Do global gross asset plus global gross liability positions increase global financial stress? : a study of 17 advanced countries.

Do global gross asset plus global gross liability positions

increase global financial stress?

A study of 17 advanced countries

Lotte Hermsen

November 22, 2013

Student ID 5933579 Faculty of Economics and Business

Supervisor prof. dr. H. Jager Department of Economics

Abstract

This thesis analyzes whether gross cross-border capital positions increase global financial stress. This is done for 17 OECD countries over the time span 1980-2010. The Financial Stress Index (FSI) is included as the dependent variable. The explanatory variables of interest that are included in the model are absolute gross capital positions, expressed in millions of US dollars, and gross capital positions as a ratio of GDP. For both variables the annual growth rate, quadratic term and the respective lags are included. A set of control variables is composed on the basis of prior research outcomes. Six models are estimated to control for various econometric threats and to check for robustness. Four variables were consistently significant and were theoretically explainable: the lag of the FSI, the lagged growth rate of gross capital positions, domestic credit and the lagged growth rate of domestic credit. The positive relation between the lag of financial stress and financial stress proves that financial stress is persistent. The height of domestic credit is negatively related to the FSI, confirming that the notion that domestic credit stimulates economic growth and thereby contributes to financial stability. The significance of the lagged growth rate of gross capital positions and of domestic credit confirms the notion that high growth rates reflect the increasing accumulation of risks that inevitably result in financial stress.

Table of Content

2.2 Net and gross capital flows……… 8

2.2.1 Balance of Payment Identity………. 10

2.2.2 Income identity………... 10

2.3 Balance sheet risks……….. 11

2.4 Transmission of crises……….. 13

3 Empirical literature review………. 16

3.1 Gross capital flows: a cause of financial stress………..……….. 16

3.1.1 Volume of gross capital flows and financial stress…….………... 16

3.1.2 Gross capital flows and balance sheet risks……….……….. 19

3.1.3 Gross capital flows and the transmission financial stress………..…….. 23

3.2 Additional causes of crises……….……… 24

4 Data and methodology………..………. 32

4.1 Data description………. 32 4.1.1 Dependent variable………. 32 4.1.2 Independent variables……… 33 4.2 Model formulation……….. 38 4.3 Model tests……….. 39 5 Estimation results………. 48

5.1 Model (1): variables of interest……… 50

5.2 Model (2): base model………... 50

5.3 Model (3): reversed causality……… 52

5.4 Model (4): multicollinearity……… 52

5.5 Model (5): again multicollinearity……….. 53

5.6 Model (6): final model……… 53

6 Conclusion……….. 55

Chapter 1 Introduction

When Lehman Brothers went bankrupt in September 2008, the financial markets were severely impaired. The collapse of the US housing market triggered the financial crisis of 2008-09, that subsequently quickly spread propagated from the US to the rest of the global financial system. The pace at which the crisis spread worldwide made it unique in its sort. Ever since the existence of financial crises, academics and policy makers have been aiming to find the causes in order to restrict or even avoid future occurrences.

Current account imbalances have always been considered to be one of the main drivers of financial instability. In the years before the crisis of 2008-09 the current account surplus and deficit of, respectively, East Asia and the US rapidly diverged. This divergence implied that investments grew faster than savings in the US and vice versa in Asia. Bernanke (2005), the president of the Federal Reserve, disagreed with the conception that increasing current account deficits are the result of factors like the deterioration of domestic import-competing sectors or unfair foreign trade policies. He stated that both the extent and the pace at which the US current account deficit widened could not be explained by traditional domestic factors. He considered net US capital inflows and the reinforcing effects that result from savings exceeding investments abroad, to be a more fruitful explanation of the widening US current account deficits.

After the Asian currency crisis of 1997-98, Asian countries and especially China substantially increased their savings of which most were invested in the US: the bulk of the international capital flows now flew from Asia toward the US. The high volume of capital flows into the US suppressed US interest rates. Low interest rates consequently gave a boost to US private expenditures, including imports, and made mortgages cheap and saving unattractive, further stimulating the US investment boom. Bernanke concluded that China’s ‘excessive’ saving behaviour lied, in the first part of 2000s, at the roots of the US current account deficit. Hence, external rather than internal factors were the cause of the diverging US current account. Many researchers agreed and also found the US investment surplus and the financial crisis to be intimately linked (Obstfeld and Rogoff, 2009; Caballero and Krishnamurthy, 2009).

Although many scholars like Bernanke (2005) believe that current account imbalances of the US and Asia, and their associated net capital flows to be causally related to the US financial conditions that ultimately created the financial crisis, contradicting empirical evidence is accumulating (Borio and Disyatat, 2011). Three flaws are noteworthy. Firstly, the excess saving theory predicts that Chinese capital inflows suppress US interest rates and would influence the current account via lower US saving rates. However, the link between US current account balances and US interest rates seems to be very

weak. Secondly, the US dollar depreciated in the period before the crisis, while the excess saving theory predicts that through the increased demand for in dollar-denominated assets the currency would appreciate. Finally, the predicted relation between the pattern of global savings and the US current account deficit is very weak. Namely, when the US current account deficit began to

deteriorate in the 1990s, the world savings rate trended downward to the end of 2003. The fact that Asian excess savings bear no significant relation with either US financial conditions or US investment behaviour, makes it very unlikely that Asia’s saving behaviour was the primary cause of the financial crisis that initially burst in the US.

The limited explanatory power of current accounts and their associated net capital flows naturally triggers the search for better explanations. The following example of a German Landesbank illustrates how, until now mostly neglected, gross capital flows might have played a role in the diffusion of financial risks (Obstfeld, 2011). Suppose a Landesbank-sponsored conduit finances the purchase of US assets by issuing short-term Asset-Backed Commercial Paper (ABCP) to a US money market fund. The Landesbank-sponsored conduit is a legally separate entity, designed to be remote in case of the bank’s bankruptcy (Anderson and Gascon, 2009). This transaction leaves for the

Landesbank the net foreign capital position, foreign assets minus foreign liabilities, unaffected because foreign assets increase with the same amount as foreign liabilities. Yet, this financial transaction raises liquidity, maturity, credit and currency risks and increases the interconnectedness and opaqueness of the global financial system.

Borio and Disytat (2011) therefore propose to also examine the cross-border gross capital flows. Cross-border gross capital flows capture the international asset trade of the Landesbank. They argue that by solely observing current accounts and the net capital flows that correspond to them1, one fails to see the risks that arise from asset trade. Instead, global gross capital flows better capture the financial risks that arise from asset trade.

Preliminary research indeed shows that gross capital flows and positions substantially increase financial stress that may ultimately lead to financial crises. However, the majority of the research done so far was theoretical (Borio and Disyatat, 2011; Obstfeld, 2011). Until now, the relation has not been tested with a regression model. This thesis aims to contribute to existing knowledge by looking for empirically supporting evidence of gross capital flows and to contribute to the following research question via a regression model:

Do global gross asset plus global gross liability positions increase global financial stress?

1

The Financial Stress Index (FSI) will be used as an indicator of financial crises. The FSI is a continuous variable: if the value of the index passes a certain threshold the country is assumed to be financially distressed. That is to say, if it passes the threshold, the country experiences a financial crisis. International gross asset plus international gross liability positions are in several forms included as explanatory variables in the model. Based upon empirical and theoretical literature a number of additional variables is included as control variables. The current account and net capital positions are included as control variables to compare their explanatory power to the explanatory power of gross capital positions. Panel data of 17 OECD countries from 1980-2010 is used. The regression results show that the lagged FSI, the lagged growth rate of global gross asset plus global gross liability positions, domestic credit and the lagged growth rate of domestic credit are consistently significant. All significant variables, but domestic credit, have as expected a positive sign.

The remainder of this thesis is organized as follows. Chapter two provides a theoretical framework in order to identify and analyse the relevant theories on financial instability and stress. Chapter three introduces the relevant empirical research that has been done so far. Chapter four introduces the model that is used in the present research. Chapter five gives an overview of the regression analysis. Chapter six concludes this thesis.

Chapter 2 Theoretical Framework

The purpose of this chapter is to outline the theoretical background of this thesis. It aims to clarify how gross international asset plus gross liability positions are potentially related to financial stress. Section 2.1 provides a historical background that helps the reader to understand why researchers were recently induced to examine the relation between gross capital flows and financial stress. Section 2.1 describes the developments that took place in the financial markets in the decade before the crisis of 2008-09. Subsequently, section 2.2 describes how gross capital flows are created and how they differ from net capital flows. Furthermore, this section describes how gross and net capital flows are captured by the balance of payments identity and the income identity. Section 2.3 contains a thorough description of the liquidity, maturity, credit and currency risks that arise from asset trade. This is of importance since the materialization of these respective risks may ultimately lead to financial crises. The last section (2.4) describes how financial crises potentially transmit from one country to the next.

2.1 Historical background

This section elaborates on the developments that took place in the global financial markets in the decade before the crisis. It describes how those developments created a financial environment that became increasingly risky.

Bernanke (2005) believes Asia’s and in particular China’ savings behaviour to lie at the roots of the worsening financial conditions in the US. According to him, the sequence of events that created the Asia’s excess saving, began at the currency crisis that struck Asia in 1997. In that year Asian countries experienced a substantial devaluation of their currencies, which posed severe pressure on their balance sheets as a consequence of currency mismatching. After the crisis the exchange rates remained at a lower level, allowing those countries to pursue an export-based growth strategy. To keep the exchange rate low Asian governments strongly intervened in currency markets by accumulating large amounts of claims on advanced economies such as the US.

Investment rates grew rapidly in Asian countries and especially in China. Yet, saving rates grew at an even higher pace. Evidently, increased savings in Asian countries by far exceeded the decreased savings elsewhere in the world. This phenomenon is where Bernanke refers to when he speaks about the Global Saving Glut (GSG) or the excess saving theory. Because of China’s underdeveloped financial markets, Chinese sought for safe assets in foreign markets, mainly in the US. The augmented demand for safe US treasury bills altered financial conditions in the US: bond prices increased and bond yields decreased. Low yields made US saving unattractive and logically made investing more attractive,

further increasing the US current account deficit. House purchases, financed with mortgages are a particular type of investments that grew explosively. The pace at which investments grew turned out to be unsustainable, in particular due to the deteriorating quality of the mortgages and other loans that were issued. As a result, the first signs of the crisis were beginning to show in 2005-07 when failures began to rise.

Bernanke (2005) mentions nothing about the effect of Europe’s demand for risky US assets on financial conditions. Obstfeld and Rogoff (2009) however do acknowledge that Europe’s demand for US risky assets also affected financial conditions. Lower yields on US treasuries induced European investors, who were looking for higher yields, to seek for riskier US assets. In order to meet the demand for riskier assets the phenomenon ‘securitization’ came to live, which resulted in huge gross flows between mainly the US and Europe. These two authors state that securitization also strongly influenced US credit conditions, ultimately leading to the collapse of the US housing market. Securitization is essentially the transformation of illiquid financial assets, e.g. residential mortgages and car loans, into marketable capital market securities. Naturally, mortgages are illiquid because each mortgage has very specific, custom-made characteristics. Financial institutions created tradable assets by bundling a portfolio of housing loans together, dividing the portfolio of loans into standardized amounts. Standardized securities are thus assets, where each sort carries a on

beforehand calculated and designated ‘standard’ amount of risk and yield. This feature makes them easily tradable. A natural consequence of improved tradability was that cross-border asset trade -represented by gross international capital flows- expanded.

An example of such a security that became more tradable is a mortgage-backed security (MBS), which is backed by mortgage claims (interest paid and repayments). Securities are rated according to the likelihood that the asset holder receives a return on its invested money, the rating AAA means absolute certainty of receiving claims, whereas CCC is the rating of the riskiest securities. Rating securities is supposed to contribute to market transparency. Additional benefits of standardized securities are that they are suited for risk diversification and have very low transaction costs, making them very suitable for trading purposes.

Seemingly, risks -incorporated in securities- were transferred from the banking sector to outside investors, thereby spreading risks across the economy (Acharya et al., 2012). In reality however this was not the case: outside investors were often conduits, also called special purpose vehicles (SPVs), created by the bank and still financially linked to it. While it seemed that a bank had eliminated risks by selling the claims on mortgages in the form of securities (MBSs) to outside investors being its own SPV, the mortgages still imposed risks on the bank and therefore on the financial system, because banks were obliged to take the MBSs back on their balance sheet when

funding of special purpose vehicles dried up.

The creation of SPVs and the associated (invisible) financial linkages resulted in an increasingly opaque financial system. The balance sheet of a bank did not reveal the risks that it was in reality exposed to, making the overall financial sector very opaque. The opaque system made it very difficult if not impossible to assess the likelihood that other parties would not be able to meet their obligations (Acharya, 2012). Moreover, to fulfil the increased demand for MBSs, the issuance of mortgages shot up. Once the market for decent mortgages became saturated, banks started to issue the infamous and low-quality NINJA mortgages: ‘No Income, No Job or Assets’ households now became eligible for a loan (Brunnermeier, 2008). Obviously, this increased the riskiness of the MBSs. Additionally, trading in securities occurred on a worldwide scale and as a result the global financial markets became more interconnected in the decade before the crisis. The increasing extent to which global asset trade occurred is reflected by the steep incline of global gross capital flows. Yet, these global gross capital flows were to a large extent offsetting since the volume of gross capital inflows and the volume of gross capital outflows were roughly equal. In order to finance the purchase of in dollars denominated US securities, Europeans issued US liabilities to obtain the amount of dollars they needed. Due to the equal volumes of inflows and outflows of capital, the net effect -gross asset minus gross liability positions- was thus relatively modest.

To summarize, this section described in what ways the developments in the financial system in general, contributed to worsening financial conditions. I suggest that the increase of gross capital flows, reflect the opaqueness, riskiness and interconnectedness that ultimately led to financial stress.

2.2 Net and gross capital flows

Whereas Bernanke (2005) believes that the movements of the US current account and associated net cross-border capital flows are a proper indicator of the deteriorating financial conditions, Borio and Disyatat (2011) propose to examine gross cross-border capital flows. To give the reader a better understanding of where their arguments are based upon, this section dives further into the

differences between international net and international gross capital flows. Section 2.2.1 introduces the balance of payments identity to clarify the connection between the current account and the financial account. Section 2.2.2 elaborates on the income identity, where especially Bernanke’s GSG theory strongly draws upon.

First, it is of importance to mention that several terms are adopted to point to cross-border financial activity. McGuire and Von Peter (2009) use the term ‘foreign assets and liabilities’, Borio and Disyatat (2011) talk about ‘gross flows’ and Obstfeld (2012) about ‘gross international asset and liability positions’. Throughout the remainder of this thesis I will use the term gross capital positions

to refer to gross international asset plus gross international liability positions, unless stated differently. Similarly, the term gross capital flows is adopted to refer to gross international asset plus gross

international asset capital flows, unless stated differently.

Figure 2.2

Source: author’s version based on Obstfeld (2011)

Obstfeld (2012) explains what types of international trade in goods and services, as well as assets take place and the net and gross cross-border capital flows that follow from it. With the help of Figure 1 it becomes visible what gross and net capital flows exactly are. In his example, he assumes a

two-country world consisting of Home and Foreign. The red arrows represent the direction of trade flows of goods and services in a specific period. Logically, the volume of goods trade is constrained by the physical amount of goods available. In the figure Home exports more goods than it imports. This implies that it runs a current account surplus. Foreign imports more goods than it exports, and has to find a way to finance its current account deficit. It does so by rising debt to Home (an inflow of assets into Home). This transaction is represented by the short orange arrow directed towards Home. The sale of assets appears on Foreign’s balance sheet as a liability because Foreign is obliged to pay a rate of return on its liabilities in the hands of Home.

Asset for asset trade, depicted by the two lower orange arrows in Figure 1, takes place for hedging purposes and is called intratemporal trade; the assets that are traded are entitlements to consumption in the future. Asset for goods trade (as the two upper arrows represent) takes place for consumption smoothening purposes and is called intertemporal trade, since the traded goods or services can be consumed today, while assets are entitlements to future consumption of goods and services (Krugman and Obstfeld, 2009, p.177).

The total sum of assets flowing from Foreign into Home is larger than the sum of assets flowing from Home into Foreign. The former (latter) sum is the increase in Home’s (Foreign’s) gross international capital position. The difference between the two sums of asset flows is the increase in Home’s net international capital position. The figure clearly shows that gross capital flows are larger than the net capital flows.

2.2.1 Balance of Payments Identity

This section aims to explain how the current account and the net capital position are related through the balance of payments identity.

In Figure 1, Foreign runs a current account deficit and finances this deficit with a positive balance of the financial account. The balance of payments records a country’s international

transactions in a specific time period and contains the following elements: the current account, the financial account and the capital account (IMF, 1993). The financial account covers the cross-border flows of financial assets of an economy. These assets include monetary gold, SDRs2 and claims on non-residents. It also covers the foreign liabilities of an economy that consist of the indebtedness to non-residents. The first component of the capital account covers capital transfers such as remittances and debt forgiveness. The second component of the capital account is which covers the acquisition of disposal of non-produced, non-financial assets such as goodwill. The above discussed gross capital flows arising from asset trade are thus captured by the financial account and not like the name suggests by the capital account. The balance of payments (BoP) identity looks as follows:

(7)

Equation (7) implies that when the country runs a current account surplus it has to run a financial account deficit of similar size (assuming that the capital account typically is a relatively small element) for the balance of payments to be balanced as it is ex post by definition.

2.2.2 Income Identity

This section explains how the current account and savings and investments are related.

The trade account can be expressed as exports minus imports, but as well as a country’s savings minus investments after rewriting the income identity. The starting point here is the income identity, which indicates that a country’s domestic production is a sum of various components. The income identity looks as follows:

(1)

Where Y is domestic production, C is consumption by residents, I is investments by residents, G is the country’s government spending and X and IM denote, respectively, exports and imports. This identity can be rewritten as:

2

(2)

Where A is the label for absorption or total spending by residents (3). Equation two is an identity, which implies it presents no causal relationships. It expresses the feature that a balance of payments deficit in the form of a trade deficit is by definition equal to a same amount of spending surplus of the county’s residents. A shortcoming of equation (2) is that it does not show the complete current account of the balance of payments. This appears if net income earned abroad by production factors and the net inflow of unilateral transfers (together N) is added in equation (2). Including this term leads to the following identity:

( ) (4)

The current account (CA) is identical to national income, domestic production plus income earned abroad and the net inflow of unilateral transfers minus absorption. We may implement government income from taxes T twice, once with a minus sign and once with a plus sign, since this does not alter the equation. Replacing A by its components gives us the following equation:

(5)

As national savings are by definition equal to national income minus total spending by residents: . Plugging S into equation (4) ultimately yields:

( ) ( ) (6)

The left-hand side of equation (6) represents the current account. The first term on the right- hand side represents private saving surplus (the difference between private savings and private

investments), the second term represents the fiscal surplus or government’s savings (the difference between government tax income and government spending). Equation (6) demonstrates that the current account, the private savings surplus and the government surplus are fully linked (Jager and Jepma, 2011).

2.3 Balance sheet risks

Intratemporal asset for asset trade takes place for hedging purposes and thus should, in principle, increase financial stability. Securities were initially created to serve these purposes. It seems however,

that due to the increased demand for securities, their quality severely deteriorated –they became more risky. Additionally, the more opaque and interconnected infrastructure of the global financial system made it very difficult to observe those risks and even more worrisome, to assess who had direct exposure to those risks. I suspect that an increased volume of gross capital flows is accompanied by a strong increase in risks that ultimately may promote financial stress. The following chapter will turn to the assessment whether this suggestion is empirically plausible. But first this section clarifies what kinds of risks apply to a bank’s assets and liabilities and how they might result in financial stress. Bank shareholders face risks that arise from mismatching a bank’s balance sheet -assets and liabilities- in four different dimensions (Greenbaum and Thakor, 2007). Despite the fact that risks might force a bank into bankruptcy, banks are still willing to hold them because of their potential profitability. In this section I will briefly touch upon each dimension. The four dimensions are depicted in Table 2.3.

Table 2.3 Dimensions of balance sheet risk

Assets (loans) Liabilities (deposits)

1. High credit risk Low credit risk

2. Long maturities Short maturities

3. Less liquid Highly liquid

4. Foreign currency Domestic currency

Credit risk

Credit mismatching enables a bank to hold positions in projects that have a large probability of

defaulting that are funded by funds that have a smaller probability of defaulting. The probability that a bank cannot fulfil its obligations, and thus defaults, when a depositor wants to withdraw money is relatively small compared to the probability that the bank’s client defaults on its mortgage or loan. In normal circumstances, it is more likely that a bank’s client cannot pay off their mortgages –because of (temporary) unemployment - than a scenario where depositors massively withdraw their money. In case of materialization of credit risks it is likely the value of the asset side is more negatively affected by credit risks than the liability side of a bank’s balance sheet.

Maturity risk

Maturity mismatching enables a bank to hold positions in long-term projects that are funded by short-term funds. Maturity risk derives from the fluctuation of market prices. Assets and liabilities that are traded are subject to revaluations. Either the level or the structure of the interest rates unequally affects the value of assets and liabilities because of their different maturities.

Generally, assets unlike liabilities (depositors can withdraw their money immediately) have long maturities (e.g. 30-year mortgages). Prices and yields are negatively related: if the first goes up the latter goes down. The long maturity of an asset implies more uncertainty, in case the interest rate increases this reflects in a strong negative effect on its value. Shorter-termed imply less uncertainty and therefore the effect on the liabilities value of the same increase of in the interest rate is less negative. Materialization of maturity risk implies that the value of the assets is more negatively affected than the liability side of the bank’s balance sheet.

Liquidity risk

Liquidity mismatching enables a bank to hold less liquid assets that are funded by more liquid liabilities. Depositors can easily withdraw their money from their bank account. It t is harder for a bank to get rid of mortgages. The unique character of a mortgage complicates the determination of its value and its tradability. Liquidity risk can be defined as the possibility that over a specific horizon the bank will become unable to settle obligations with immediacy (Drehmann and Nikolaou, 2010). An extreme form of liquidity risk is that the seller is unable to sell the asset at any price (Greenbaum and Thakor, 2007).

Liquidity risk can be quantified by observing how much banks are willing to pay to obtain secure liquidity from the central bank to prevent itself from becoming illiquid. While bidding for funds at the central bank (CB) auction, the bank is constantly facing the trade-off to obtain either funds at the CB or in the market (Drehmann and Nikolaou, 2010). Materialization of liquidity risk implies that the value of assets is more negatively affected than the value of the liability side of the bank’s balance sheet.

Exchange rate risk

Currency mismatching enables a bank to hold in foreign currency denominated assets that are funded by in domestic currency denominated liabilities. Currency risk arises when it is impossible or when the bank is unwilling to hedge fully against exchange rate risk of currency mismatching. In case of a large appreciation of the domestic currency the value of the liabilities increases relative to the value of the assets, negatively affecting the bank’s balance sheet (Ranciere et al., 2010).

2.4 Transmission of Crises

This section describes how financial crises transit from one country to the other. Securities are increasingly traded worldwide, making the global financial system increasingly interconnected (Borio, 2007). I suggest that the increased interconnectedness of the financial system –reflected by increased international gross capital flows- facilitates the transmission of a national crisis to rest of the world. Broadly speaking two channels of transmission exist: a trade and a financial channel. Because

of the high correlation however, it is hard to distinguish between both channels (DeGregorio and Valdes, 2001).

Trade Channel

Two important channels of transmission that can be classified as the trade channel are direct trade links and trade competition in third markets (De Gregorio and Valdes, 2001). Rose and Spiegel (2009) state that two countries strongly linked through international trade might experience contagion if one of them suffers from a crisis marked by a depreciating currency. The cheaper currency improves price competitiveness of the import-competing sector of the crisis country, negatively affecting the exports of the trade partner. This effect is known as the substitution effect. Additionally, downturn of

economic activity that is typically a feature of a crisis inhibits imports of the crisis country. This effect is known as the income effect (Caramazza et al., 2004).

Financial Channel

The higher the degree of financial integration, the more extensive could the international contagious effects be. The most direct source of transmission of crisis via financial linkages occurs through a deterioration of a country’s balance sheet because of capital losses on assets with international exposure. A crisis in one country may induce investors to rebalance their portfolios and to reduce their exposure to assets that are highly correlative with the performance of that country (Caramazza et al., 2004).

Furthermore, investors’ behaviour allows shocks to spill over from one country to the other. Often, their behaviour is individually rational but lead to excessive co-movements. This type of investment behaviour can be divided into five problems: liquidity, incentive, asymmetrical problems, multiple equilibriums and changes in the international financial system (Dornbusch et al., 2000). First, liquidity problems may occur when banks have their lending concentrated in one region. When suffering losses, banks might reduce overall lending in their portfolios by reducing exposure in other high-risk areas, which reduces liquidity in those areas.

Second, banks also face an incentive to sell off all high-risk assets when suffering losses from one specific high-risk asset. Namely, as a result to losses from one specific asset, overall risk appetite might decrease.

Third, another cause of contagion is related to imperfect information and differences in investor expectations. Investors may believe that a financial crisis in one country could lead to similar crises in other countries. Obviously, this channel presumes that investors are imperfectly informed about a country’s true characteristics and thus decide on the basis of some known indicators, which may not reflect the true state of a county’s vulnerabilities. One example is herding behaviour; if the costs of gathering and processing of country specific information are high, small investors might

decide to follow the patterns of better informed investors. When the perfectly informed investors decide to leave the markets, the smaller uninformed investors will follow irrespective of their own information. This results in excessively volatile financial markets (Haile and Pozo, 2008).

Fourth, a more general explanation of contagion involves changes in expectations that are self-fulfilling in financial markets subject to multiple equilibriums. Pessimistic expectations may result in a new bad equilibrium outcome.

Fifth, contagion may result if investors change their assessment of the rules under which international financial transactions occur. New financial architecture may for example lead to concerns about the supply of funds from the central bank. Investors may expect that in the new financial

architecture, central banks are more hesitant to perform their function as a lender of last resort. These concerns could make investors reluctant to investor and fear may even paralyze the financial markets and thereby trigger a financial crisis.

Chapter 3 Empirical Literature Review

The previous chapter contains the theoretical framework: it outlined the historical background of the global financial crisis of 2008-09, what gross capital flows are and how they differ from net capital flows, what the associated risks are and how crises transmit from one country to the other. This chapter has been divided in two sections. The first one aims to provide an overview of the empirical findings that support the notion introduced in Chapter 2 that gross international capital flows are important drivers to financial stress. Obviously, gross capital flows are not the only potential cause of the financial stress. Therefore, the second section of this chapter summarizes the relevant articles that empirically lay emphasis on other causes of financial stress.

3.1 Gross capital flows: a cause of financial stress

This section is roughly structured like the previous chapter. Subsection 3.1.1 explores whether the current account and the associated net international capital flows, like Bernanke (2005) suggested, are an important indicator of worsening financial conditions. Or are gross international capital flows a better indicator? Subsequently, subsection 3.1.2 describes how, along with the growth of gross international capital flows, the four dimensions of balance sheet risks expanded. Ultimately, subsection 3.1.3 discloses that crises have mostly propagated through the financial channel.

3.1.1 Volume of gross international capital flows and financial stress

Seven observations below show that besides the current account, gross international capital flows should be taken into account to explain the incidence and the scope of the 2008-09 financial crisis (Borio and Disyatat, 2011). Even though the seven observations below focus in particular on the 2008-09 crisis, they provide sufficient reason to further examine the relevance of gross capital

positions to financial stress for a larger time span. In fact each of the seven observations show that the excess saving view, and thus the current account, has less explanatory power than gross international capital flows.

Firstly, total gross cross-border capital flows as a fraction of world GDP rose from 5% in 1998 to over 20% in 2007 (Figure 3.1.1). This expansion can mostly be accounted to flows between advanced economies, despite a decline in their share in world goods trade. The gross capital flows among emerging market countries were more modest. Yet, the excess saving theory sees emerging economies and especially China as the driver of financial crisis. Lane and Milesi-Ferretti (2008) support this observation. They impute the lesser degree of cross-border integration of the financial sectors in emerging market countries to the fact that they have not experienced the same degree of financial

innovation support this observation.

Figure 3.1.1 1995-2010. Gross capital flows as a percentage of world GDP

1 Gross flows equals sum of inflows and outflows of direct, portfolio and other investments. 2 Australia, Canada, Denmark, the euro area, Japan, New Zealand, Sweden, the United Kingdom and the United States. 3 Algeria, Angola, Azerbaijan, Bahrain,

Democratic Republic of Congo, Ecuador, Equatorial Guinea, Gabon, Iran, Kazakhstan, Kuwait, Libya, Nigeria, Norway, Oman, Qatar, Russia, Saudi Arabia, Sudan, Syrian Arabic Republic, Trinidad and Tobago, the United Arab Emirates, Venezuela and

Yemen. 4 China, Chinese Taipei, India, Indonesia, Korea, Malaysia, the Philippines, Singapore, Thailand and the 20 smaller Asian countries. 5 Bulgaria, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia. Source: Borio and Disyatat, 2011, p. 14

Second, gross flows into and out of the US expanded more rapidly than its current account deficit (Figure 3.1.2, left). The increase of gross asset claims on the US was namely three times as large as the increase of net asset claims on the US. Bear in mind that according to the balance of payments identity this increase in net claims on the US is mirrored by the current account deficit. Even if the US had not run trade deficits, there still would have been large capital flows into the US.

Figure 3.1.2 1996-2010. US balance of payments as a percentage of US GDP.

2 Sum of US Treasury securities, foreign assets in US dollar and US liabilities to unaffiliated foreigners. Source: Borio and Disyatat, 2011, p. 14

Third, the excess savings view mostly focused on the role of the Asian demand for safe Treasuries issued by the official sector. However, a large part of the capital inflows into the US originates from

the private sector: cross-border banking activity grew substantially in the years before the crisis. Overall, the foreign expansion of gross purchases of private US securities and US bank liabilities is striking and indicates an unsustainable boom in global banking (Figure 3.1.2, right). This boom would have been overlooked if one had only looked at the path of the current account balance.

Fourth, the most important source of capital inflows into the US was Europe, not the emerging countries. Half of the inflows originating from Europe was accountable to the United Kingdom, a country running a current account deficit. Hence, this is inconsistent with the excess saving view that stated that surplus countries invested their excess savings in the US asset market.

Fifth, by looking at net capital flows one does not capture the disruption that took place in the cross-border interbank lending. For example, in 2008 net capital flows decreased with only $20 million, whereas gross inflows decreased substantially with $1.7 trillion, a 75% decline compared to 2007. Sixth, in the evening before the crisis total holdings of US debt securities like Treasury bonds were particularly high in China and Japan. Holdings of privately issued mortgage-backed securities were concentrated advanced economies like the US and Europe. Borio and Disyatat (2011) conclude that Asia was only to a limited extent accountable for depressing US interest rates. It follows that Europe’s share, captured by the volume of international gross capital flows, in financing the US housing boom was more substantial in relative terms than initially was perceived.

Finally, consolidated balance sheets of banking systems show that foreign claims account for a substantial share of total assets, which highlights the boom in global banking, especially in Europe. In Figure 3.1.3 one can see that in the upper right box that at end-2007 about 90% of Swiss assets are foreign claims, implying large global gross capital positions.

Figure 3.1.3 Positions at end-2007. Size and structure of bank’s foreign operations: nationality basis.

1. in trillions of US dollars. 2 Foreign plus foreign currency claims vis-à-vis residents of the home country booked by home offices excludes inter-office claims; in percent. 3 Share of total assets booked by offices outside the home country, in percent. 4 Total claims (cross-border claims plus claims on residents in host country) booked by offices in each location over total worldwide consolidated foreign claims. Excludes banks’ “strictly domestic” claims, or their claims on residents of the home country in the domestic currency; in per cent. Source: Borio and Disyatat, 2011, p. 17

3.1.2 Gross capital flows and balance sheets risks

The prior section (3.1) mostly focused on the expanding volume of gross capital flows in the decade before the crisis of 2008-09. It also compared the empirical relationship of excess savings and gross capital flows to financial stress. Even though the seven observations may induce the reader to suspect that there is, in particular, an empirical relationship between the manifestation of financial stress and the volume of gross capital flows, it does not show causality. This section shows how each dimension of balance sheet risk was affected by gross capital flows, to support the view that gross capital flows are related to the risks that cause financial stress. Since a bank’s balance sheet is affected through four dimensions of risks, each dimension is examined separately.

Credit Risk

Credit risk materializes when a bank’s balance sheet is negatively affected due to credit risk

mismatching. Credit risk mismatching implies that credit risk is larger at the asset side of the balance sheet than on the liabilities side. Figure 3.2.2 below shows how the TED spread, the difference between the interbank Libor rate and the T-bill rate, rose after the occurrence of several events that create high uncertainty in the global markets.

losses in the subprime market. After several meetings of the Federal Reserve and potential buyers, it was decided to let Lehman go bankrupt. The bankruptcy of this internationally operating bank proved that credit risks were present and substantial. The sensitivity of the credit spreads is the result of the fact that it was well known that the global markets were full of low quality assets that could easily result in another default (Melvin and Taylor, 2009).

Figure 3.2.2 2007-09. TED spread

Source: Melvin and Taylor, 2009, p. 1322

Interest/Maturity Risk

Maturity risk materializes when a bank’s balance sheet is negatively affected due to maturity mismatching. Maturity mismatching implies that a bank’s assets have a longer maturity than its liabilities. An increase in interest rates therefore affects the bank’s assets more negatively than it affects its liabilities.

Figure 3.2.3 shows the compilation of European bank’s gross US dollar positions. European banks hold on aggregate a positive net dollar position in non-bank claims, which includes retail and corporate lending, lending to hedge funds etc. This implies the asset side of a bank’s balance sheet predominantly consists of non-bank claims. European banks hold a negative net dollar position in interbank lending. This implies that the liabilities side of a bank’s balance sheet predominantly consists of interbank lending. Because interbank loans tend to be shorter termed than non-bank claims this exposes them to cross-border maturity risk. While in 1999 the difference between interbank and non-bank net positions and therefore also the exposure to maturity risks were relatively modest, they exploded in the subsequent ten years.

Figure 3.2.3 1999-2008. European banks’ balance sheet positions, net positions in trillions of USD dollars by counterparty sector

4 Cross-border positions in all currencies and local positions in foreign currencies vis-à-vis official monetary authorities. 5 The blue line tracks estimated net interbank lending to other (unaffiliated) banks. 6 The net position vis-à-vis non-banks is estimated as the sum of net

international positions vis-à-vis non- banks and net local US positions (vis-à-vis all sectors). 7 Implies cross-currency funding, which equates gross US dollars and liabilities. Overall, European banks have a negative net position in USD. Source: McGuire and Von Peter, 2009, p. 56

Liquidity Risk

Liquidity risk materializes when a bank’s balance sheet is negatively affected due to liquidity

mismatching. In case of liquidity risk materialization a bank is unable to meet its obligations in a short period of time due to lack of liquid assets or access to liquid means elsewhere.

After the first credit risks had liquidity risks increased. In July 2007, commercial paper programmes began to face roll-over difficulties as investors nervously began to pull back after credit risk surfaced. Liquidity risks materialized when on July 30, 2007, IKB, a German bank, was unable to take over the obligations from its SPV. IKB had to be supported with cash injections from its main shareholder bank (KfW). A month later Paribas, a French bank, announced ‘the complete evaporation of market liquidity’ interbank lending markets: banks had been hoarding liquidity and had become very reluctant to lend to other banks reflecting the fear that counterparties would default in the near future (Brunnermeier, 2008). The reduced supply of liquidity increased its price, not only in the US but also in the Euro area and in the UK interbank lending rates rose sharply mid-2007 (Figure 3.2.4). As a response to the increased funding and roll-over costs of assets, banks tried to reduce their holdings of such assets. It turned out that US treasuries were relatively easy to sell, but that other non-bank asset were very illiquid and caused banks to suffer tremendous losses, reinforcing the existing credit risks (McGuire and Von Peter, 2009). Both liquidity risk and gross international capital positions reached unprecedented level right before the outbreak of the crisis.

Figure 3.2.4 2007-08 interbank markets. Left: three-month Libor-OIS spread, right: money market rates

1 Libor rate minus Overnight Index Swap (OIS) rates in basis points. 2 For the United States, effective federal funds rate; for the euro area, EONIA; for the United Kingdom, overnight Libor. 3 For the United States, federal funds target rate; for the euro area, minimum bid rate in the main refinancing operation; for the United Kingdom, official Bank rate.Source: Borio, 2008, p. 7

Currency Risk

Currency risk materializes when the value of a bank’s balance sheet is negatively affected due to currency mismatching. Currency mismatching implies that assets and liabilities are on net denominated in different currencies. In case of currency risk materialization, exchange rate depreciation affects the banks’ assets more negatively than it affects the assets.

Figure 3.2.5 below shows that UK banks for example, hold negative net positions in their domestic currencies, pounds sterling in their case, in order to finance their long positions in foreign currencies. This exposes them to currency risk since pound depreciation increases the value of their liabilities relative to their assets, negatively affecting the net value of their balance sheet. This figure also displays that net currency positions during the 1990s were relatively modest, but widely diverged in the subsequent years for most European countries. Logically, this also built up credit risk.

Figure 3.2.5 2000-09. Net foreign positions, by currency

1 Cross-border positions in all currencies and local positions in foreign currencies vis-à-vis official monetary authorities. Excluding liabilities to Japanese monetary authorities placed in banks located in Japan. 2 The solid blue line tracks net interbank lending to other (unaffiliated) banks. The dashed blue line is an alternative measure of interbank positions which makes use of the available information on inter-office positions (see box). 3 The estimated net position vis-à-vis non-banks is the sum of net international claims on non-banks and net local claims on US residents (vis-à-vis all sectors) booked by the US offices of the reporting bank. See footnote 9 in main text. 4 Implied cross-currency funding (ie FX swaps) which equates gross US dollar assets and liabilities. 5 Prior to Q4 2005, local liabilities in local currency vis-à-vis some large European countries are estimated. 6 Local positions vis-à-vis advanced economies are available from Q4 2002. The contraction in positions in Q3 2008 in part reflects the sale of some business units of ABN AMRO.Source: McGuire and Von Peter, 2009, p.52 3.1.3 Gross capital flows and the transmission of the crisis

Cross-border financial linkages have greatly expanded via the enlargement of international activities of financial institutions, as well as via the provision of services abroad as with offices abroad. The most

prominent factor that increased financial interconnectedness was the growth of cross-border portfolio investments. Figure 3.2.1 shows how the stock of international assets and liabilities as a fraction of GDP grew sharply (Borio, 2007). Spiegel (2009) recognizes as well the substantial increase of

international cross-border holding of assets. A natural consequence of this international expansion is the increasing interconnectedness of bank’s balance sheets.

Claessens and Van Horen (2011) find indeed that the financial crisis mostly propagated through cross-border connections. The creation of sophisticated financial intermediaries and instruments deeply integrated markets and let financial stress quickly spread across the global financial system through large cross-border positions.

Figure 3.1.2 1970-2005. Cross border stocks and transactionsas a percentage of GDP

2 Sum of external assets and liabilities of 22 industrial countries 3 Gross purchases and sales of bonds and equities between residents and non- residents; G7 countries excluding the UK. Source: Borio, 2007, p.5

3.2 Additional causes of crises

The second chapter and section 3.1 above explored whether gross international capital flows should be included for the explanation of the cross-country incidence and severity of the recent financial crisis. There appears to be a lot of evidence. But obviously, gross capital flows are not the only relevant factor. The aim of this section is to grasp what other factors are relevant for the explanation of the financial crises.

Even though crisis research has already been in the centre of interest for decades, this section only contains the articles that empirically researched the crisis of 2008-09, in order to keep the framework sharply defined and to include only the most up-to-date knowledge. Presumably, the decision of the authors’ to include or exclude explanatory variables is based upon prior empirical research of financial crises, and that consequently no crucial information is lacking. Across the models the majority of the variables is included in more than once, this proves that there exists consensus among researchers about what the potential causes of financial crises are. Research concerning crises

differs in its focus. In the research surveyed here, periods, dependent and independent variables and sample countries vary.

Lane and Milesi-Ferretti (2010)

The authors focus in their empirical research on the cumulative shift in output, consumption and total domestic demand over 2008-09 and how international financial linkages have influenced the state of the macro economy. The sample they use consists of about 175, developed and developing, countries. Their definition of crisis is either a negative growth of output, consumption or domestic demand during the period 2008-09 or a decline in growth rate of under a certain threshold in the period 2008-09 relative to the 2005-07-growth rate. Noteworthy, they find almost similar results if they use the 1990-2007 growth rate as a threshold.

First, the authors investigate output slowdown by estimating the average GDP growth in 2008- 09 using average GDP growth in 2005-07, the average growth rate over 1990-2007 and vectors of real-side variables and financial-side variables as explanatory variables. The real-side variable vector includes trade openness, which potentially increases the exposure to external demand shocks. Also manufacturing share in GDP in 2007 is included because manufactured goods tend to be more cyclical, letting their producers severely suffer during crises. Manufacturing share thus serves as an indicator of financial vulnerability. Interestingly, they also add a dummy variable for oil-producing countries to take into account the impact of the shift in oil prices on economic activity. Due to a collapse in demand for oil, prices may plummet and hit economic activity in oil-exporting countries more than other countries.

The financial variables vector consists of the ratio of private credit to GDP over the period 2004-07. This is included because of the potential structural vulnerabilities generated by credit growth much above the GDP growth rate. The 2007 current account value is also included as a financial variable, since the surge in risk aversion may have affected surplus and deficit countries differently. Namely, output in deficit countries may have been affected by a reversal in capital inflows. A dummy variable for exchange rate regimes is included to investigate the idea that countries with pegged exchange rates were more vulnerable to stops in capital flows and trade disruptions. The sum of foreign assets and liabilities as a ratio of GDP is used to measure financial openness. Financial openness can have either a positive or a negative effect on the balance sheet’s strength. A country’s exposure to a decline in foreign asset value affects its balance sheet proportionally negatively. However, a higher degree of financial openness may also serve as a way of risk diversification in case of domestic financial instability.

imbalances on the one hand and a decline in output and especially domestic demand on the other hand exists. Trade openness and the manufacturing share are also positively correlated with output and demand declines, although they are not always significant. The negative, significant relation between GDP per capita and output growth emphasizes the ‘advanced economy’ nature of the crisis. Countries with pegged exchange rate regimes experienced weaker output growth. The variable financial openness is only positively significant when low-income countries are excluded from the sample.

Second, the authors investigate demand growth and consumption. While the decline in output growth is a key indicator, financial integration theory implies that domestic consumption will be affected by international wealth shocks even if domestic GDP growth is unaffected. The same set of control variables is used. Overall, they find a strong correlation between consumption and GDP growth. For domestic demand growth they also find a very strong correlation with GDP growth.

Rose and Spiegel (2012)

The authors try to gain a deeper understanding of the international scope of the crisis and are especially interested in modelling the causes of the crisis and why if severity differs across countries. The adopted Multiple-Indicator Multiple-Cause Model (MIMIC) methodology simultaneously links indicators of a financial crisis with potential causes. One of the attractive features of the MMIC model is that it treats the severity of a crisis as a continuous rather than a discrete phenomenon. Another positive feature of the model is that it recognizes and tries to deal with the possibility that the severity of crises can be measured with an error.

They observe four indicators of the severity of the crisis: three financial indicators they include are the percentage change in the stock market index in 2008, the percentage change in the exchange rate, measured as the domestic currency price of one Special Drawing Right (SDR), and the change in the country credit rating. The fourth indicator is macroeconomic of nature: real GDP growth in 2008. Initially the MMIC model consists of two sets of equations. The first equation measures the

unobservable severity of the crisis by using the four crisis indicators and, thus, is a definition equation. The second equation models the behaviour of the crisis as a function of the severity of the crisis calculated in the first equation.

The sample includes 107 countries, aiming to include all the countries that have dramatically been affected, but as well countries that have not been affected as badly. Since high-income countries were most affected they are all included, plus a number of developing countries. The authors examine countries with an annual real GDP per capita of at least $10.000 in 2003, plus those countries with an annual real GDP per capita of at least $4000 and a population of one million or more. Rose and Spiegel try to link the crisis incidence and its causes by using a list of possible determinants of the crisis, by

using data from 2006 and earlier.

The included explanatory variables can be divided into roughly eight categories: the country’s size and income, financial policies, financial conditions, asset price appreciation, the current account, macroeconomic policies, institutions and geography. Rose and Spiegel (2012) illustrate the effects of the explanatory variables as follows: relatively small countries tend to depend more on trade, increasing their vulnerability. Income can have an ambiguous effect since on the one hand it seems that richer countries might have better means to respond to crisis, but on the other hand the

availability of these means could have encouraged agents to take risky assets on their balance sheets. Low capital requirements allowing banks to take on excessive leverage may increase the probability of financial instability. Deteriorating lending standard increases financial fragility. The housing market boom seems to be a typical example of an unsustainable asset price appreciation, increasing risk of a bursting bubble and thus a crisis. A country’s current account surplus can function as a war chest. The interest rate target could also have fuelled the credit boom. The existence of regulating bodies could help reducing macroeconomic volatility. Ultimately, a country located closely to a financial centre makes it an attractive country to invest in.

The regression results are in general very weak and disappointing. From the wide range of over sixty variables almost none affects significantly the severity of the crisis. There are however a few explanatory variables that were found to be strongly significant related to incidence of the crisis: countries that experienced a large equity appreciation before the crisis were more likely to be hit by the crisis; other pre-crisis determinants such as large current account deficits and lower holdings of international reserves also made countries vulnerable. Weaker evidence, only significant at a 5% level, exists that highly leveraged and high-credit growth are associated with crisis-incidence. Eastern European and Baltic countries have been hard-hit, which came apparent by the inclusion of geographic dummies.

The authors come up with several potential reasons that might explain why their results are so disappointing. The first explanation of the weak linkage is that the data on crisis manifestations were collected in Spring 2009 and may therefore not fully measure the total extent of the financial crisis. Nonetheless, the authors believe that the problem seems to be explaining the crisis, rather than measuring it. Hence, two more likely explanations remain. The second explanation lies in the fact that they only examined national characteristics, while the crisis may be international in nature, for instance because crises may spread contagiously or are the results of a common external shock. The third plausible explanation the authors provide is that it is rather difficult to quantify the opaque causes and linkages of the crisis.

The authors attempt to explain the differences in forecasts due to the recent crisis. They focus on the factors driving revisions of projections for GDP growth in 2009, comparing growth forecasts prior to and after the intensification of the crisis in September 2008. The explanatory variables can be divided into roughly four broad groups trade linkages, financial linkages, underlying vulnerabilities plus financial structure and, the fourth channel, the overall policy framework.

Again I will elaborate on them respectively. Countries with stronger trade linkages are more exposed to an immediate drop in their export demand. Also the composition of exports plays a role, because countries that have a diversified export portfolio are better able to absorb demand shocks of one specific good. Emerging countries with close financial ties to advanced countries are expected to be affected severely. The low yields in advanced countries combined with easier cross-country borrowing also fuelled borrowing in emerging markets. However once the crisis broke out capital flows reversed and currencies of emerging countries depreciated. This translated in financial

constraints or worse, severe balance sheet problems. Finally, a broad range of variables is included like inflation levels and exchange rate regimes to capture the strength of the policy framework.

Regression analysis shows that for emerging market countries, the main transmission channel of shocks appears to have been financial channels. Countries that borrowed more from advanced countries were significantly hit harder; the mostly affected countries had liabilities to advanced countries of about 66% of GDP, while the least affected countries of only 19% of GDP.

Vulnerabilities, particularly leverage (credit-to-deposit ratio) and the cumulative growth in bank credit in the period 2005-07 are clearly positively affect the extent of the growth revision. Leverage explains basically all the revisions for the least affected countries, roughly two thirds of the revisions for the averagely and the most affected countries. There is also weak evidence that countries’ current account balances are significantly positively related to growth revisions. Thus, this implies that countries with a positive current account tended to experience more modest downward growth revisions than countries with deficits. Significantly weak evidence suggests that countries with higher levels of international reserves experienced smaller downward growth revisions.

Countries with exchange rate pegs were hit particularly strongly. Also countries with wider fiscal deficits experienced strong growth revisions.

Claessens et al. (2010)

The authors examine how a country’s performance during and following the financial crisis of 2008-09 depend on pre-crisis conditions. Different from Rose and Spiegel they include different points of time as starting points of the crisis because they suspect that the crisis did not hit all countries at the same moment. Countries are divided into five groups based on the date they entered the recession that is identified by a negative real GDP growth rate.

The authors measure how the performance of the macro economy and of the financial sector was dependent on initial conditions. The main indicator of financial sector stress is the Financial Stress Index (FSI) that contains seven indicators: banking sector beta; TED3 spread (that measures the spread between the interest rates on interbank loans and interest rate of T-bills issued by the US government); the slope of the yield curve4 (that measures the spread between short- and long-term yields on

government securities); stock market return; stock market return volatility; sovereign debt spread, and exchange market volatility.

The performance of the macro economy is measured with the following indicators: the duration of the recession; the severity of income loss following the crisis; and the change in average growth rate in the crisis years compared to the pre-crisis period (the change in average growth comparing 2003-09 to 2008-09).

They include the following explanatory variables as indicators of initial conditions: house price appreciation; growth in bank credit-to-GDP; domestic credit/GDP; mortgage debt-to-GDP; wholesale funding dependence; fiscal balance; current account balance; foreign bank claims; trade openness and log per capita income. Most of these variables are included for the same reasons as described in the previous articles. The two newly introduced variables are wholesale funding dependence and fiscal balance. Due to the short-term nature of wholesale funding, funds need to be rolled over quickly or else a bank could be forced into liquidation. Wholesale dependence is therefore expected to be positively related to financial stress. Weak fiscal positions might reduce the government’s ability to intervene in times of downturn.

House price appreciation, bank credit growth prior to the crisis and the current account deficit are significant predictors for all three macroeconomic performance indicators. The negative relation between the current account balance and all macroeconomic indicators is also significant. Trade openness significantly affects the severity of the crisis and decline in growth in a positive manner. Foreign bank claims and per capita income are positively related to the financial crisis.

The relation between FSI and house price appreciation; domestic credit/GDP; mortgage debt-to-GDP and per capita income are -as expected- all found to be positive and significant. Remarkably, trade openness, on the contrary, is negatively related to FSI.

3

TED is an acronym formed by T-bill and ED (the symbol for Eurodollar contracts).

Table 3.2 Consolidation of all researched variables

L & M-F R&S C & Al. B & Al.

Signficance* Dependent Variable Demand Growth 2008- 09 Consump. Growth 08 - 09 Output Growth

08-09 Severity Severity Duration

Decline in Average Growth FSI Revisions for GDP growth in 2009 # SUM

Explanatory Variable Year

Trade Openness 2006 . . . . 1 0 1 1 . 3

2007 0 . 1 . . . 1

Manufacturing /GDP 2007 1 . 0 . . . 1

Oil- producing (dummy) 0 1 0 . . . 1

Change in Private Credit/ GDP 2004-07 1 0 1 . . . 2

2000-06 . . . . 1 1 1 0 . 3

2006 1 . . . 1

Domestic Bank Credit / GDP 2006 . . . 1 0 0 0 1 . 2

Mortgage Debt /GDP 2006 . . . . 0 0 0 1 . 1

Cumm. House Price Appreciation 2000-06 . . . . 1 1 1 1 . 4

Cumm. Credit Growth 2007 . . . . 0 0 0 0 1 1

Change in Stock Market / GDP 2003-06 . . 1 . . . 1

Current Account /GDP 2006 . . . 1 1 1 1 0 . 4

2007 1 0 1 . . . 2

Pegged exchange rate (dummy) 2007 0 . 1 . . . 1 2

Financial Openness 2007 0 0 1 . . . 1

Foreign Claims 2006 1 1 . 2

Trend Growth 1990-07 0 . 0 . . . 0

Growth before/during crisis 2005-07 1 . 1 . . . 2

2008-09 . 1 . . . 1

GDP per Capita 2006 . . . 1 0 1 1 1 . 4

More Leveraged Banking Sector 2006 . . . 1 . . . 1

2007 . . . 1 1

Geographic (dummy) 2006 . . . 1 . . . 1

Whole Sale Funding Dependence 2006 . . . . . 0 0 0 0 . 0

Fiscal Balance 2006 . . . . 0 0 0 0 . 0

Primary Gap 2007 . . . . 0 0 0 0 1 1

EU accession dummy 2007 . . . 0 0 0 0 1 1

* Significant with expected sign (=1) Insignificant (=0).

The table contains all the variables that were included in the four discussed articles. When a field is marked with a '. ' the variable was not included in that specific article/model. In some cases the authors used various variables as dependent variable. In that case, a separate column for each dependent variable is included in the table. The right yellow column displays the number of times a variable was found significant across the discussed articles and the tested regression models.