction of the main capital flows is actually from developing countries to developed, and mainly to the US (that contradicts with the opinion of Ba(2014) and Calvo et al (1993,1996) cause from their point of view main inflows take the opposite direction, from developed countries to developing). The current account deficit of the US grew from 5,3% of GDP in 2004 to 6% in 2006. Gourinchas and Rey claim that since 1982 the Unites States has been the largest net capital importer. Liabilities exceed the assets and if the foreign investors do not fund the US external deficit, the US will fail in its obligations. The same will happen in case of the crisis situation in the economy of countries where the US has invested. Hence it creates a systemic risk. FDI investments (which the authors call “risky” assets) on average form a significant part of assets for advanced economies (49% for the US, 50% for Canada, 26% for the UK) and the volume of such investments for developing countries much smaller (5% for India, 9% for China, 5% for Indonesia). So, developed economies bear more risk since they make riskier investments.
9
Cœuré states that considering not the whole world but Europe, Germany has the same position as the US in the world. (Cœuré, 2013). It produces short-term bonds with low risk and invests in risky assets of other countries. Thus, like in the case with the US, FDI outflows are risky for Germany.
Herrero A.G. and Simon D.N. (2003) review the theoretical and empirical literature about the determinants of FDI and its influence from the home country prospective (a country producing FDI). They argue that there is a gap in the literature about the influence of FDI on the home country (the country which is the source of investments). So, this thesis might help to fill this gap.
The summary of the literature review above can be represented in the following table:
STUDY
SAMPLE
RESULTS
Ba (2016) Britain and 25 economic partners, the XIX century
Probability to have a crisis in two years after foreign capital inflows surge is two times higher than without capital surge Calvo et al (1993,1996) Asia and Latin America,
1980-s-1990-s.
significant capital inflows in these regions ($850bln to Brazil stock market and $600mln in Argentine market in 1991) occur before stock market booms and cause them Edwards (1990) 58 less developed
countries (Bolivia, Botswana, Brazil, Ecuador, Mexico, etc), 1980-s
Countries with lower income per capita, larger internal markets, and higher domestic
investment ratios tend to be more attractive to FDI. Gourinchas and Rey
(2007),
Dooley et al (2004)
Worldwide data (around 190 countries, 1970-2011)
Developed countries have almost half of their assets as risky FDI investments, while for developed countries its only around 5-10%. So, FDI is more significant problem for advanced economies. So, the main direction of FDI should be from developed countries to developing. Dooley et al (2004) Worldwide data Developing economies
finance current account deficit in the US trough FDI. If the crisis happen
10
in one of these countries it will have spillover on the US and may affect all interconnected
economies.
Coeuré (2013) Europe Analogously with
Gourinchas and Rey (2007
These papers analyze why FDI can lead to crises. In the thesis I would like to investigate why FDI can lead to the increase in systemic risk, which can be a step before crisis. Then FDI could be considered as an important indicator. The main questions are the existence of connection between FDI and Systemic Risk, the direction of FDI which can lead to negative consequences and countries for which it is important.
Methodology
To assess empirically the impact of FDI on Systemic risk the following model is used: 𝑆𝑅𝐼𝑆𝐾𝑖𝑡= 𝛽0+ 𝛽1𝐶𝑟𝑜𝑠𝑠_𝑏𝑜𝑟𝑑𝑒𝑟 𝑖𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑖𝑛𝑓𝑙𝑜𝑤𝑖𝑡
+ 𝛽2𝐸𝑚𝑒𝑟𝑔𝑖𝑛𝑔𝑖+𝛽3𝐶𝑟𝑜𝑠𝑠 𝑏𝑜𝑟𝑑𝑒𝑟_𝑖𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑖𝑛𝑓𝑙𝑜𝑤𝑖𝑡∗ 𝐸𝑚𝑒𝑟𝑔𝑖𝑛𝑔𝑖+ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀𝑖𝑡5
Analogously with Cross-border investments’ outflow.
The most important coefficient is 𝛽3 , the coefficient of interaction term. It is expected for 𝛽3 to be significant and positive Ba H. (2015), Calvo et al (1993,1996) found empirically positive influence of FDI on crises (the higher FDI inflow, the lower interest rates, the higher probability of credit and asset bubbles and following crises). If 𝛽3 >0, then FDI inflow (outflow) increases systemic risk for emerging (developed) countries in Europe.
As control variables macroeconomic indicators were chosen as they also can affect systemic risk and change it before crisis: inflation, exchange rate and GDP.
Systemic risk
Systemic risk is measured by laboratories of HEC Lausanne and NYU Stern Volatility Institute. The methodology is based on papers of Acharya et al (2010), Brownlees and Engle (2010) and
5First of all, the type of model suited for the given data was chosen. There are three types of models for panel data – pooled regression, regression with fixed effects or random effect. The results of tests (Breush-Pagan test, Lagrange Multiplier test and Hausman test) have revealed that the model with fixed effects is appropriate for current data.
11
Engle et all (2012). Systemic Risk is the propensity of a firm to be undercapitalized while the whole financial system is undercapitalized (in financial crisis). For a bank it means to have market value of equity below a fraction 𝜃 of assets set by a regulator. Thus, if 𝐴𝑖,𝑡 − 𝑊𝑖,𝑡 > 0 , where 𝐴𝑖,𝑡 – assets, 𝑊𝑖,𝑡 – equity.
Expected capital shortfall is defined in the following way:
𝐶𝑆𝑖,𝑡:𝑡+𝑇 = 𝐸𝑡−1(𝜃𝐴𝑖,𝑇− 𝑊𝑖,𝑇|𝐶𝑟𝑖𝑠𝑖𝑠𝑡:𝑡+𝑇),
where 𝐶𝑆𝑖,𝑡:𝑡+𝑇 is an indicator whether there was a crisis between dates t and t+T. Capital shortfall measures the amount of capital needed to be correctly capitalized after crisis (in
accordance with regulator’s requirements). It can be rewritten in terms of units from the balance sheet or estimated econometrically:
𝐶𝑆𝑖,𝑡:𝑡+𝑇=[𝜃(𝐿𝑖,𝑡− 1) − (1 − 𝜃)(1 − 𝐿𝑅𝑀𝐸𝑆𝑖,𝑡:𝑡+𝑇)]𝑊𝑖,𝑡, where 𝐿𝑖,𝑡 = 𝐴𝑖,𝑡 𝑊
𝑖,𝑡
⁄ is a financial leverage, 𝐿𝑅𝑀𝐸𝑆𝑖,𝑡:𝑡+𝑇 𝑖𝑠 long-run marginal expected shortfall of
the bank. The latter is the sensitivity of bank’s equity return to the world market changes in case of financial crisis (to 40% semiannual market decline).
𝐿𝑅𝑀𝐸𝑆𝑖,𝑡:𝑡+𝑇 = −𝐸𝑡−1[𝑅𝑖,𝑡:𝑡+𝑇|𝑅𝑀,𝑡:𝑡+𝑇 ≤ −40%], where T = 6 month, R – cumulative returns:
𝑅𝑖,𝑡:𝑡+𝑇 = 𝑒𝑥𝑝 (∑ 𝑟𝑖,𝑡+𝑗 𝑇
𝑗=1
) − 1
𝑅𝑀,𝑡:𝑡+𝑇 = 𝑒𝑥𝑝(∑𝑇𝑗=1𝑟𝑀,𝑡+𝑗) − 1, where
𝑟𝑖,𝑡 is the daily log-return of firm i and 𝑟𝑀,𝑡 the daily log − return of the market at date t. Bank daily return depends on country 𝑟𝑐,𝑡, region (Europe) 𝑟𝐸,𝑡 and world indexes 𝑟𝑊,𝑡. World market can affect particular country with time-lag of 1 day because of time-zones, that is why the model includes both 𝑟𝑊,𝑡 and 𝑟𝑊,𝑡−1.To receive 𝑟𝑖,𝑡 the laboratories userecursive multifactor model with time-varying parameters
𝑟𝑖,𝑡 = 𝛽𝑖,𝑡𝐶𝑟𝐶,𝑡+ 𝛽𝑖,𝑡𝐸𝑟𝐸,𝑡+ 𝛽𝑖,𝑡𝑊𝑟𝑊,𝑡+ 𝛽𝑖,𝑡𝐿 𝑟𝑊,𝑡−1+ 𝜀𝑖,𝑡 𝑟𝐶,𝑡 = 𝛽𝐶,𝑡𝐸 𝑟𝐸,𝑡+ 𝛽𝐶,𝑡𝑊𝑟𝑊,𝑡+ 𝛽𝐶,𝑡𝐿 𝑟𝑊,𝑡−1+ 𝜀𝐶,𝑡 𝑟𝐸,𝑡 = 𝛽𝐸,𝑡𝑊𝑟𝑊,𝑡+ 𝛽𝐸,𝑡𝐿 𝑟 𝑊,𝑡−1+ 𝜀𝐸,𝑡 𝑟𝑊,𝑡 = 𝛽𝑊,𝑡𝐿 𝑟 𝑊,𝑡−1+ 𝜀𝑊,𝑡, Where L is for Lagged world index.
12
There is a problem to measure LRMES because only three declines of 40% exist in history (1929, 2000, 2008). Brownless and Engle (2010) suggest following approach to measure LRMES in Europe:
LRMES is estimated as the expected return of the firm in case of 40% semiannual market decline. Simulation of the model:
𝐿𝑅𝑀𝐸𝑆
𝑖,𝑡:𝑡+𝑇= −
∑ 𝑅𝑖,𝑡:𝑡+𝑇 (𝑠) ∗𝐼(𝑅𝑀,𝑡:𝑡+𝑇(𝑠) ≤−40%) 𝑆 𝑠=1 ∑𝑆𝑠=1𝐼(𝑅𝑀,𝑡:𝑡+𝑇(𝑠) ≤−40%) , I(*)=1 if condition inside brackets is true and 0 otherwise. S=50000 draws. Then, Systemic risk of a bank is calculated as𝑆𝑅𝐼𝑆𝐾𝑖,𝑡:𝑡+𝑇 = max (𝐶𝑆𝑖,𝑡:𝑡+𝑇, 0),
If 𝐶𝑆𝑖,𝑡:𝑡+𝑇 < 0 then bank has more equity than it is required by regulator (as 𝐶𝑆𝑖,𝑡:𝑡+𝑇 = 𝐸𝑡−1(𝜃𝐴𝑖,𝑇− 𝑊𝑖,𝑇|𝐶𝑟𝑖𝑠𝑖𝑠𝑡:𝑡+𝑇) and 𝜃 is regulator’s requirement) and there is no risk for it. LRMES for the entire financial system is:
𝐿𝑅𝑀𝐸𝑆𝐹,𝑡:𝑡+𝑇 = −𝐸𝑡−1[𝑅𝐹,𝑡:𝑡+𝑇|𝑅𝑀,𝑡:𝑡+𝑇≤ −40%],
where 𝑅𝐹,𝑡:𝑡+𝑇 = ∑𝑁𝑗=1𝑤𝑖,𝑡𝑅𝑖,𝑡:𝑡+𝑇, 𝑤𝑖,𝑡 = 𝑊𝑖,𝑡⁄∑𝑁𝑖=1𝑊𝑖,𝑡 because cumulative return for the financial system is the value-weighted individual institutions’ returns.
So, LRMES for a country obtained by aggregation is:
𝐿𝑅𝑀𝐸𝑆𝐹,𝑡:𝑡+𝑇 = ∑ 𝑤𝑖,𝑡𝐿𝑅𝑀𝐸𝑆𝑖,𝑡:𝑡+𝑇 𝑁
𝑖=1
Systemic risk of the whole financial system (for one country) is aggregation of individuals by weight: 𝑆𝑅𝐼𝑆𝐾𝐹,𝑡:𝑡+𝑇 = ∑𝑁𝑖=1𝑤𝑖,𝑡𝑆𝑅𝐼𝑆𝐾𝑖,𝑡:𝑡+𝑇, where 𝑤𝑖,𝑡 = 𝑊𝑖,𝑡 ∑ 𝑊𝑖,𝑡 𝑁 𝑖=1 ⁄
Thus, the data from HEC Lausanne for systemic risk was gathered at the bank level for each country and then they aggregated it at the country level by equity weights.
Data
The data is aggregated at the country level. Sample consists of 28 European countries with observations for the period of 2002-2011. Europe was chosen because of data availability. The sources of data are database from Lane and Milesi-Ferretti (2007) (for FDI) and CFRM (Lausanne) for systemic risk.
13
Data for FDI is taken on an asset/liability basis: for a country assets include equity investment of resident company abroad (outflow). Liabilities are investments from abroad into the resident country, so the resident country is a recipient (inflow).
Table 1. Desciptive statistics Developed
(mean)
Transition (Mean)
mean median min max sd
SRISK 36348.2 30.2 27268.7 85.7 0 449194.1 70972.52 FDI inflows 364264.2 109640.1 300608.1 134084.5 2413.4 2653457 381303.6 FDI outflows 436277.2 51834.3 340166.5 118629.7 144 2731302 484150.9 GDP 732676.9 343074.7 635276.3 283357.6 4467.1 3752110 840625.4 Infl 2.213241 7.9 3.64315 2.440271 -4.5 44.96413 4.378523
There are 280 observations for 28 countries for 10 years. As can be seen from the table, the mean value of systemic risk is $27268,7mln. At the same time, the median is only $85,7mln which is far from the mean. It occurs because there are several economies with large amount of systemic risk (like United Kingdom) which moves the mean to higher value.
For FDI inflow and outflow the results are comparable, the means are $300bln for inflows and $340bln for outflows. However standard deviation is significantly higher for outflows.
Also, the table represents the data for two subsamples – for two types of countries. All values are higher for developed countries. It seems logical since stronger economies can operate with larger amounts of money, and their risks are higher because the size of their financial systems is bigger. It should be noticed that for developed countries FDI outflows are higher than inflows on average ($436277,2mln vs $364264,2mln), and for the emerging countries the situation is opposite ($51834,3mln vs $109640,1mln). This coincides with hypotheses that inflows should affect more developing countries, while outflows should have higher risk for developed if to compare these values with GDP size. So, for developing countries, for example, FDI_outflows/GDP is higher than FDI_inflows/GDP.
14
Pic1. FDI and systemic risk in developed countries Pic2. FDI and systemic risk in emerging (transition_ countries
At the graphs above co-movements between FDI and SRISK can be seen. Before 2007, there was an increase in both FDI inflows and outflows and after with a small lag it pushed sharply SRISK. After 2007, if to look at SRISK can be noticed that it repeats movements of FDI flows with the lag of one or two years for either developed or developing countries. This might mean that FDI can be a leading indicator of SRISK, or it is the factor determines the movement of SRISK. This fact will be checked by regression analysis.
There is a moderately strong statistically significant correlation between FDI and SRISK (0,5330*** for inflows and 0.6590*** for outflows, for the whole table see appendix). So, there is a connection between these two terms and coefficients in following regression analysis should be significant. Also it should be important to set causal relationship rather than just correlation.
15
From the given scatterplot can be concluded that there should be a significant positive link between FDI (either inflow or outflow) and Systemic Risk.
Results
This section presents the results of regression analysis. After the results of tests for model specification it was revealed that the model with fixed effects is the most appropriate for current dataset. Test for unit roots revealed a stochastic trend in systemic risk. To deal with this problem the model with first differences was chosen:
𝐹𝐷_𝑆𝑅𝐼𝑆𝐾𝑖,𝑡−(𝑡−1)= 𝛽0+ 𝛽1𝐹𝐷_𝐶𝑟𝑜𝑠𝑠_𝑏𝑜𝑟𝑑𝑒𝑟 𝑖𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑖𝑛𝑓𝑙𝑜𝑤𝑖,𝑡−(𝑡−1)
+ 𝛽2𝑇𝑟𝑎𝑛𝑠𝑖𝑡𝑖𝑜𝑛𝑖+𝛽3𝐶𝑟𝑜𝑠𝑠 𝑏𝑜𝑟𝑑𝑒𝑟_𝑖𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑖𝑛𝑓𝑙𝑜𝑤𝑖𝑡∗ 𝑇𝑟𝑎𝑛𝑠𝑖𝑡𝑖𝑜𝑛𝑖+ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 + 𝜀𝑖𝑡
Where 𝐹𝐷_𝑆𝑅𝐼𝑆𝐾𝑖,𝑡−(𝑡−1)= 𝑆𝑅𝐼𝑆𝐾𝑖,𝑡− 𝑆𝑅𝐼𝑆𝐾𝑖,𝑡−1, and analogously for FDI (cross-border investment).
Table 2: FDI flows influence on systemic risk Dependent variable: systemic risk
VARIABLES 1 2 3 4 5 6 7 8 Fd FDI inflows -0.113*** -0.152*** -0.109*** -0.159*** (-5.69) (-7.118) (-5.75) (-6.823) interact_infl 0.090* 0.189*** (1.63) (3.259) Inflation -423.848 792.5 -360.959 924.9 (-0.94) (1.009) (-0.28) (1.122) Exchange rate -16.937 149.3 -12.394 203.0 (-0.40) (0.279) (-0.28) (0.355) GDP 0.027*** 0.013*** 0.0330*** (2.853) (6.89) (3.415) Fd FDI outflows -0.070*** -0.106*** -0.725*** -0.0935*** (-3.88) (-5.416) (-4.29) (-4.620) interact_outfl 0.038 0.148 (0.43) (1.527) Observations 252 252 252 252 252 252 252 252 R-squared 0.185 0.185 0.233 0.248 0.116 0.116 0.145 0.168 Number of id 28 28 28 28 28 28 28 28
Year fixed effects No Yes No Yes No Yes No Yes
Country fixed effects No Yes No Yes No Yes No Yes
16
Notes: Fd means first differences, interact_infl - interaction term of variables. interact_outfl – interaction term of variables Transition and fdFDI_outfl.
In the table above models 1,2 and 5,6 are the simplest versions without any controls, Model 3,4 and 7,8 are models with main independent variables (first differences of FDI inflow and outflow) and controls (inflation, exchange rate and GDP). Models 1,3,5,7 are without fixed effects, models 2,4,6,8 are with fixed effects. As can be seen, the coefficients of interests in front of FDI and interaction terms are significant. The sign of FDI coefficients is opposite from the one supposed in the hypothesis. For example, according to model 2 the increase in FDI inflows leads to decrease in systemic risk (-0.152). Analogously with an FDI outflow in model 4 (-0.106). However, the coefficient of the interaction term interact_infl (0.189) has a positive sign and coincides with the suggestion that there is a difference in influence of FDI between Developed and Developing countries. Thus, for transition countries, in contrast with developed, the influence of FDI inflows on systemic risk is positive: the higher is the first difference of FDI, the higher the first difference of systemic risk. For the interaction term interact_outfl the coefficient (0.148) is significant only at 13% level, but also positive. Variable Transition is omitted from the initial model because of perfect collinearity with interaction terms.
These results come along with results of the first group of articles considered in literature review section (Ba (2015), Calvo et al (1993, 1996), Kaminsky and Reinhart (1999)). FDI can negatively affect developing countries and be the reason of increase in systemic risk followed by the crisis. FDI influence on credit causing credit expansion, also the surge in FDI can be the reason of abnormal currency appreciation. These consequences can lead to credit or assets’ bubbles and be followed by the increase in systemic risk and crisis in the end.
Concerning the second group of papers (Dooley et al (2004), Gourinchas and Rey (2007)) with the common idea about the possible a negative effect of FDI on developed countries, the results of regression analysis do not confirm this hypothesis. As coefficient in front of the first differences of FDI is negative (either for inflows or outflows) and most of the countries in the sample are developed, the conclusion is even opposite, FDI can be a stabilizing factor for these economies.
The main advantage of the new approach of investigating the link between FDI and systemic risk (rather than the link between FDI and crisis in current literature) is that it can help to prevent crisis. Systemic risk is an indicator of an economic situation in the country,
17
especially in banking sector. So it is extremely important to explore what can affect this indicator and cause the movements in it. The prevention of sharp and uncontrolled rise in systemic risk can save economy of particular country from possible crisis.
Calvo et al (1993) suggested a possibility that till some point FDI can have a positive effect on a country (raising the availability of capital, stimulating GDP growth), but after, when the surge of FDI becomes abnormal and difficult to control, there can be problems for economy followed by crisis. Thus, based on this idea the hypothesis about non-linear dependence should be checked.
Table 3: FDI squared influence on systemic risk Dependent variable: systemic risk
VARIABLES 1 2 3 4
Fd FDI inflows -0.188*** -0.153***
(-5.208) (-3.931)
Fd FDI inflows squared 1.60e-08 8.35e-09
(1.250) (0.638) Inflation 512.5 846.3 (0.636) (1.020) Exchange rate -14.34 67.61 (-0.0262) (0.119) GDP 0.0238** 0.0330*** (2.431) (3.287) Fd FDI outflows -0.124*** -0.0659 (-2.652) (-1.336)
Fd FDI outflows squared 5.94e-09 -7.24e-09
(0.418) (-0.498)
Observations 252 252 252 252
R-squared 0.191 0.213 0.117 0.161
Number of id 28 28 28 28
Year fixed effects Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes
t-statistics in parentheses *** p<0.01, ** p<0.05, * p<0.1 Notes: Fd means first differences
The table above disapproves the hypothesis about squared dependence of systemic risk on FDI. Coefficients of FDI squared are insignificant for either inflows or outflows. That can mean unambiguity of FDI influence on different type of countries, stabilizing factor for developed countries and risk factor for developing. As coefficients of squared FDI are insignificant there should not be a critical point after which the influence of FDI on particular country could change. The coefficients of not squared FDI are again significant and negative approving that
18 FDI can be a stabilizing factor for economy.
Robustness check
To check for endogeneity instrumental variables method and Durbin-Wu-Hausman test are used. From the work of Passari and Rey(2015) the trade openness can be taken as the instrumental variable for FDI flows. According to the authors, trade openness can mitigate the dependence of a market from local products for developing countries. Financial and trade liberalization, and trade openness as a consequences, lead to an increase in FDI. To be a good instrument, trade openness should not relate to the dependent variable. The dependent variable is systemic risk, the propensity to be undercapitalized while the entire financial system is undercapitalized, and trade openness cannot influence this indicator.
Trade openness can be measured as the sum of export and import divided by GDP volume. The data is taken from the UNCTAD database. As the proxy for the size of financial market total market capitalization for a country is taken from The World Bank Database. The size of financial system also cannot influence the probability to be undercapitalized.
For regression with instrumental variables the results were the same in the qualitative sense – negative sign of FDI coefficients and positive sign for coefficients of interaction terms.
Table 4. Comparison of results of 2SLS and OLS models.
2SLS OLS VARIABLES fd_infl -0.407*** -0.159*** (-8.856) (-6.823) Exch_rate -90.56 149.3 (-0.446) (0.279) Infl 90.33 792.5 (0.0771) (1.009) GDP 0.00962 0.0266*** (1.232) (2.853) interact 0.396*** 0.189*** (2.821) (3.259) Observations 203 252 Number of id 27 28 t-statistics in parentheses *** p<0.01, ** p<0.05, * p<0.1
19
To check for the endogeneity the Durbin-Wu-Hausman test was held with results below:
Coefficients (b) (B) (b-B) iv fe Difference S.E. fdFDI_infl -.431439 -0.1591899 -0.2722491 0.0501719 interact 0.4273199 0.1891455 0.2381744 0.1487087
b is consistent under H0 and Ha; obtained from instrumental variables regression B is inconsistent under Ha, efficient under Ho; obtained from OLS regression
chi2(2) =
(b-B)'[(V_b-V_B)^(-1)](b-B)
= 4.11
Prob>chi2 = 0.2501
The result of test shows that there is no endogeneity in original model and simple model without instrumental variables can be used. The difference in coefficients is not systematic and simple OLS model also gives not biased results. The results of instrumental variable approach confirm the results of simple OLS model.
Conclusion
The aim of the current thesis is to explore the influence of cross-border investments on systemic risk and to reveal the difference in that influence for developing and developed countries in Europe. This is an important question since in current literature there was no common opinion about whether this influence is positive or negative, or is it the same for developing and developed countries. Ba (2016) and Calvo et al (1993, 1996) claim that developed countries can cause crises in developing though FDI, because sharp surge in FDI lead to the decrease in the interest rate in the country recipient of investments. At the same time, Dooley et al (2004), Gourinchas and Rey (2013), Cœuré B. (2013) argue that these are developed countries who suffers from FDI, because for them FDI are too risky investments and take significant part of their assets (around 50% vs only 5% for developing countries). Also, most researches cover (Calvo et al (1993,1996), Gourinchas and Rey (2013))Asian and Latin American countries since there were lots of currency and economic crises. That was an option as the crisis was usually taken as the dependent variable. The systemic risk as the dependent variable allows taking another group of country for research and also allows avoiding time limitations (taking into
20
consideration only years with crises). There were not so many crises in history, but systemic risk exists every day on financial markets. Regression analysis revealed statistically significant differences between transition and developed countries. For transition countries increase in FDI inflows can cause increase in systemic risk. The results for FDI outflows significant only at 12,8% level. For developed countries results are opposite, so FDI can be considered as stabilizing factor. Perhaps the reason is in the differences of financial markets levels between the countries. While transition countries should have appropriate regulation to control FDI inflows and to accept them in a not damage way, for developed countries FDI can be already considered as the way of portfolio diversification for investments.
Since the systemic risk can be considered as the leading indicator for the future crises and FDI affects systemic risk, FDI also can be considered as an indicator of a current situation in a particular economy and a signal of the following crisis. Also, on the base of it the new regulation can be created to control FDI flows among the countries, as FDI inflows can heave negative impact on transition (developing) economies and simultaneously stabilizing factor for developed. For future research and to check the results of current paper another approach of measuring systemic risk and FDI can be taken. For example, for systemic risk there are following approaches (Rodriguez-Moreno and Pena, 2012): Principal Component Analysis (measuring risk as CDS spreads from the pool of CDS), LIBOR spreads, Structural model (measuring a systemic risk based on at the probability of default of a particular proportion of banks in a particular financial system), CDO indexes and tranches, Banking System Multivariate Density. The approach suggested by HEC Lausanne was chosen for current thesis because of data availability for Europe with properly explained methodology.
21
Appendix
1. Countries included in research
Developed Developing
Austria Croatia
Belgium Hungary
Cyprus Poland
Czech Republic Romania
Denmark Russian Federation
Finland Turkey France Ukraine Germany Greece Ireland Italy Luxembourg Malta The Netherlands Norway Portugal Slovakia Spain Sweden Switzerland United Kingdom 2. Cross-correlation table SRISK FDI_inflow FDI_outflo w GDP Infl SRISK 1.0000 FDI_inflow 0.5330** * 1.0000 (0.0000) FDI_outflo w 0.6590** * 0.9633*** 1.0000 (0.0000) (0.0000) GDP 0.6873** * 0.5590*** 0.6623*** 1.0000 (0.0000) (0.0000) (0.0000) Infl -0.1408** -0.11946** * -0.2258*** -0.1148* 1.000 0 (0.0185) (0.0011) (0.0001) (0.0550 )
22
References
Acharya et al (2010), Measuring Systemic Risk. Available at SSRN: http:// ssrn.com/abstract=1573171.
Alfaro L.,Charlton A. (2007). “Growth and the Qaulity of Foreign Direct Investment: Is All FDI Equal”, - HBS working paper.
http://www.hbs.edu/faculty/Publication%20Files/Growth%20and%20the%20Quality%20of%20 Foreign%20Direct%20Investment-%20Is%20all%20FDI%20Equal_8de61b6b-1bb6-491c-a750-9f32d251a4ce.pdf (accessed 08.04.2016)
Aliber R., Kindleberger C. (2005). “Manias, panics and crashes. A history of Financial Crises”. – 5th edition. Hoboken, New Jersey: Wiley & sons inc.
Ba H. (2016). “The Systemic Causes of FinancialCrises in the Long 19th Century”. Midwest Political Science Association Conference.
http://heatherba.web.unc.edu/files/2015/07/TheSystemicCausesofFinancialCrises_HBa_3_23_20 16.pdf (accessed 09.04.2016)
Borensztein e. et al (1998). “How Does Foreign Direct Investment Affect Economic Growth”, - Journal of International Economics, vol.45, pp.115-135
Brownlees, C.T., and R.F. Engle (2010), Volatility, Correlation and Tails for Systemic Risk Measurement. SSRN:http://ssrn.com/abstract=1611229 (accessed online 07.05.2016).
Calvo G et al (1993). “Capital Inflows and Real Exchange Rate Appreciation in Latin America: The Role of External Fators”, - MPRA Paper No. 7125, posted 13 February, 2008.
http://mpra.ub.uni-muenchen.de/7125/ (accessed online 06.04.2016)
Calvo G.et al (1996). “Inflows of Capital to Developing Countries in the 1990s”,- The Journal of Economic Perspectives, Vol. 10, No.2, pp.123-139
Carkovic M., Levine R. (2002). “Does Foreign Direct Investment Accelerate Economic Growth?”, - University of Minnesota, Department of Finance, Working Paper
23
Cœuré B. (2013). “Global Liquidity and International Risk-sharing in the Post-crises Environment”. – Speech at the Bank of Korea International Conference:“Assessing global liquidity in a global framework”.
https://www.ecb.europa.eu/press/key/date/2013/html/sp130603_1.en.html (accessed 16.05.2016)
Dooley, M. P., Folkerts-Landau, D., and Garber, P. (2004). “The revived bretton woods system: The effects of periphery intervention and reserve management on interest rates and exchange rates in center countries”, NBER Working Paper No. 10332.
Edwards S. (1990). “Capital Flows, ForeignDirect Investment, and Debt-Equity Swaps in Developing Countries”, - NBER Working Paper No. 3497
Engle, R.F., E. Jondeau, and M. Rockinger (2012), “Conditional Beta and Systemic Risk in Europe”. HEC Lausanne working paper.
http://www.crml.ch/fileadmin/contenu/fichiers/REEJMR_SystRisk-Euro.pdf (accessed 20.03.2016).
Evanoff D.D.,Hoelscher D.S., Kaufman G.G.(2009). “Globalization and Systemic Risk”, - World Scientific Publishing Co. Ptc. Ltd., ISBN-13 978-981-283-337-2
Gourinchas P-O., Rey H. (2013). “External Adjustments, Global Imbalances, Valuation Effects”.-NBER Working paper 19240, National Bureau of Economic Research, Inc.
Herrero A.G., Simon D.N. (2003). “Determinants and Impact of Financial Sector FDI to Emerging Economies: a Home Country’s Perspective”,- BIS CGFS working paper.
Jaye N., interview of R. Aliber (2014). “Go with the Flow”. - CFA Institute Magazine.
Kaminsky G.L., Reinhart C.M.(1999). “The Twin Crises: The Causes of Banking and Balance-Of-Payments Problems”, - The American Economic Review, Vol. 89, No.3, 473-500
Makki S.S. (2004). “Impact of Foreign Direct Investment and Trade in Economic Growth”,- Journal of Economic Literature
24
Martin P., Rey H..(2006), "Globalization and Emerging Markets: With or Without Crash?." The American Economic Review, Vol. 96, No.5 (2006): 1631-1651.
Passari E., Rey H. (2015),”Financial Flows and The International Monetary System”,- NBER working paper 21172
Pena J.I., Rodriguez-Moreno M. (2013). “Systemic Risk Measures: The Simpler The Better?”, - Journal of Banking and Finance, Vol.37, pp. 1817-1831
Portes R., Rey H. (2005). “The determinants of cross-border equity flows”,- Journal of International Economics, vol.65, pp.269-296.
Reinhart M., Rogoff K.S. (2009). “This Time Is Different”, - Princeton University Press, 496p.
Repullo R., Saurina Salas J. (2011) "The Countercyclical Capital Buffer of Basel III: A Critical Assessment", - CEPR Discussion Paper No. DP8304, pp. 1 – 32
White C., Fan M. (2006). “Risk and Foreign Direct Investment”,- the UK: Palgrave Macmillan, ISBN 1-4039-4564-0.
