“An Investigation of Financial Fragility within the Economic and Monetary Union of the European Union”

(1)

1

“An Investigation of Financial Fragility within the Economic and

Monetary Union of the European Union”

Eoin Carroll

carroll.eoin@hotmail.com

(2)

2

This paper examines if members of the Economic and Monetary Union (EMU) of the European Union (EU) suffered from increased levels of financial fragility as a direct result of EMU membership. In this paper I analyze which factors contribute to an increased level of financial fragility within a country. I examine whether a larger volume of capital inflows results in higher amounts of private sector credit, as well as the effect of that credit growth on financial fragility. To that end, I utilize impaired loans as a percentage of total gross loans as a measure of that financial fragility. I find that capital inflows encouraged private sector credit growth within the EMU. Furthermore, my results show that private sector credit growth has a significant effect on financial fragility throughout the entire sample and the magnitude of this effect is greater within the EMU.

(5)

5

1. Introduction

1.1 An Introduction to Financial Fragility

Minsky (1982, 1986), explained the concept of financial fragility as a measure of an economy’s capacity or incapacity for dealing with shocks to its conditions of financing, without a widespread breakdown in the flow of payments amongst economic agents in an economy. Similarly, further literature on the topic suggests that financial fragility refers loosely to a financial system’s vulnerability to a widespread financial crisis, caused by a relatively small shock (Lagunoff and Schreft, 2001; Allen and Gale, 2004). Financial fragility can be observed at both the microeconomic and macroeconomic level. At the microeconomic level, the term financial fragility suggests that elements on both the assets and liabilities side of an economic unit’s balance sheet are exposed to shocks in their conditions of financing. At the macroeconomic level, the term financial fragility generally refers to the tendency of financial troubles to develop into episodes of financial instability (Bezemer, 2014). This paper examines financial fragility purely at the macroeconomic level and thus for the purpose of this paper, I posit that:

Financial Fragility is the risk of a financial shock developing into a full-blown financial crisis.

1.2 The Global Financial Crisis

The Global Financial Crisis (GFC) in 2007-08 resulted in the most acute and devastating financial crisis since the Great Depression. Global financial imbalances increased significantly in the years preceding the recent GFC for three main reasons. Firstly, incentives such as low global interest rates incentivized economic units to take on excessive amounts of risk and leverage. Secondly, the replacement of the traditional banking model with the new “originate and distribute” model was perceived to make the banking system more stable due to banks’ professed ability to transfer risk to the economic units most able and willing to bear it. This belief provided a disincentive for banks to build and hold adequate capital buffers. Thirdly, the financial innovation of securitization enabled the movement of large international capital flows, and this coupled with reduced lending standards and plentiful liquidity in the global financial system, led to large supply-driven credit expansions in a number of countries (see Appendix I ). In addition, financial markets had failed to discipline both lenders and borrowers for many years, up until the triggering of the GFC in 2007 (see Appendix II for a graphical display of Government Bond Yields over the period). Global imbalances unwound suddenly as a result of the above factors. (Klaas Knot, 2014).

1.3 The Economic and Monetary Union (EMU) of the European Union (EU) and Financial Fragility

(6)

6 the EMU and this raised questions regarding its financial fragility. Opponents of the EMU formation argued that it was ‘incomplete’ due to its lack of political and budgetary union, and thus inherently fragile. Moreover, Optimal Currency Area (OCA) theory proposes that in situations in which financial markets fear the insolvency of a nation, they will sell off its sovereign bonds in large amounts, which drives up the interest rate and liquidity exits the country as a result. This creates a vicious cycle between liquidity and solvency crises, as an increase in the need for liquidity coupled with a rise in interest rates will make it more difficult for the nation to roll-over its debt. In an independent nation there exists an equilibrating mechanism, in the form of control over the currency in which its debt is issued. In that case, the sell-off of the government bonds leads to currency depreciation and the liquidity does not leave the country However, without a national central bank capable of increasing liquidity the country will inevitably default on its sovereign debt. (De Grauwe, 2014).

1.4 The Aim and added value of the Research

In this paper I analyze the economic factors that may lead to an increased level of financial fragility within an economy. Specifically, I examine whether; financial liberalization, capital flows, credit to Gross Domestic Product (GDP) growth, the lending interest rate, the level of economic development and macroeconomic stability have any impact on financial fragility. In Model 1, Ianalyze the combined effect of Capital Flows, Financial Liberalization, and the Lending Interest Rate on the level of Credit-to-GDP growth. In Model 2, I analyze the combined effect of Credit-to-GDP growth, GDP per capita, Macroeconomic Stability and Financial Liberalization on the level of financial fragility. Models 1 and 2 are analyzed for both a group of 180 nations (‘All Nations’) and for the 28 nations of the EMU (EMU Nations) for comparative purposes.

The main findings of this paper suggest that capital flows between EMU member states put upward pressure on private sector credit levels within the receiving nations. Furthermore, private sector credit growth is a robust predictor of financial fragility across the entire sample, and the magnitude of this effect is greater within EMU member states. The possible reason for this is discussed in detail later in the paper.

(7)

7

2. A Review of the Literature

2.1 The Theory of Financial Fragility

Minsky (1992) built his ‘Financial Instability Hypothesis’ upon the previous work of authors such as: Fisher (1933), Schumpeter (1934), Kindleberger (1978), and Levy (1983). Minsky’s Financial Instability Hypothesis explains how the decisions made by investors regarding how their investment financing has a significant effect on the level of financial stability and thus fragility within an economy. The decisions can be classified into three different financing profiles: Hedge, Speculative, and Ponzi. Each profile carries a different level of risk which can be viewed as a different degree of financial fragility within an economy. Hedge finance refers to the situation in which the predicted near-term income stream from the investment is adequate to cover both the interest and principal borrowed. With Speculative finance, the predicted income stream will only cover the interest due and thus the debtor will roll-over the debt regularly. Finally, with Ponzi finance, the near-term predicted income stream will not even cover the interest payments. In this case, the debtor believes that the predicted increase in the value of the asset will be sufficient in order to roll over the debt. It is evident that the greater proportion of Hedge relative to Speculative and Ponzi financing profiles in an economy, the less financial fragility that will be present in an economy and vice versa. Furthermore, as Keynes (1936) outlines, the predicted cash flow of an investor is not known with absolute certainty and thus is based upon subjective expectations, i.e. speculation. Minsky (1992) goes on to explain how such speculation amongst investors can result in an increased level of financial fragility within an economy: In an investment boom (possibly induced by a credit boom), profits are increasing along with investment, which leads investors to speculate about future profits, and the level of investment increases accordingly. As profits are increasing together with investment, this encourages investors to take on greater amounts of risk, and thus their financing profiles may move away from Hedge and more towards Speculative and Ponzi finance, and as a result produce a greater level of financial fragility within an economy. If this economy suffers a shock to its conditions of financing, such as a rise in interest rates, a reduction in the availability of credit, or a combination of the two, the investors with more Speculative and Ponzi positions will most likely be unable to repay their debts. If the proportion of Ponzi and Speculative financing profiles are greater than the amount of Hedge financing profiles in this economy, the financial shock will most likely propagate into a full blown financial crisis. Minsky believed this gradual evolution from a period of stability to instability is inherent in the modern capitalist economy of today.

(8)

8 view, alongside Minsky’s Financial Instability Hypothesis. I find that malfunctioning economic fundamentals such as the combination of excessive private credit growth and lax monetary policy encouraged risk taking by both borrowers and lenders and culminated in a speculative frenzy which rendered economies extremely vulnerable to a shock in their conditions of financing. In addition, the magnitude of this effect was greater for EMU member states, which suggests that EMU membership is associated with an increased level of financial fragility, this mechanism will be described-in depth later in the paper.

2.2 The Economic and Monetary Union of the European Union 2.21 Background of the EMU

The Bretton Woods system collapse of 1971, coupled with the diverging economic policies between the members of the European Economic Community (EEC), caused various exchange rate tensions between the EEC member states at that time. As a response, an exchange rate agreement called the ‘currency snake’ was introduced in 1972 with the aim of exchange rate stability between European countries. Seven years later in 1979, the European Monetary System (EMS) was created with the primary goal of reducing the unsettling impact of significant exchange rate devaluations and regulating changes in parities. The central fundamentals of the EMS were: the introduction of the European Currency Unit (ECU) as a basket of currencies and the Exchange Rate Mechanism (ERM), which laid out fixed currency exchange rate margins, with flexible exchange rates within these margins for member currencies. The Single European Act was adopted in 1986 and introduced the goal of the Single Market as another objective of the community. The following year proposals were made to create a full Economic and Monetary Union. This was realized in 1999 with the launch of the Euro. (Mongelli, 2008)

2.22 Formation of the EMU

At the time of writing, all 28 Member States of the European Union represent a certain stage of the EMU, depending on when they joined and their desired level of integration and participation within the union.

The first stage involved complete liberalization of capital movements between the member states commencing 1st July 1990 and included all twelve member states of the EEC. In addition, this period also saw the increased co-operation, collective use of the ECU, and improved economic convergence between member states. Finally, the “Maastricht Treaty”, signed on the 7th of February 1992 set out the economic convergence criteria relating to long-term interest rates, inflation, fiscal deficit, and public debt levels which member states must meet in order to participate in the union.

(9)

9 Italy forcing their withdrawal from the ERM. However, Italy re-joined in 1996. (Mongelli, 2008)

A number of lessons were taken from this experience. It became apparent that complete liberalization of capital controls with states maintaining individual currencies pegged to one another will lead to exchange rate tensions. A potential solution to this problem was the introduction of a single currency that would also increase the economic welfare of all member states through greater price transparency for investors and consumers, the elimination of exchange rate risk, and the reduction of transaction costs. This encouraged the progression to Stage 2 of the union beginning in 1994. This stage established the European Monetary Institute (EMI) that was responsible for: (i) reinforcing monetary policy coordination and cooperation between the Central Banks of member states and (ii) planning the creation of a European Central Bank (ECB) responsible to conduct monetary policy for the single currency.

In May 1998, it was announced that 11 members of the EMU: Belgium, Germany, Spain, France, Ireland, Italy, Luxembourg, the Netherlands, Austria, Portugal and Finland were considered to have met the criteria to adopt the Euro as their single currency the following year and the ECB announced the operational framework and strategy for the common monetary policy it would carry out as of the 1st of January 1999.

On the 1st of January 1999 the exchange rates of the 11 EU member states that joined the Euro were permanently fixed, and the Euro currency became their official legal currency with their NCBs authority transferred to the ECB. In addition, the Stability and Growth pact, and the intra-EU exchange rate mechanism (ERM II) came into being. (Mongelli, 2008)

(10)

10 Figure 1. The Three Stages of EMU

Since adoption in 1999, the Euro area and the EU have experienced an expansion. Under the Maastricht Treaty, all EU member states are required to join the Euro area once they satisfy the necessary convergence criteria. As mentioned previously, Denmark and the United Kingdom have received a special ‘opt-out’ clause which allows them to remain in the EMU framework with no obligation to join the third stage. Sweden failed to meet all four convergence criteria on three different assessment dates (1998; 2000; 2002) but is still expected to join the Euro area once it meets the criteria. Furthermore, the remaining seven countries are expected to join the Euro area as soon they meet the criteria (Lithuania is scheduled to do so in 2015). See Figure 2 below for a full outline of when each Member State joined the European Union, the Economic and Monetary Union and the Euro area respectively.

(11)

11 Figure 2. EMU Member State’s Accession Dates

Country EU Membership EMU Membership Euro Area Membership Austria 1995 1995 1999 Belgium 1957 1990 1999 Bulgaria 2007 2007 Croatia 2013 2013 Cyprus 2004 2004 2008 Czech Rep. 2004 2004 Denmark 1973 1990 Estonia 2004 2004 2011 Finland 1995 1995 1999 France 1957 1990 1999 Germany 1957 1990 1999 Greece 1981 1990 2001 Hungary 2004 2004 Ireland 1973 1990 1999 Italy 1957 1990 1999 Latvia 2004 2004 2014 Lithuania 2004 2004 2015* Luxembourg 1957 1990 1999 Malta 2004 2004 2008 Netherlands 1957 1990 1999 Poland 2004 2004 Portugal 1986 1990 1999 Romania 2007 2007 Slovakia 2004 2004 2009 Slovenia 2004 2004 2007 Spain 1986 1990 1999 Sweden 1995 1995 United Kingdom 1973 1990 *Scheduled

2.3 Measuring Financial Fragility

(12)

12 may be used as ‘indicators’ or ‘predictors’ of financial fragility (Bezemer, 2014). I will now review the literature on these potential ‘indicators’ of financial fragility.

2.31 Financial Liberalization

Financial liberalization generally refers to the elimination of financial regulations such credit and interest rate ceilings, reserve requirements, lending requirements. It is also the full or partial opening of the country’s capital account (external liberalization), and thus capital flows are expected to increase post liberalization (Weller, 2001). Financial liberalization may produce costs or benefits to an economy depending on the circumstances. Chinn (2002) finds that financial systems derive greater benefits from financial liberalization if also accompanied by a high level of legal and institutional development. At the micro level, financial liberalization opens the doors for increased competition between financial institutions. Kelly and Everett (2004) show that the Central Bank of Ireland encouraged greater competition between retail lending financial institutions during the years of financial liberalization in Ireland. The increased competition may reduce the ‘franchise value’ or ‘charter value’ of banks and may result in the adoption of riskier policies in an attempt to boost declining profits, which will lead to an increase in financial fragility as a result (Demirgüç-Kunt and Detragiache 1998; Gonzalez 2005). Lorenzoni (2008) confirm that competitive financial contracts may cause unwarranted credit booms, which could lead to excess volatility. Similarly, Weller (2001) and Tornell et al. (2004) find that financial liberalization has contributed to an increased level of financial fragility within liberalized countries.

The banking system experienced a significant transformation in recent decades as a consequence of financial liberalization. The traditional banking model of ‘originate and hold’ was replaced with an ‘originate and distribute’ model. Where banks had originally retained loans on their balance sheets, the new model allowed banks to repackage loans and pass them on to a wide range of investors, and effectively off-load the risk as a result. This financial practice of pooling various types of debt together is a financial innovation known as Securitization (Brunnermeier, 2008). OECD (2012) states that the contagion which spread throughout financial markets during the GFC was in fact a direct result of the financial institutions’ strategies of creating securitized products on a global scale. Furthermore, Gennaioli et al. (2012) posit a strong connection between financial innovation and financial fragility, particularly in the creation of new kinds of securitized products.

2.32 International Financial Flows

(13)

13 inflation’. Furthermore, Allen and Gale (2000) argue that such asset price inflations, induced by capital inflows often culminate in financial crisis.

Capital inflows can also harm an economy through their resulting effect on the real exchange rate, and thus competitiveness of a nation. In a flexible exchange rate setting, capital inflows may cause an appreciation of the currency. A real appreciation of the currency will result in a country losing competitiveness, as their exports have become relatively more expensive than their competitors’ prices abroad. In a fixed exchange rate setting such as the Euro area, capital inflows will influence the real exchange rate through its affect on the price and wage levels in an economy. An increase in either of the two will also results in a loss of competitiveness. The mechanism through which this occurs is as follows. Capital flows into a country and investment in capital occurs. As the capital per effective worker ratio increases so does their productivity. As workers are paid according to their productivity, real wages increase as a result. As the money supply increases, so does the general level of consumer prices i.e. inflation. The resulting loss of competitiveness will lower the profitability of investment in the tradable sector and thus deter the private sector from further investment. Rodrik and Subramanian (2009) confirm that if countries are ‘investment demand constrained’ rather than ‘savings constrained’, foreign capital inflows will likely worsen the investment constraint through their unfavourable effect on the real exchange rate.

Within the EMU in the years preceding the GFC, Alessandrinin et al. (2012) posit that capital inflows rather than being used for productive investments were used to fuel consumption and housing bubbles, the consequence of an excessive supply-driven credit boom. Lane (2013) outlines how the mass of capital flows within the EMU were extraordinary in comparison to historical records and that debt-type (credit) instruments dominated cross-border capital flows during this pre-crisis period. It is evident that the EMU experienced disproportionately large capital flows (in the form of debt) between member states in the decades prior to the crisis, which were reflected by large and persistent current and capital account imbalances between member states. This was presumably due to the unique characteristics of the EMU such as a reduced exchange rate risk and a fully integrated financial and goods market, along with the presumption the ECB would not allow any member state to get into financial difficultly. This presumption was proven to be false. Calderon and Kubota (2012) find that mass capital inflows are a considerable indicator of financial fragility, due to their influence on credit booms, which was exactly the case within the EMU. This consequence of capital inflows along with others will be discussed in-depth in the following section.

See Figure 3 below for a graphical representation of Credit Growth and Net Foreign Debt Flows 2003-2008.

(14)

14 Figure 3

Source: LANE (2013)

Figure 4

(15)

15

2.33 Financial System Development

The financial system is the channel through which capital flows become domestic production in a country. If this channel is underdeveloped and inefficient, capital inflows are more likely to be put to unproductive uses within an economy. A common measure for the level of financial development within an economy is the ratio of Credit-to-GDP (Levine, 1997). Prasad et al. (2007) find that financial underdevelopment causes an ineffectiveness of foreign capital in an economy. If foreign funds are channelled to non-tradable sectors such as real estate, or simply used for immediate consumption, this hampers the economy’s ability to earn the income required to repay their foreign debt. Bailliu (2000) studies the relationship between the level of financial development in a country and the effect of private capital inflows on economic growth. The results show that capital inflows may only promote economic growth in countries where the banking sector has attained a certain level of development and thus is effective in allocating capital to the most productive uses.

Beck et al. (2013) discovers a positive relationship between the size of the financial system and volatility in high-income countries, and a negative relationship in low-income countries. The analysis of Allen and Gale (2000), Schularick and Taylor (2009), Rosseau and Wachtel (2011), and Jorda et al. (2012) confirm that excesses in private credit levels are a clear predictor of future financial turmoil. Boissay et al. (2013) suggests that this occurs as the level of private credit reaches and exceeds the ‘absorptive capacity’ of an economy, and any credit over this limit will be invested unproductively. Attempting to put a figure on this threshold, Easterly et al. (2001) and Arcand et al. (2012) posit that a Credit-to-GDP level anywhere over 80-100% will have negative effects on economic growth. Büyükkarabacak and Valev (2010) take it one step further, separating household credit from corporate credit. Their findings suggest that although excesses in both household and corporate credit have the potential to cause financial turmoil, the magnitude of this effect is significantly stronger for household credit.

2.34 Interest Rates

(16)

16 adopted by the U.S. Federal Reserve and ECB, which ultimately led to the financial instability of 2007-2008 (Taylor, 2009; Carmassi et al. (2009).

2.35 Macroeconomic Stability

According to The Reut Institute (2006) the expression ‘macroeconomic stability’ is used to describe the situation in which a national economy has successfully suppressed its susceptibility to external shocks. A common measure for macroeconomic stability is the general increase in consumer prices i.e. the inflation rate. Schwartz (1995) posits that price level stability is crucial for financial stability. The author goes on to explain how bank failure rates in the 1970’s and 1980’s occurred at times of general price level instability, as when banks make loans and investments, they do this based upon predictions of future inflation. Bordo and Wheelock (1998) confirm that macroeconomic instability was certainly a contributing factor in episodes of acute financial instability throughout U.S. financial history. It is price instability (and expected future price instability) that has often caused or at least exacerbated financial instability.

(17)

17

3. Data

3.1 Data Structure and Sources

The dataset used has 8 variables: Financial Liberalization (finlib), Capital Flows (capflows), Financial Development – Credit-to-GDP (credit_gdp), Lending Interest Rate (intrate), Financial Fragility (finfrag), Gross Domestic Product (gdp_pcap), Macroeconomic Stability (macro_stab), and one indicator variable EMU Membership (emu_mem) to indicate whether a country is an EMU member state or not.

There are 180 countries in total, with data from years 1970-2013 (44 years). Thus, the dataset has 7920 observations. However, not all the variables are available for every year or every country, which leads to a large amount of missing observations in the dataset. See the table in Appendix III for a description of each variable by the measure, years and source.

3.2 Definition of Variables 3.21 Financial Liberalization

Financial liberalization generally refers the period in time in which the elimination of financial regulations such as credit and interest ceilings, reserve requirements, lending requirements, or entry restrictions, occurs in a country. In addition, financial liberalization usually implies external liberalization, which is the full or partial opening of the country’s capital account (Weller, 2001). In this paper the level of financial liberalization is measured using the Chinn-Ito index of financial openness which ranges from -2.5 which is a completely closed financial system to +2.5 which is a fully liberalized financial system.

3.22 International Capital Flows

International Capital Flows may be defined as financial flows of credit and ownership claims between countries. The most general description of a country’s balance of trade is called its current account balance. If the country has a surplus or deficit on its current account, there is an offsetting net financial flow consisting of currency, securities, or other real property ownership claims. This net financial flow is called its capital account balance. Due to the lack of data available for capital flows, the author was forced to collect data for the current account balance and switch the sign. Capital flows are measured in billions of U.S. dollars.

3.23 Financial Development - Credit-to-GDP

Credit refers to private sector loans and GDP refers to the national income of the residents of a country for one year. The ratio of Credit-to-GDP measures the stock of credit in an economy as a percentage of GDP. In the dataset this ranges from between 1% and 319% of GDP.

3.24 Lending Interest Rate

(18)

18 Interest rate targets are an important tool of monetary policy instruments and these are taken into account when considering important variables such as inflation, investment and unemployment. The lending interest rate in this dataset ranges from 1% to over 1000% of the principal borrowed when all nations are included. When focusing on EMU member states the mean falls to 8% and whilst the minimum is still 1% and the max is 29.45%.

3.25 Financial Fragility

For the purposes of this paper the author defines financial fragility as the risk of a financial shock developing into a full-blown financial crisis. Financial fragility is measured by the amount of impaired loans as a percentage of total gross loans in a financial system. The figures for financial fragility range from a minimum of less than 1% to a maximum of 157.4%.

3.26 The General Level of Development

For the purposes of this paper, I use the measure of GDP per capita as a measure for the general level of development of a country. The World Bank refers to countries with low and middle national incomes per capita as ‘Developing Countries’ whilst high national income per capita countries are referred to as ‘Developed Countries’. GDP is frequently used as an indicator for a nation’s economic health, in addition to an estimate for a country’s standard of living. GDP comprises total private consumption, government spending, investment and plus or minus net exports in a country in one year. This data is then divided by the population of a country to determine the per capita figures. In this dataset, GDP per capita ranges from $142.01 in Liberia in 1995 to $133,733.90 in Qatar in 2011.

3.27 Macroeconomic Stability

I measure macroeconomic stability as the inflation rate in this paper. I take the natural logarithm of the inflation and deflation rate after I add a constant of 1 to make sure all the figures are positive (a common procedure in the literature (Dollar and Kraay, 2002)).

3.28 EMU Membership

The EMU is the successor of the EMS and is the combination of the European Union member states into a cohesive economic system. This system is most notably represented with the adoption of the Euro as the national currency of the participating members at the third stage of the Euro area development. This has been agreed to by all initial EU member states with the exception of the UK and Denmark, who received a special ‘opt out’ of joining the third stage of EMU membership. This will be captured by a dummy variable in this paper with EMU members taking a 1 and all other nations taking a zero 0.

3.3 Data Summary Statistics

(19)

19 year 2000 to 2011. This paper is at the mercy of data availability. The main reason I chose a worldwide panel of 180 countries was to maximize coverage, however it is evident that developed countries are more reliable at reporting data. This fact causes somewhat of an issue for the analysis. Furthermore, I chose to build an annual panel dataset in order to maximize sample size and the degrees of freedom. However, a tradeoff I faced when compiling a worldwide dataset is the fact that countries with more reliable and historic data are represented more in the dataset than others who do not.

Table 1 – Summary Statistics

Variable Obs. Mean Std. Dev. Min Max

finlib 6197 -.0043408 1.530606 -1.863972 2.439009

capflows 5474 8.90e+07 3.55e+10 -4.21e+11 7.98e+11

credit_gdp 6286 .3947338 .3861787 .01 3.194609

intrate 4698 .4544144 17.8023 .01 1219.06

finfrag 1342 .0680681 .086882 0 1.574

gdp_pcap 3907 13537.83 16588.4 142.0188 133733.9

macro_stab 5884 1.343594 5.047416 .6679397 245.1103

Financial Liberalization (finlib) has 6197 observations here. The minimum value of -1.863972 suggests that no country in the dataset is fully closed whilst the maximum value of 2.439009 suggests that a number of countries in the dataset are fully open. Capital Flows (capflows) has 5747 observations. The minimum in this case represents the largest capital outflow of $421 billion, whilst the maximum represents the largest capital inflow of $798 billion. The standard deviation in this case is legitimate as there are large differences in the level of capital flows depending on the countries. Credit-to-GDP (credit_gdp) has 6286 observations. The mean is a Credit-to-GDP ratio of 39.5%, with a minimum of 1.0% in Azerbaijan in 1996 and a maximum of 319.5% in Iceland in 2006. This is not an anomaly as Iceland experienced a banking sector collapse soon after this date.

(20)

20 of .66 which represents a deflation rate of approximately 34% in Uganda in 1993 and a maximum of 245.11 which represents a hyperinflation rate of 24,511% in Zimbabwe in 2007, which was shortly before the Zimbabwean dollar was abandoned.

3.31 Correlation of Variables

credit_gdp capflows finlib intrate finfrag gdp_pcap macro_stab

credit_gdp 1.0000 capflows -0.1828 1.0000 finlib 0.2868 -0.0562 1.0000 intrate -0.7052 0.1862 -0.2530 1.0000 finfrag -0.3446 0.0741 -0.2283 0.3221 1.0000 gdp_pcap 0.7662 -0.2161 0.2780 -0.6803 -0.3929 1.0000 macro_stab -0.3867 0.1403 -0.2014 0.4842 0.0787 -0.3877 1.0000

The above table displays the correlation between all the variables included in the dataset. Most notably there is a high negative correlation of -0.7052 between credit (credit_gdp) and the lending interest rate (intrate) which would be expected due to people borrowing more as the cost of borrowing falls. In addition, there is a high positive correlation between credit (credit_gdp) and GDP (gdp_pcap) which is expected due to the finance and growth nexus, in which developed nations both require and demand greater financial deepening for growth.

3.4 Outlier Analysis

With regards to Capital Flows, there are no clear outliers. However, the analysis shows that range of capital flows of non-EMU countries is much larger than that of EMU countries. This is expected as there are many more non-EMU countries and within this group there are developed and developing countries. Iceland in 2006 is far higher than other countries in the GDP figures. Also, the analysis shows that EMU countries have higher Credit-to-GDP ratios than non-EMU countries. This is expected as the EMU group has a greater proportion of developed nations compared to the non-EMU nations group, as developed nations have greater amounts of credit available in comparison to developing nations. Nicaragua in 1988 is the highest in terms of the lending interest rate. In terms of financial fragility, Peru in 2004 is a certainly the highest percentage of impaired loans to total gross loans. Although these observations are influential and possibly outliers, I will not discard this data as it is already scarce and removing it would constitute data manipulation. Please see Appendix IV for the graphs accompanying this analysis.

3.5 Check for Multicolliniarity

(21)

(22)

22

4. Methodology

4.1 Theoretical Model

The theoretical model proposed in this paper suggests that capital inflows led to excessive levels of credit growth. As a consequence, the excessive credit levels resulted in increased levels of financial fragility. In addition, I posit that both effects were stronger in EMU member states in the run up to the GFC, and as such these nations were more vulnerable to shocks in their conditions of financing which eventually propagated into the Euro crisis. To test this, I collected data for a number of variables from an assortment of sources and constructed a panel dataset of 180 countries for the years 1970-2013. I discussed the patterns of international capital flows and the resulting effect in the build-up of the Credit-to-GDP ratios in Model 1. Following this, I analyzed the pattern of credit build-ups and the resulting effect on the level of financial fragility within countries in Model 2. The sample is comprised of a wide range of countries, both EMU and non-EMU countries in order to have a sizable control group. This allows a comparison of experiences between members of the EMU and other countries worldwide.

4.11 Model 1

Y

it

**= K + a*X**

it

+ bZ1

it

+ cZ2

it

+ e

it

Y = Credit Growth

X = International Capital Flows Z1 = Financial Liberalization Z2 = Lending Interest Rate e = Error term

In Model 1 the dependent variable Y is Credit Growth and K is the intercept term. The main explanatory variable is International Capital Flows denoted by X. The additional control variables in this case are the level of Financial Liberalization and the Lending Interest Rate, denoted by Z1 and Z2 respectively. Finally, the error term is denoted by the letter e.

4.12 Model 2

Y

it

**= K + a*X**

it

+ bZ1

it

+cZ2

it

+ cZ3

it

+ e

it Y = Financial Fragility X = Credit Growth Z1 = GDP Z2 = Macroeconomic Stability Z3 = Financial Liberalization e = Error Term

(23)

23

4.2 The Method

Firstly, I provided some summary statistics and correlations of the variables. Secondly, I checked for multicolliniarity through analyzing the pairwise correlation of the explanatory variable and control variables, and a VIF test. Third, I tested for the presence of heteroskedasticity. A collection of random variables is understood to be heteroskedastic if there are sub-populations that have different variabilities than others. I tested for this using both a graphical inspection and two formal tests. If heteroskedasticity is found to be present in a model it can nullify the results of statistical tests. If heteroskedasticity is present, the solution is to report the robust standard errors. Please see Appendix VII for the graphical inspections and formal tests for heteroskedasticity for Models 1 and 2.

Furthermore, I used the fixed effects (FE) model here as I believe individual country specific effects will render the Ordinary Least Squares (OLS) biased, as some countries will have time invariant characteristics that will result in them having lower financial fragility. For example, cultural and historical differences between Germany and Ireland which are difficult to model may result in one having a natural intolerance to inflation whilst the other having an appetite for risk, and thus fragility may be more prevalent in the latter. These unexplained country specific components are contained in the error term and thus why the FE model is favored. This will also be formally tested using the Hausman test for fixed or random effects. Pleases see Appendix VIII for the statistical output for the Hausman test.

There is reason to believe that this model may be dynamic. It is possible that financial fragility in a preceding year or in a given country, may impact another through contagion in the future. For instance, if an investor makes some poor investment decisions, he or she may gamble by taking on more leverage in the hope of recouping her or his losses, but may actually increase the number of impaired loans and thus financial fragility. As such, I believe this is good motivation to posit that this model may in fact be dynamic. In addition, as mentioned previously I also believe there may be a problem of endogeneity in this model based on simultaneity, as there may be a loop of causality between the dependent variable Financial Fragility and the independent variables Credit, Income, Macroeconomic Stability and Financial Liberalization.

(24)

24

5. Empirical Results

5.1 Results – Model 1

Table 1: Model 1- Credit-to-GDP

Dependent Variable: Credit-to-GDP

Method: Panel Least Squares Total Sample: 1970-2013

Variable All Nations EMU Member States

Capital Inflows .0206509 .0499987** (0.205) (0.043) Financial .0654034 .1830966 Liberalization (0.128) (0.100) Lending interest -.3360452*** -.4766645*** rate (0.000) (0.000) Hausman Test (0.000) (0.000) Observations 1639 363 Countries 107 28

Fixed Effects Yes Yes

Notes: Robust standard errors reported due to heteroskedasticity Numbers in brackets denote p-values

(*) (**) (***) Denotes (10) (5) (1) % level of significance.

The Hausman p-value tests the null hypothesis that the Random effects (RE) is the more efficient estimator, while under the alternative fixed effects (FE) is at least consistent and thus preferred. In this instance I reject the null hypothesis and conclude the FE model is more efficient here. As the p-values for the Hausman test are <0 this confirms the FE model is the most efficient and therefore preferred estimator to use here. A post estimation check of the distribution of the residuals was normal. (Please see Appendix X for the a graphical display of the distribution)

(25)

25 Financial liberalization for EMU member states enters with a positive coefficient, however, just short of being significant at the 10% level. This suggests that the convergence criteria which required member states to fully liberalize their financial systems upon joining the EMU may have been a determining factor in allowing the build-up of credit-to-GDP levels in certain EMU nations.

The lending interest rate is highly significant at the significance level of 0.01 in both cases. The estimated coefficient is negative, which suggests a negative relationship between Lending Interest Rate and Credit. This is expected as the cheaper it is to borrow funds the more people will be willing to take on debt. This suggests that low global interest rates and thus cheap credit can be attributed to the global credit boom, and this effect seemed to be slightly stronger in the EMU. As a robustness test I used another capital inflow measure and the results remained largely unchanged. Please see Appendix IX for the output using Other Investment (OI) inflow data.

5.2 Results - Model 2

Table 2: Model 2 - Financial Fragility

Dependent Variable: Financial Fragility

credit_gdp .9401878*** 1.898922*** (0.000) (0.000) gdp_pcap -3.710255*** -6.648683*** (0.000) (0.000) macro_stab -2.142501 -5.935293* (0.011)** (0.063) finlib -.0136428 -.8389948 (0.857) (0.120) Hausman Test (0.000) (0.000) (p-value) Observations 634 200 Countries 82 28

Note: Numbers in brackets represent p-values

(*) (**) (***) Denotes (10)(5)(1)% level of significance.

(26)

26 Table 2 shows the effect of Credit, Income, Inflation and Financial Liberalization on the level of Financial Fragility within an economy. The variable of interest Private Credit is highly significant with positive estimated coefficients in both cases. The magnitude increases in EMU countries, possibly due to the existence of a common currency and an integrated financial market which may have fostered a stronger lending relationship.

Income is also significant in predicting financial fragility. The negative estimated coefficient of -3.71 indicates a negative relationship between income and financial fragility. For a one unit increase in income, financial fragility decreases by 3.71 units. This suggests that the wealthier a nation is in terms of National Income, the more resources that nation will have at its disposal to tackle financial fragility.

The inflation variable which measures the macroeconomic stability of a nation’s economy has a negative coefficient and is significant at the 0.05 and 0.10 level of significance for all nations and EMU members respectively. This is somewhat surprising as I would expect that in more stable macroeconomic environments, financial fragility would be lower as a result. However, this could possibly be explained by the fact that in times of unstable macroeconomic environments, financial institutions may lend less due to worries regarding the strength of their balance sheet. In this instance, the level of impaired loans and thus financial fragility may decrease as credit standards tighten, with the extreme scenario of this referred to as a ‘credit crunch’. On the other hand, this negative coefficient could be construed as clear evidence of an endogeneity problem based on simultaneity. This scenario will be revisited in the following section.

The financial liberalization variable has a negative coefficient and is insignificant in both cases. Focusing on the negative sign, this is surprising as theoretically the financial liberalization increases the level of financial fragility, in particular if liberalization leads to increased competition and resulting riskier behaviour from banks attempting to regain lost profits. However, this situation may be subject to a time lag as financial liberalization is usually quite a gradual process, which takes many years to fully complete. It is possible that the increased levels of financial fragility resulting from financial liberalization will not become apparent until a number of years after a country fully liberalized its financial system, and this would explain the negative sign in this model.

In the static model a robustness test was performed using time dummy variables for the years 2000-2013 for Model 2 to discover if any common shocks exist. The results of this suggest that time dummies are the insignificant party and therefore I do not reject the null that the time dummies had an effect on financial fragility. Occasionally, financial fragility fell in 2009 which is expected as mass recapitalization of banks occurred at this point.

5.3 Dynamic Panel Data

(27)

27 form of endogeneity using instrumental variable techniques whilst still controlling for individual specific effects.

Table 3: Model 2 - Financial Fragility

Dependent Variable: Financial Fragility

Method: Dynamic panel-data estimation, tow-step system GMM Total Sample: 1970-2013

L. finfrag .9485164*** .5560822*** (0.000) (0.000) credit_gdp .9240297 1.490962** (0.192) (0.012) gdp_pcap -.7992031 -1.755152** (0.392) (0.015) macro_stab 3.472992 .0516865 (0.151) (0.988) finlib .0442225 .1173388 (0.655) (0.438) Observations 518 161 Countries 80 28 Instruments 11 11 Hansen p-value 0.020 0.032 AR(2) 0.34 0.47

Note: Numbers in brackets represent p-values

(*) (**) (***) Denotes (10)(5)(1)% level of significance.

(28)

28 The variables in this model now have a coefficient with the expected sign, although insignificantly. With regards to the diagnostics, there is a presence of first order serial correlation which is required for this estimator, and the p-value here is less than 0.1. The AR2 is required to be greater than 0.1 here, which it is. With regards to the instruments, the Hansen test is an overidentificantion test of all instruments. Using this test the author does not reject the null hypothesis that the instruments are valid at the 1% level of significance.

The coefficients for the EMU members now also carry the expected sign, and Private Credit remains positive and highly significant whilst also a similar magnitude to that of the fixed effects regression. There is once again a presence of first order serial correlation, with a p-value of less than 0.1. The AR2 is greater than .1 at 0.473. With regards to the instruments, using the Hansen test I do not reject the null hypothesis that the instruments are valid at the 1% level of significance.

(29)

29

6. Concluding Remarks

In this paper I have found evidence which suggests that capital inflows encouraged private sector credit growth within the EMU. Furthermore, my results show that private sector credit growth has a significant effect on financial fragility throughout the entire sample, and this effect is greater within the EMU. These results are consistent with the literature relating to excessive amounts of capital inflows and private sector credit growth on financial fragility through the mechanisms outlined previously.

In this paper I provide a clear hypothesis of how EMU member states came to suffer from increased levels of financial fragility, which rendered them extremely vulnerable to the initial liquidity shock of the recent crisis. The removal of capital controls, an integrated financial market, and a significant reduction in exchange rate risk led to an explosion of capital flows between EMU member states. Capital in the form of debt flowed from the Northern member states such as The Netherlands, Finland, Germany, Belgium, and Austria, to Ireland and the Southern member states of Spain, Portugal and Greece. This large amount of capital inflows to these countries, in combination with prolonged low interest rates, led to an excessive supply of cheap and easily obtainable credit. As private credit levels went beyond what could be productively invested, excessive credit fuelled consumption and speculative bubbles. This practice was unsustainable as excess credit to non-tradable sectors increased debt whilst also reduced the ability to repay. When financial markets took stock, the receiving nations experienced a shock to their conditions of financing. Interest rates rose along with lending standards and the amount of Speculative and Ponzi financing positions (and thus the level of financial fragility within countries) became apparent. The number of impaired loans to total gross loans increased significantly and banks were pushed to insolvency. Government bailouts were required in order to save failing banking systems. The enormous cost of these bailouts led to a sovereign debt crisis within a number of EMU nations, and consequently resulted in the Euro crisis. On 14th March 2014, Mr. Klaas Knot President of ‘De Nederlandsche Bank’ and member of the Governing Council of the ECB agreed with the hypothesis put forward in this paper.

Hyman Minsky’s Financial Instability Hypothesis highlights the dangers of speculative frenzies which may be caused by a range of factors, such as excessive private credit growth. The results of this paper provide further evidence in support of this hypothesis, and furthermore show that this effect is greater amongst EMU members. As a result, I suggest that policies must be put in place to monitor and interpret the level of financial development within an economy in order to prevent the economy reaching a Speculative and Ponzi frenzy, and thus reduce the level of financial fragility. Furthermore, I believe a banking union within the EMU would significantly increase financial stability as it would effectively break the dangerous link between the governments and national banking systems which became apparent during the GFC.

(30)

30 for what purpose in the economy. The same problem exists for credit, as I cannot distinguish where the credit is invested in the economy. However in the case of Spain and Ireland we know it went to the asset markets such as real estate, whilst credit in Portugal and Greece was predominantly used for immediate consumption.

(31)

31

Bibliography

Alessandrini, P., Fratlanni, M., Hallett, A. H. & Presbitero, A. F., 2012. External Imbalances and Financial Fragility in the Euro Area. Mo. Fi. R Working Paper No. 66.

Allen, F. & Gale, D., 2000. Bubbles and Crisis. The Economic Journal, No.110.460, Volume Vol.110, pp. 236-255.

Allen, F. & Gale, D., 2004. Financial Fragility, Liquidity, and Asset Prices. Journal of the European Economic Association, Volume 2.6, pp. 1015-1048.

Anon., 2006. The Reut Institute. [Online]

Available at: http://reut-institute.org/Publication.aspx?PublicationId=1299 [Accessed 19 July 2014].

Arcand, J. L., Berkes, E. & Panizza, U., 2012. Too Much Finance?. International Monetary Fund .

Arellano, M. & Bover, O., 1995. Another Look at the Instrumental-Variable Estimation of Error-Components. Journal of Econometrics, Volume 68, pp. 29-52.

Bailliu, J. N., 2000. Private Capital Flows, Financial Development, and Economic Growth in Developing Countries. Bank of Canada Working Paper 2000-15.

Berk, J. M., 2013/ 14. Monetary Policy and Financial Regulation Lecture Notes. University of Groningen.

Bezemer, D., 2014. Financial Fragility Paper (in progress).

Blundell, R. & Bond, S., 1998. Initial Conditions and Moment Restrictions in Dynamic Panel Data Models. Journal of Econometrics, Volume 87.1, pp. 115-143.

Boissay, F., Collard, F. & Smets, F., 2013. Booms and Systemic Banking Crises. European Central Bank, No.1514.

Bordo, M. D. & Wheelock, D. C., 1998. Price Stability and Financial Stability: The Historical Record. Federal Bank Reserve of St. Louis Review , Volume 80, pp. 41-62.

Brunnermeier, M. K., 2008. Deciphering the Liquidity and Credit Crunch. NBER Working Paper No. 14612.

Büyükkarabacak, B. & Valev, N. T., 2010. The Role of Household and Business Credit in Banking Crises. Journal of Banking and Finance, 34(6), pp. 1247-1256.

Calderon, C. & Kubota, M., 2012. Gross Inflows Gone Wild: Gross Capital Inflows, Credit Booms and Crises. The World Bank Policy Research Working Paper 6270.

Carmassi, J., Gros, D. & Micossi, S., 2009. The Global Financial Crisis: Causes and Cures. Journal of Common Market Studies , 47(5), pp. 977-996.

(32)

32 Crotty, J., 2009. Structural Causes of the Global Financial Crisis: A Critical Assessment of the "New Financial Architecture". Cambridge Journal of Economics, 33(4), pp. 563-580. De Grauwe, P., 2014. The Economics of Monetary Union. Tenth Edition ed. s.l.:Oxford University Press.

Demirgüç-Kunt, A. & Detragiache, E., 1998. Financial Liberalization and Financial Fragility No. 1917. The World Bank Annual Conference.

Dollar, D. & Kraay, A., 2002. Growth is Good for the Poor. Journal of Economic Growth, 7(3), pp. 195-225.

Drukker, D., 2008. Econometric Analysis of Dynamic Panel-Data Models Using Stata. Easterly, W., Islam, R. & Stiglitz, J. E., 2001. Shaken and Stirred: Explaining Growth

Volatility. Annual World Bank Conference on Development Economics, Volume 191, p. 211. Fisher, I., 1933. The Debt-Deflation Theory of Great Depressions. Econometrica: Journal of the Econometric Society, pp. 337-357.

Gennaioli, N., Schleifer, A. & Vishny, R. W., 2012. Neglected Risks, Financial Innovation and Financial Fragility. Journal of Financial Economics, 104(3), pp. 452-468.

Gonzalez, F., 2005. Bank Regulation and Risk-Taking Incentives: An International Comparison of Bank Risk. Journal of Banking and Finance, 29(5), pp. 1153-1184.

Jorda, O., Schularick, M. & Taylor, A., 2011. When Credit Bites Back: Leverage, Business Cycles and Crises. NBER No. w17621.

Kelly, J. & Everett, M., 2004. Financial Liberalisation and Economic Growth in Ireland. Financial Services Authority of Ireland, Quarterly Bulletin, pp. 91-112.

Keynes, J. M., 1936. The General Theory of Employment, Interest and Money. New York: Harcourt Brace.

Kindleberger, C., 1978. Manias, Panics and Crashes. New York: Basic Books.

Knot, K., 2014. Monetary Policy of the Eurosystem and the Financial Crisis, Guest Lecture at The University of Groningen, Netherlands on 14 March.

Lagunoff, R. & Schreft, S. L., 2001. A Model of Financial Fragility. Journal of Economic Theory, 99(1), pp. 220-264.

Lane, P. R., 2013. Capital Flows in the Euro Area. CEPR Discussion Paper No. 9493. Le Goff, M. & Singh, R., 2013. Does Trade Reduce Poverty? A View From Africa. The World Bank Poverty Reduction and Economic Management Unit Working Paper No.6327. Levine, R., 1997. Financial Development and Economic Growth: Views and Agenda. Journal of Economic Literature, 35(2), pp. 688-726.

(33)

33 Lorenzoni, G., 2008. Inefficient Credit Booms. The Review of Economic Studies, 75(3), pp. 809-833.

Minsky, H. P., 1982. Can "It" Happen Again?: Essays on Instability and Finance.. Armonk, NY: ME Sharpe.

Minsky, H. P., 1986 (cop. 2008). Stabilizing an Unstable Economy. New York: McGraw-Hill.

Minsky, H. P., 1992. The Financial Instability Hypothesis. The Jerome Levy Economics Institute Working Paper No. 74.

Mongelli, F. P., 2008. European Economic and Monetary Integration and the Optimum Currency Area Theory. No. 302 Directorate General Economic and Monetary Affairs (DG ECFIN), European Commission.

Obstfeld, M. & Rogoff, K., 2009. Global Imbalances and the Financial Crisis: Products of Common Causes. London: Centre for Economic Research.

OECD, 2012. Debt and Macroeconomic Stability. OECD Economics Department Policy Notes No. 16.

Prasad, E. S., Rajan, R. G. & Subramanian, A., 2007. Foreign Capital and Economic Growth. NBER Working Paper No. w13619.

Rousseau, P. L. & Wachtel, P., 2011. What is Happening to the Impact of Financial Deepening on Economic Growth?. Economic Inquiry, 49(1), pp. 276-288.

Schularick, M. & Taylor, A. M., 2009. Credit Booms Gone Bust: Monetary Policy, Leverage Cycles and Financial Crises 1870-2008. NBER Working Paper No. w15512.

Schumpeter, J. A., 1934. Theory of Economic Development. Cambridge, Mass: Harvard University Press.

Schwartz, A. J., 1995. Why Financial Stability Depends on Price Stability. Economic Affairs, 15(4), pp. 12-25.

Taylor, J. B., 2009. The Financial Crisis and the Policy Responses: An Empirical Analysis of what Went Wrong. NBER Working Paper No. w14631.

Tornell, A., Westermann, F. & Martinez, L., 2004. The Positive Link Between Financial Liberalization, Growth and Crises. NBER Working Paper No.w10293.

(34)

34

8. Appendices Appendix I

Average Annual Private Credit Growth %

(35)

35

Appendix II

Source: Klaas Knot (2014)

-1 0 1 2 3 4 5 6 7 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

10-year government bond spreads

Netherlands France Italy Spain

26 juli 2012

(36)

36

Appendix III – Variable Table

Variable Variable Code

Measure Years Source

1. Financial System Development credit_gdp Domestic Credit to Private Sector % GDP 1970-2013

International Monetary Fund International Financial Statistics

World Bank Databank OECD GDP Estimates 2. International

Capital Flows capflows

Capital Account Balance (Billions US $)

1980-2013

World Bank Databank 3. Financial Liberalization finlib Chinn-Ito Index of Financial Openness 1970-2011 Chinn-Ito Index

4. Interest Rate intrate Lending Interest Rate (%)

1970-2012

World Bank Databank European Central Bank International Monetary Fund 5. Financial Fragility finfrag Number of Impaired Loans as a % of Total Gross Loans 2000-1013

International Monetary Fund Global Financial Stability

Report

World Bank Databank 6. National Income gdp_pcap

Gross Domestic Product Per Capita (real PPP at constant $)

1990-2012

International Comparison Program

World Bank Databank 7. Macroeconomic Stability macro_stab Inflation Rate (Consumer prices annual %) 1970-2013

International Monetary Fund World Bank Databank International Financial Statistics 8. EMU

(37)

37

Appendix IV - Outlier Analysis 1. Financial Liberalization 2. Capital Flows -2 -1 0 1 2 3 F in a n cia l L ib e ra liza ti o n (f in lib ) 0 1

Financial Liberalization of Non-EMU vs. EMU countries

-5 .0 e + 1 1 0 5 .0 e + 1 1 1 .0 e + 1 2 C a p it a l F lo w s (c a p fl o w s) 0 1

(38)

38 3. Credit-to-GDP 4. Interest Rate Iceland, 2006 0 1 2 3 C re d it G D P (c re d it _ g d p ) 0 1

Credit GDP of Non-EMU vs. EMU countries

Nicaragua, 1988 0 5 0 0 1 ,0 0 0 1 ,5 0 0 In te re st R a te (in tra te ) 0 1

(39)

39 5. Financial Fragility Summary Peru, 2004 0 .5 1 1 .5 F in a n cia l F ra g ili ty (f in fra g ili ty) 0 1

Financial Fragility of Non-EMU vs. EMU countries

0

5

10

15

0 1

Non-EMU vs. EMU countries

(40)

40

Appendix V - Check for Multicolliniarity

Model 1

Pairwise correlation of predictor variables: | capflows finlib intrate ---+--- capflows | 1.0000

finlib | -0.0878 1.0000

intrate | 0.1515 -0.2281 1.0000

There is no sign of an overly high degree of correlation between the predictor variables.

VIF – Test:

The VIF test reported no score >4 which confirms Multicolliniarity is not present.

Model 2

Pairwise correlation of predictor variables:

| credit_gdp gdp_pcap macro lfinlib ---+--- credit_gdp | 1.0000

gdp_pcap | 0.6692 1.0000

macro_stab | -0.2548 -0.1500 1.0000

finlib | 0.2043 0.2265 -0.2311 1.0000

Again, there is no sign of an overly high degree of correlation between predictor variables. VIF – Test

(41)

41

Appendix VI – Theoretical Mechanism

(42)

42

Appendix VII – Check for Heteroskedasticity

Model 1

Firstly through graphical inspection:

Through graphical inspection, it is suspected that heteroskedasticity may be present in Model1. The author will now attempt to confirm this using the two following formal tests: 1) Breusch-Pagan / Cook-Weisberg test:

Ho: Constant variance

Variables: capflows finlib intrate chi2(3) = 208.67

Prob > chi2 = 0.0000

2) White’s test for heteroskedasticity

White's test for Ho: homoskedasticity

against Ha: unrestricted heteroskedasticity chi2(9) = 207.66

Prob > chi2 = 0.0000

(43)

43 Model 2

Firstly through graphical inspection:

Through graphical inspection it would appear that there is no presence of heteroskedasticity in Model 2. The author then attempts to confirm this using the two following formal tests: 1) Breusch-Pagan / Cook-Weisberg test:

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance

Variables: credit_gdp gdp_pcap macro_stab finlib chi2(4) = 4.39

Prob > chi2 = 0.3553 2) White’s test for heteroskedasticity

White's test for Ho: homoskedasticity against Ha: unrestricted heteroskedasticity chi2(14) = 11.94

Prob > chi2 = 0.6108

As both p-values from the two formal tests are >0.05, I can conclude that heteroskedasticity is not present in Model 2.

(44)

44

Appendix VIII - Hausman test for fixed or random effects

The Hausman Test examines whether the fixed or random effects model is the more efficient and therefore appropriate estimator to be used in this model.

Model 1

---- Coefficients ----

| (b) (B) (b-B) sqrt(diag(V_b-V_B)) | fixed random Difference S.E.

---+--- capflows | .0206509 .0210293 -.0003784 .

finlib | .0654034 .0679652 -.0025618 .0030171 intrate | -.3360452 -.3673881 .0313429 .0029104

--- b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic

chi2(3) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 123.29

Prob>chi2 = 0.0000

Model 2

| fixed random Difference S.E.

---+--- credit_gdp | .9401878 .2059875 .7342004 .0646454 gdp_pcap | -3.710255 -.836741 -2.873514 .2524875 macro_stab | -2.142501 -3.940921 1.79842 . finlib | -.0136428 -.1093797 .095737 .0473236 --- b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg

Test: Ho: difference in coefficients not systematic chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B)

= 151.77 Prob>chi2 = 0.0000

(45)

45

Appendix IX - Other Investment Inflows

For a robustness test of Model 1, I ran the regressions again using a different measure for capital inflows: OI inflows. This data is somewhat more limited as it is only available for 40 countries for the years 1980-2013.

Table 4: Model 1- Credit % GDP

Dependent Variable: Credit % GDP

EMU Members EMU Members

(Original capital (OI Flows)

Variable flow data used)

Capital Inflows .0499987** .0436626*** (0.043) (0.001) Financial .1830966* .1032043 Liberalization (0.099) (0.155) Lending interest -.4766645*** -.382205*** rate (0.000) (0.000) Observations 363 261 Countries 28 28

Notes: Robust standard errors reported due to heteroskedasticity Numbers in brackets denote p-values.

(*) (**) (***) Denotes (10) (5) (1) % level of significance.

(46)

46

Appendix X – Check the distribution of the residuals

(47)

“An Investigation of Financial Fragility within the Economic and Monetary Union of the European Union”