MASTER THESIS
INTERNATIONAL ECONOMICS AND BUSINESS
“AN ASSESSMENT OF FINANCIAL STABILITY IN CENTRAL
AND EASTERN EUROPE”
H.P. Colijn 1386727
hpcolijn@yahoo.com University of Groningen
Faculty of Economics and Business Supervisor:
CONTENTS
Preface 3
Abstract 4
1. Introduction 5
2. What is financial stability? 6
3. Why does financial stability matter? 8
4. The European situation 9
4.1 Central and Eastern Europe 9
4.2 Euro Area 11
5. Benchmark indices 12
6. Research questions 16
6.1 Behavior of financial stability and its effects on the real economy 16
6.2 Causes of financial stability 17
7. Components of the indices 18
7.1 Central and Eastern European index components 19
7.2 Euro Area index components 23
8. Index construction 26
9. Financial Instability Indices 28
10. Estimation specification 33
10.1 Convergence 33
10.3 Causes of financial stability 35
11. Empirical results 37
11.1 Behavior of financial stability and its effects on the real economy 37 11.1.1 Financial stability in the CEE countries has converged 37 11.1.2 Financial stability in the CEE countries has converged 38
towards that of the Euro Area
11.1.3 Financial stability is a good forecaster of GDP growth 39
11.2 Causes of financial stability 42
12. Conclusions 44
13. Considerations 46
References 48
Appendix I – Tables 53
PREFACE
This thesis shines light upon a matter that has been a hot topic for the past few years in economics, financial stability. The idea for this thesis came about while I was working as a temp at The Conference Board, where this was well received. When my time at TCB had come, I suggested this as a master’s thesis, which was accepted.
I want to express my great gratitude to The Conference Board for allowing me to use their data and for their support while working on the thesis. Two people I want to especially thank, being Gad Levanon and Jean-Claude Manini, that have both been very helpful in giving feedback on the matter of construction of the indices, the selection of components and the assessment of the time series. It comes without saying that this thesis would not have been anywhere near completion without them.
I would also like to thank Ralph de Haas at the European Bank for Reconstruction and Development for his kind help on the data search and component selection for the Central and Eastern European countries.
Another person I want to thank is Chris Klaver, who has been of great help regarding the analysis in this paper. His knowledge on the possible methods of index construction and exercises has been very helpful in finding the right procedures to follow.
Last but not least, I want to thank Prof. dr. Garretsen who has supervised this thesis. His guidance throughout the whole process and his knowledge of the subject have been very valuable and his enthusiasm has helped me experience this thesis as an enjoyable process and a nice end to my time at the University of Groningen.
ABSTRACT
1. INTRODUCTION
At the moment of writing this thesis, world headlines have been dominated by news on financial stability for over two years straight. The 2008 credit crunch has been followed by the European sovereign debt crisis, which has yet to be fully contained. Draconic measures all around the world have been taken to keep the system afloat and new agreements on how to prevent systemic failure are imminent.
Financial stability has been interesting for much longer as there have of course been periods of financial turmoil before. When looking at Europe for example, the Central and Eastern European (CEE) countries have seen the fall of the Berlin Wall as the starting point of a free-market economy with all of the financial stability issues related to that up to the point that most of them entered the European Union and some even the Euro Area resulting in a common currency with the countries once their political and economical opposites.
There are therefore many attempts to look at financial stability in the literature that have all contributed to the subject in their own way. The problem with the topic is that there is very little consensus on what financial stability actually is, which makes it challenging to quantify. There are many attempts available and in the literature review of this paper, a definition will carefully be chosen.
The way this study will contribute to the literature is that it creates a monthly index of financial stability for the CEE countries and an analysis of those compared to an index for the Euro Area and the United States. Besides this, the effects on real economy variables will be tested to see whether the indices created will have forecasting abilities. The general question that this thesis will center around is therefore:
2. WHAT IS FINANCIAL STABILITY?
Even though financial stability is a construct that dominates the headlines, there is very little consensus on its definition. As mentioned by Aspachs et al. (2006), in an internet search for the best definition of financial stability ‘the lack of financial instability’ turned out to be by far the most popular. As much work has been done on the definition already, a better one can definitely be found, but this states the divide in the academic world on the subject.
Stability has to relate to a certain concept that it attributes to, something is stable or unstable. In this case, the financial system is the concept that financial stability relates to. So therefore, the definition of the financial system should be laid down first to come to the definition of financial stability.
The financial system can be classified as being either bank- or market based as has been considered in Levine (2002) and Beck and Levine (2002). The conclusion is that there is no argument to make for choosing one of the two. This shows from current day systems as a hybrid version is most popular among the larger economies (Boot and Thakor, 1997). The difference between the two is that the one system relies on credit supply by the banks, while in the other the market takes over this function. Especially in the United States, there is a large market in which commercial paper is traded, while in less developed economies this function is mostly bank-based.
This means that Mishkin essentially argues that there are no general causes for instability in the financial system. There is much to say for this as studies into financial crises like Laeven and Valencia (2008) shows that there are many reasons why a financial crisis started.
Laeven and Valencia (2008) do point out types of financial crisis, so not a specific reason, but an area or sector in which the crisis occurs. They define three types of crisis which are banking crisis, currency crisis and sovereign debt crisis. A slightly different approach is used in Claessens et al. (2008) which looks at credit contractions, house price declines and equity price declines. This might be catered to the recent 2008 United States credit crisis and less applicable to general financial stability theory. Illing and Liu (2003) use the same types of crises as Laeven and Valencia (2008), but add equity crises. As this is a study that creates an index that measures financial stability, they use market variables that capture the four possible crisis areas.
3. WHY DOES FINANCIAL STABILITY MATTER?
When financial stability would not have an effect on the real economy, this would not be an interesting concept to measure. Levine (2004) argues on this matter that financial markets and intermediaries have a positive effect on growth. As he also accounts for reverse causality and still finds positive results, this can be considered rather convincing evidence. This means that financial stability is a valid construct that contributes to growth.
There is another stream of literature examining the relationship between financial and price stability. English et al. (2005) examine the forecasting abilities of financial variables on economic activity and inflation, but come to the conclusion that the variables are not good forecasters of price stability. Grabowski (2009) finds evidence for financial system data to be forecasting real economy variables for the specific case of Poland. Schwartz (1995) comes to the conclusion causality runs the other way, that price stability results in financial stability. This is because less asymmetric information exists in the credit markets and more stability exists on the balance sheets of businesses and financial institutions. Therefore, the causality runs from price stability to financial stability to economic growth.
As Crockett (1997) concludes, it still remains to be seen whether financial stability is actually something that needs to be addressed by policy makers. There is a strong line of thought that pleas for a laisser-faire attitude. Minsky (1992) argues that financial crises are inevitable. When more risk is taken, economic growth will be spurred, but this is at the cost of higher incidence of financial crises (Ranciere et al., 2006). On the other hand, Mishkin (1996) argues the other way and concludes that financial crises are undesirable and that they need to be contained as much as possible.
predictor of GDP and inflation and in Bernanke (1990) that has considered interest rates and interest rate spreads as forecasters of economic growth. Estrella and Mishkin (1997) have performed an out-of-sample exercise on a range of financial indicators as predicting variables and conclude that the yield curve spread and stock prices are good forecasters of economic growth. This would mean that an aggregate of financial variables has a promising chance of being a good forecaster of economic growth as well; this is a subject that will be touched upon later in this study.
4. THE EUROPEAN SITUATION
As this study focuses on financial stability in Central and Eastern Europe, this part of the literature will concern a short history of the region and considerations on integration and convergence within the continent as integration with the Western European countries have a large positive effect on financial stability. After this, the Euro Area as an integrated economic entity will be considered.
4.1 Central and Eastern Europe
Over the course of time, the economic and financial systems and institutions adapted to the free-market economy and its implications and therefore financial instability was less prevalent. At the turn of the millennium, the Russian crisis of 1998 had worn off and most Central European countries were gearing up for admission to the European Union. To be able to enter the EU, countries have to adhere to the Maastricht Treaty, which includes strict financial demands. This means that financial stability was one of the major focuses of the countries that entered the Union. Eastern European countries trailed the more culturally and economically Western central countries like Hungary, Czech Republic and Poland (Svenjar, 2002).
This shows from the literature as well. Babetskii et al. (2007) found evidence for stock market integration on national as well as sectoral levels between Czech Republic, Hungary, Poland and the Euro Area. Ebner (2009) looks at the yield spread between the benchmark bonds for CEE countries and Germany and finds strong heterogeneity within Central and Eastern Europe. Poghosyan (2009) finds that financial integration has occurred between Germany and a number of CEE countries and finds evidence for a gradual integration of financial markets over the period of time from 1994-2006. Brada et al. (2005) compare France and Germany to a selection of CEE countries and find cointegration for money supply and prices, but not for monetary policy and industrial production. Convergence of GDP between the Euro Area and the CEE countries is found in Matkowski and Próchniak (2005), but convergence in productivity is negligible according to Gardiner, Martin and Tyler (2004). This leads to the conclusion that there is integration and convergence between the Euro Area and the CEE countries and for the CEE countries among themselves, especially in the financial sector, but not for all macroeconomic variables.
4.2 Euro Area
In Western Europe, the euro had replaced local currencies and the Euro Area was created. Baele et al. (2004) argues that this meant that an integrated stable economic power came to existence. Baele et al. test the equity, government-bond, corporate-bond, money and credit markets for financial integration and their results indicate that the money market is completely integrated and that integration is high in the bond and equity markets. Only the credit market is not very integrated according to their findings. Hardouvelis et al. (2006) come to a different conclusion about the stock market integration. The conclusion of an exercise in which forward interest and inflation differentials were compared was that both shrank towards zero hinting at full integration. Poghosyan and De Haan (2007) investigate interest rate linkages and financial-market integration between EMU countries using rolling threshold vector error-correction models and find increasingly strong interest rate linkages between countries.
Concluding, the Euro Area has become an integrated financial system, which should not be confused with a single financial system. All the different financial systems in the Euro Area still have their own characteristics, but are now integrated. This is also what shows from the sovereign debt crisis, all the Euro Area financial systems are affected, but they do behave differently. For the sake of comparison between ‘East’ and ‘West’, the Euro Area will be treated as one financial system in this thesis, so that the differences between the CEE countries and the mature Euro Area will be clear.
5. BENCHMARK INDICES
To quantify the concept of financial stability earlier defined, an index of financial variables should be created. In this case, this will result in a number which can be used to do analysis with or warn against possible episodes of financial instability. There have been many attempts to create such measures and the most important ones are mentioned in this section.
Hakkio and Keeton (2009) have created the Kansas City Financial Stress Index, which is an index that measures financial stress in the United States’ economy. The index consists of price and yield variables that are at least available on a monthly basis. This is rather similar to the Bank of Canada Financial Stress Index, explained in Illing and Liu (2003) in which several price and yield variables that relate to the relevant financial markets are indexed by using principle components analysis.
The IMF has created a different index, because it standardizes the variables and then uses equal weights, it also uses fewer variables than in the U.S. and Canada indices as it is comparing across countries. The markets captured are the foreign exchange market, banking sector and the securities market. This makes the indices relevant for countries with a large market-based part of the financial system, but less for developing economies that do not have a large securities market.
The OECD Financial Conditions Index is explained in Guichard and Turner (2008) and Guichard, Haugh and Turner (2009) and is a weighted sum of six variables. Contrary to the indices measured before, the weighting is derived from the effect on GDP in the next four to six quarters. The Bloomberg Financial Conditions Index uses sub indices for the debt, securities and equity markets that have equal weights in the final index (Rosenberg, 2009).
detail below. It is updated on a monthly basis and is constructed using principle components analysis.
The Citi Financial Conditions Index is created using The Conference Board Coincident Economic Index component forecasting equations and is available from 1983 onwards (D’Antonio, 2008). This index measures U.S. financial conditions. Recently, Hatzius et al. (2010) have created an index that measures financial conditions in the U.S. as well, which uses a large amount of variables, 45. This index also uses the principle components analysis method of indexation and slightly outperforms some of the earlier mentioned indices in forecasting real economy variables.
The Conference Board Financial Instability Index consists of the inverted 1-month growth rate of the S&P 500 and of the S&P 500 small cap, the implied volatility of the S&P 100 index, the spread between high yield U.S. corporate bonds and the 10-year U.S. Treasury note, the spread between index BAA-rated U.S. corporate bonds and the average of the 20- and 30-year U.S. Treasury bonds, two survey variables, the spread between the Federal Funds Rate and the two-year U.S. Treasury note, the two-year U.S. treasury swap rate (fixed versus floating) and the TED spread, which is the spread between the three-month London Interbank Offering Rate (LIBOR) and three-month U.S. Treasury bill.
seems to behave similar to the economic events that have taken place and therefore it seems to capture the disruption of the normal functioning of markets quite well.
The general outcome of the indices mentioned above is good as most of them arrive at similar outcomes. There are many differences between them even though they all try to capture financial stability or stress through financial variables that are available on a timely basis. There are negative sides to all of these country indices of which three flaws are mentioned below.
Figure 1: The Conference Board Experimental Financial Instability Index.
-4 -2 0 2 4 6 8 10 12 14 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 The Conference Board Experimental Financial Instability Index
Index, normalized
have represented such a big market at the beginning of the time series. Consider the asset-backed securities market for example; this was very small in the 1980s, while it is very big at the moment. Another issue that is considered a flaw of the indices is that some markets are overrepresented in some indices. This is the case when a lot of variables that essentially measure the same movement in stability are used in the index. When this is the case and no proper weighing is used, there will be an overrepresentation of a certain market which makes the financial stability measure skewed.
The negative sides pointed towards are not easy to overcome. It might even be argued that it is not necessary to overcome as the signals provided by the indicators are good enough as they are, but it has to be said that there are downsides to the measurement of financial stability in the conventional ways that are explained above.
6. RESEARCH QUESTIONS
For this thesis, financial stability in Central and Eastern Europe will be examined and compared to that in the Euro Area. The measurements that will be used as a proxy for financial stability will be the Financial Instability Indices that will be explained later in this thesis. The main research question will be:
How has financial instability in Central and Eastern Europe developed and how does it interact with the CEE’s the real economy?
This will be examined using of six sub questions that will be answered by means of testing hypotheses.
6.1 Behavior of financial stability and its effects on the real economy
The first two questions relate to the behavior of financial stability in Central and Eastern Europe and its relationship to the Euro Area. The literature in section 4 mentions that integration of systems and convergence in stability between the systems has emerged since the CEE countries changed to the free market system. This is not only among each other, but also with the Euro Area financial system. Absolute convergence is hard to measure with the indices that are used in this thesis, as they are normalized and therefore cannot be compared on levels. This does not mean that convergence of the growth pattern of financial stability cannot be tested. As the integration of the systems means that shocks in the system of one country will be felt in the other country’s system as well, this will mean that the pattern of the financial instability indices of the CEE countries will be more alike now than 10 or 15 years ago. This will be test in the first and second question.
Has the growth pattern of financial stability in the CEE countries converged?
Hypothesis 1: The growth pattern of financial stability in the CEE countries has converged.
Has the growth pattern of financial stability in the CEE countries converged towards that in the Euro Area?
converged towards that of the Euro Area.
The third question that will be answered relates to the theoretical relationship between financial stability and GDP. One would expect financial stability to be a good forecaster of GDP as financial conditions largely influence the real economy.
Is financial stability a good forecaster of GDP growth?
Hypothesis 3: Financial stability is a good forecaster of GDP growth. 6.2 Causes of financial stability
The last questions relate to the causes of financial stability in the selected countries. Where the first three research questions are related to the behavior and effects of financial stability, these questions try to find possible causes of it. The relationship between the selected variables and financial stability will be examined. As has been highlighted from the theory, price stability is considered an important prerequisite for financial stability and the membership of the European Union and the Euro Area are supposed to have positive influences as well. As there are many other variables that explain financial stability, like for example efficiency of institutions or corruption in the system, there is no attempt to explain financial stability in full. The reasons that the three variables are used are the relationship found in the literature between price stability and financial stability on the one hand and the European history explained before on the other. Therefore, the last three sub questions are the following:
What effect does inflation have on financial stability?
Hypothesis 4: Inflation has a negative effect on financial stability.
What effect does membership of the European Union have on financial stability?
Hypothesis 5: Membership of the European Union has a positive effect on financial stability.
What effect does membership of the Euro Area have on financial stability?
7. COMPONENTS OF THE INDICES
To be able to answer the research questions, a measure for financial stability has to be developed. In this case, an index will be created to measure different signals that correspond to the various markets in which stability can be disrupted. The index should contain variables that measure signals from all the four markets, so signals that represent the currency, equity, debt and banking market. The variables considered had to meet several criteria to be included in the index.
available for a large amount of the countries used in this thesis. When this is not the case and the indices are made up of changing variables, there is no possibility for comparison. 7.1 Central and Eastern European index components
The components of the indexes for the Central and Eastern European countries will not be the same as the ones for the United States that The Conference Board uses for their experimental index, nor for the Euro Area Financial Instability Index. This is because the data frequency and availability in the more emerging CEE markets are less than in the advanced economies. The financial markets also substantially differ and therefore a different range of variables have been selected to best capture the financial markets in CEE. The variables that have been chosen met all of the before mentioned criteria and are the best available at capturing the movements in stability in the financial markets. The variables selected are the following:
• Spread between the 3-month interbank rate and the 3-month Euribor
• Spread between the 3-month interbank rate and the overnight interbank rate • Spread between the benchmark bond yield and the German benchmark bond yield • Inverted 1-month growth rate of the main stock market index
• Volatility of the options on the exchange rate against the euro • Credit Default Swap
• Euro Area Financial Instability Index • (Implied exchange rate)
Spread between the 3-month interbank rate and the 3-month Euribor
Spread between the 3-month interbank rate and the overnight interbank rate
Like the previous variable, this is a measure of financial stress in the banking system. The spread measures the difference in interest that is required to borrow for 1 day and for 3 months. When the 3-month rate is high, this means that there is less trust because probability of default in those three months is assessed higher. When the overnight rate is much higher than the 3-month rate, this means that overnight lending is considered very risky and trust is at an acute low. This happens in the most severe and sudden crises and therefore this is also a sign of instability in the system. For this reason, the spread is absolute, so that the number turns up positive when the spread becomes larger in both cases.
Spread between the benchmark bond yield and the German benchmark bond yield
This is a common indicator of stability in the debt market as it captures the government debt yield against the relatively risk free German benchmark bond. The higher the yield, the higher the chance of default and this is why this is a good measure of financial stability.
Inverted 1-month growth rate of the main stock market index
This variable shows shocks in share prices over a month. When this gives a very high value, share prices have plummeted and financial instability in the equity market is indicated. This is an indicator that might not be a very reliable signal of financial stability in many CEE countries, as the markets are small and sometimes dominated by a small amount of participants. This means that shocks in the market might not necessarily reflect financial instability, but possibly just the pulling out of a big player. Nonetheless, it has been included in the index because even though it is volatile and might give false signals of financial instability, it does meet the other requirements very well. The volatility of this component will be smoothed because of the indexation.
Volatility of the options on the 1-month forward exchange rate against the euro
activity in a country, which leads to financial instability. As the foreign exchange market is a market that detects activity early, this is a good measure of financial stability. The variable is unfortunately only available for four countries, but as it behaves well, it is included in the index for those countries.
Credit Default Swap (1-year seniority)
This variable is a controversial one to include as the CDS might be forbidden in the European markets and because they are usually not very liquid. Besides this, the data does not go further back than 2003 at best which is not long either. In this small time series it does surprisingly well in picking up on financial stress in the bond market as it is an indicator of sovereign default. The 2007 financial crisis is picked up very well and the current sovereign debt crisis shows a new peak that indicates financial instability for many countries, especially Greece.
This good performance of the past episodes of financial stress is confirmed by Hekuran (2008), which studies CDS as a predictor of financial crises in emerging markets and concludes that indeed it is a good predictor. It also shows that there are significant links with the currency- and equity markets as investors use this as a general measure of stability in the emerging economy. As can be seen in figure 2, the pattern of the CDS variable is similar to that of the other variables included as there is not much change in the index when the CDS variable is left out.
Euro Area Financial Instability Index
Figure 2: Financial Instability Index for Czech Republic with and without the Credit Default Swap variable included.
-4 -2 0 2 4 6 8 10 12 14 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 With CDS Without CDS Index, normalized
Implied exchange rate
The 3-month forward exchange rate against the euro less the spot rate against the euro is used to measure the differences in interest rate expectations of investors. This is a variable that measures essentially the same as the spread between the interbank rate and the Euribor does and therefore it is used as a substitution for this variable when it is not available for a country. When compared for countries that have both, it becomes evident that there is little difference in movement and therefore high correlation between the variables exists.
7.2 Euro Area index components
For the more advanced Euro Area financial markets, different variables are available to give an accurate view of financial stability in the area. This is because markets are more mature and therefore provide more accurate information about the financial situation and because more data is available for the advanced financial markets. The variables resemble the variables included in The Conference Board U.S. Financial Instability Index mentioned in the literature review. This means that the variables included are the equivalent of the U.S. variables, except for the credit default swap and the volatility of the option on the 1-week forward exchange rate of the euro that have been added as an improvement of the U.S. basket of variables and the survey variables have been left out due to the lack of availability. The mimicking of the U.S. index is to be able to make adequate comparisons with the U.S. financial market that is also advanced and mature, like the Euro Area. The variables included are the following:
• Credit Default Swap
• Spread between high yield corporate bonds and the 10-year German Treasury note
• Spread between the BAA-rated corporate bonds and the average of the 20-year and 30-year German Treasury bonds
• Inverted 1-month growth rate of the Euro Stoxx price index • Inverted 1-month growth rate of the Euro Stoxx small cap index • Implied volatility of the Euro Stoxx price index (Vstoxx)
• Spread between the European Central Bank short term repo rate and the 2-year German Treasury note
• German 2-year Treasury swap rate (fixed versus floating)
• Volatility of the option on the 1-week forward exchange rate of the euro
Credit Default Swap (1-year seniority)
This variable is the same as is used in the CEE Financial Instability Indices.
Spread between high yield corporate bonds and the 10-year German Treasury note
This variable measures the probability of default in the corporate sector together with the spread between the BAA-rated corporate bonds and the average of the 20-year and 30-year German Treasury bonds. This measure shows the difference in yield that the market requires to compensate for default risk. When the debt market becomes disturbed, this yield will rise. The German note has been used, this is the case for more variables, as it is considered the most risk free in Europe and this variable is supposed to show the risk of high yield corporate bonds.
Spread between the BAA-rated corporate bonds and the average of the 20-year and 30-year German Treasury bonds
Similar to the above mentioned variable, even though at times the behavior differs.
Inverted 1-month growth rate of the Euro Stoxx Price Index
This variable measures the financial stability in the stock markets together with the small cap index. The number is inverted because a fall in stock prices represents negative expectations and therefore hints at financial instability. This is a very volatile measure as it is a 1-month growth rate and therefore has a lot of false movement. Because of the fact that this is aggregated into an index, the noise will be filtered.
Inverted 1-month growth rate of the Euro Stoxx small cap index
Similar to the above mentioned variable, even though at times the behavior differs.
Implied volatility of the Euro Stoxx price index (Vstoxx)
time to expiration. When the measure is high, this means that uncertainty about equity prices is high, which has a negative influence on financial stability.
Spread between the European Central Bank short term repo rate and the 2-year German Treasury note
This is a variable that shows the expectations of the ECB rate to change. This is high in periods of financial instability because people will anticipate a fall in the rate. When the ECB rate is close to zero, there is the problem of giving a troubled view of the financial situation as the rate cannot go below zero.
German 2-year Treasury swap rate (fixed versus floating)
This variable represents exchanging German treasury notes with fixed interest payments for notes with floating payments and therefore the number represents the risk associated with holding the notes with the floating payment. When the risk is higher, the yield demanded for holding notes with floating rates has to increase and therefore this measure increases with financial instability.
Spread between the 3-month euribor and the 3-month eonia swap index
This variable is the European equivalent of the TED spread, as the euribor is the interest rate at which banks lend to each other and the eonia swap index is the daily average of overnight rates, like the Federal Funds Rate. When this spread widens, banks charge each other higher interest, which means that the trust between banks has decreased and this negatively influences financial stability.
Volatility of the option on the 1-week forward exchange rate of the euro
This variable is the same as the variable used in the indices for the CEE countries only for a week ahead.
8. INDEX CONSTRUCTION
The several independent variables that give out signals of financial instability will be taken together and indexed. The method that is used to create the indices is principal components analysis (PCA). This is an exercise for exploratory data analysis and in this case it will be used to reduce dimensions without sacrificing much accuracy. This means that it creates one variable out of all of the financial stability components and therefore one number is created. For this to happen, all variables will be normalized first as most variables are on different scales and variables with high numerical values will dominate total variance and trouble analysis. This type of indexation is conform prior research as can be found in the literature reviewed earlier.
For every country that the indices will be created for, there are n variables that will be used, with n≥3. ni i i i Υ Υ Υ Υ1, 2, 3 ,K,
In order to be able to distinguish which variables are used for a particular country j, we introduce an index set Ijfor each country. If for example the relevant variables for country 3 are the variables Υ1i,Υ2i,Υ3i,Υ4i, then I3 ={1,2,3,4}. A concise formulation for the factor model can now be presented.
ij ki kj I k ij j
Z
=
Υ
+
∈
∈∑
α
, (1)where the sought after weights for the j-th country are represented by αkj , the term ij
Z holds the factor score for measurement i and country j and the term εijdenotes the error term. Also note that i∈{1,...,T}.
T ij
U
V
X
=
Σ
,with U and V both column-wise orthonormal matrices and ∑ a diagonal matrix, with its element ordered such that the largest value is top left and the smallest bottom right. Now an elementary result from Eckart and Young (1936) shows us that the best one factor approximation can be derived from the truncated version of the singular value decomposition. In particular, in order to find a solution to (1) we compute the singular value decomposition and choose the values of Zij such that it coincides with the first column of U times T1/2 and the values of αij, such that they coincide with the first column of V times T-1/2 times the top left value of ∑ .
9. FINANCIAL INSTABILITY INDICES
The financial instability indices that have been created will be assessed in this section. The fact that there is no reference for the series makes it very difficult to properly assess. When the index goes up even though there was no notable event that might explain this, it seems that the index has falsely assessed the financial system unstable. On the other hand, one might argue that apparently the system was more unstable than the world was aware of. This issue is one that is hard to find a solution for and with the lack of a better assessment, the indices will be weighed by comparing them against events in the financial system. This will not explain every small movement of the indicator, but it will filter out the larger peaks. In this section, this will happen by assessing groups of indices, an individual assessment of all indices can be found in the appendix.
Most of the indices show a rather similar pattern, which comes from high financial instability and gradually moves downward towards a state of relatively stable times midway through the 2000s. This can be seen in figure 3 in which the Central European states are pictured. This pattern is roughly in line with the history of the CEE economies that has been explained earlier in this thesis.
Figure 3 – Financial Instability Indices for Czech Republic, Hungary, Poland and Slovakia. -2 -1 0 1 2 3 4 5 6 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 Czech Republic Hungary Poland Slovakia Index, normalized
It also becomes clear from figure 3 that the most recent episode of financial instability, the 2007 credit crunch, is picked up perfectly in every index. Even though some of the indices show that the effects were less than they have been in other episodes in the 1990s, it has been the most severe peak in instability in the financial system in years for all countries. The fact that there seems to be a joined movement in the indices also shows from table 9 in Appendix I, that there is indeed high correlation between the growth patterns of all the indices. It has to be kept in mind that levels cannot be compared, because of the normalization, but growth patterns are comparable as they are not dependent on levels.
Figure 4 – Financial Instability Indices for Cyprus, Lithuania and Romania. -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Cyprus Lithuania Romania Index, normalized Euro Area
Figure 5 – Financial Instability Index for the Euro Area -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Index, normalized
Figure 6 - Financial Instability Indices for the United States, Euro Area and Poland. -4 -2 0 2 4 6 8 10 12 14 16 -3 -2 -1 0 1 2 3 4 5 6 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 United States (The Conference Board, experimental) Euro Area (right scale)
Poland (right scale)
10. ESTIMATION SPECIFICATION
The indices that serve as a measure for financial stability will now be used to answer the research questions posed earlier in this thesis. The estimation strategy used will be explained in this section, in order of relevance for the hypotheses.
10.1 Convergence
The test that will be used to measure whether convergence exists in the growth path of financial stability in the CEE countries is the Barro convergence test (Barro, 1991). This is a test in which an ordinary least squares framework is used to test whether financial stability growth in the CEE countries has converged or diverged. It uses the growth rate of the period for which convergence is tested as the dependent variable and the log of the level of financial stability in the first period as independent variable (2). This is usually used to compare levels, but as this is not a comparison of levels of financial stability, the interpretation of the outcome is somewhat different. When the indices have converged towards each other, it means that the growth paths of financial stability have synchronized. That is what is being tested in this case, if the CEE financial stability patterns have synchronized over the period that is being tested.
∈ + +
=
∆FII2000−2010 β0 β1FII2000 (2)
Another tool that will be used to test convergence is correlation between the indices for separate time groups. This is useful because the correlation is larger when the series move in the same direction and therefore it shows convergence of growth paths.
10.2 Forecasting GDP
tested by turning point analysis. The difference lies in the fact that turning point forecasting serves as a warning signal for the economy to turn and growth forecasting shows whether the forecasting variable is able to forecast the growth of the economy per month.
In-sample forecasting is an exercise that uses an ordinary least squares framework to determine the amount of variation explained by the independent variables. To see whether the independent variable that is assumed to have forecasting ability performs, a base regression is run using just the dependent variable and lags of the dependent variable as independent variables. This shows the amount of variation of the dependent variable explained by the previous movement of the variable itself. Now the forecasting variable, in this case the financial instability index, can be added to the equation as an independent variable.
The exercise (3) performed uses the change in GDP as dependent variable and therefore the lags of GDP as independent variable. As a period of financial instability to take effect on the real economy two quarters is assumed, which means that two lags of GDP growth are used. This exercise tests how well the trend of the dependent variable forecasts the dependent variable itself is.
∈ + ∆ + ∆ + = ∆GDP β0 β1 GDP−1 β2 GDP−2 (3)
After this, the second exercise (4) will be performed, which is essentially the first exercise, but two lags of the financial instability index will be added.
∈ + ∆ + ∆ + ∆ + ∆ + = ∆GDP β0 β1 GDP−1 β2 GDP−2 β3 FII−1 β4 FII−2 (4) The increase of the r² will be the measure of performance of the index. This will be measured against a benchmark of variables, being the variables of which the index is constructed. The variable’s inclusion that leads to the highest increase explains the most variance in the dependent variable and is therefore the best forecasting variable.
other as they measure the average of the distance from the forecast to the actual number. The exercises are the same as (3) and (4) as the lags of the dependent variable are used as independent variables in the base regression and the lags of the forecasting variable are added in (4).
Out-of-sample forecasting means that a certain period is used to estimate the coefficients of the independent variables and that for every quarter after that the value for the dependent variable is estimated. The estimation period will be from the beginning of the time series to the first quarter of 2003 in this case and for every quarter after that will be forecasted. This will therefore first be done for quarter 2 of 2003. After this, the regression will be reestimated and the new coefficients will be used for the forecast for the subsequent quarter. This repeats itself until the last quarter with available data. When the MSE of one forecasting variable is lower than the other, less distance from the actual number was measured in the forecasts and therefore better forecasting has been done. Turning point analysis is the exercise that measures the ability of the financial instability indices to detect a turning point in GDP growth. This means that it essentially measures the ability to detect declining growth and recessions. The peaks and troughs in both GDP and the financial instability indices are compared and false signals and missing turns in the indices are taken into account. To determine the peaks and troughs in the series, the Bry-Boschan algorithm is used (Bry and Boschan, 1971).
When the before mentioned three exercises are performed, a conclusion can be drawn about the forecasting ability of financial instability on GDP growth.
10.3 Causes of financial stability
performed, depending on the result of the Hausman test. This is a test in which both regressions are performed and the estimates of the coefficients are then compared. Under the null-hypothesis both estimators are consistent, when this is not the case, the null-hypothesis is rejected. In that case, the random effects regression will be used, but when the null-hypothesis is accepted, a fixed effects regression will suffice as the omitted variables may differ between cases but are constant over time. When this is completed, the proper test can be carried out, which in this case is the fixed effect model (5).
it kit k it b b X e FII = 0 +∑ + (5) where it t i it u v w e = + + (6)
In this model, k indexes the independent variables, i indexes the countries used in the model, t the time periods and e the error terms. In the second equation, the error terms are explained by u, which is the country component of the error term, v, which is the time component and w, which is the random error component.
11. EMPIRICAL RESULTS
11.1 Behavior of financial stability and its effects on the real economy.
In this section, the results of the empirical research will be presented. The six hypotheses tested will be discussed.
11.1.1 Financial stability in the CEE countries has converged.
For the convergence hypotheses, a selection of the countries has been made, relating to the beginning of the time series of the FII for the country and the geographic location. Iceland for example, has been left out of the first two hypotheses because it is not located in Central and Eastern Europe. The countries used can be found in table 1. It has to be noted that levels of financial stability cannot be compared, because of normalization of the indices. The convergences examined are on financial stability growth, does financial stability growth follow a more or less similar pattern among countries at the end than at the beginning of the time series.
large sub periods of 1999 - 2004 and 2005 – 2010. As can be seen from the smaller sub periods of three years, this is largely due to the high correlations during the credit crisis and the European sovereign debt crisis. The period beforehand shows lower correlation.
Table 1 – Correlation between the Financial Instability Indices of the Central and Eastern European countries.
Country 1999 - 2004 2005-2010 1999 - 2001 2002 - 2004 2005-2007 2008-2010 Cyprus 0.02 0.57 -0.20 -0.20 0.04 0.29 Czech Republic 0.44 0.91 -0.11 0.50 0.42 0.83 Estonia 0.60 0.89 0.17 0.52 0.33 0.78 Greece 0.12 0.81 -0.14 0.47 0.31 0.57 Hungary 0.55 0.89 0.22 0.13 0.26 0.83 Latvia 0.62 0.84 0.25 0.52 0.25 0.69 Lithuania 0.54 0.86 0.16 0.47 0.42 0.73 Poland 0.56 0.88 -0.13 0.45 0.31 0.82 Romania 0.47 0.90 0.11 0.28 0.08 0.84 Russia 0.56 0.86 0.16 0.54 0.05 0.83 Slovakia 0.58 0.87 0.19 0.54 0.21 0.80 Average 0.46 0.84 0.06 0.38 0.24 0.73
When looking at convergence between the CEE Financial Instability Indices, a mixed view emerges. On the one hand it is clear that financial growth is more synchronized in the 2000s than before. On the other hand, the most recent period shows that the financial crisis has resulted in different growth patterns of financial stability throughout Europe. All in all, the Barro convergence test gives weak signs for convergence, while the correlation test gives a stronger argument for correlation. As this is not a clear case for convergence, the first hypothesis can be rejected.
11.1.2 Financial stability in the CEE countries has converged towards that of the Euro Area.
Table 2 – Correlation between the Financial Instability Indices for the CEE countries and the Euro Area.
Country 1999 - 2004 2005-2010 1999 - 2001 2002 - 2004 2005-2007 2008-2010 Cyprus 0.19 0.78 0.85 -0.40 0.53 0.74 Czech Republic 0.53 0.91 0.72 0.44 0.61 0.81 Estonia 0.55 0.90 0.13 0.95 0.89 0.76 Greece 0.85 0.81 0.81 0.89 0.53 0.51 Hungary -0.07 0.93 -0.60 0.27 0.23 0.88 Latvia 0.23 0.77 -0.24 0.88 0.78 0.42 Lithuania -0.13 0.79 -0.76 0.86 0.90 0.52 Poland 0.26 0.88 0.26 0.28 0.27 0.75 Romania -0.25 0.91 -0.78 0.15 -0.08 0.85 Russia 0.05 0.87 -0.75 0.73 -0.15 0.81 Slovakia -0.05 0.95 -0.78 0.71 0.60 0.94 Average 0.20 0.86 -0.10 0.52 0.47 0.73
In conclusion, the hypothesis can be accepted, as the correlation points towards convergence between the regions. This means that the financial systems in the Central and Eastern European countries integrated somewhat with that of the Euro Area, as the signals from the several markets have been picked up in both the CEE systems and that of the Euro Area.
11.1.3 Financial stability is a good forecaster of GDP growth.
The third hypothesis concerns the forecasting ability of GDP of financial stability. The three exercises performed to assess this ability will be shown below, starting with the in sample forecasting results.
forecaster, while only in the case of Bulgaria and Iceland the country’s own FII performed best. This hints at the fact that the Euro Area index could serve as a forecaster for GDP in all the CEE countries as well.
Table 3 – Average r² for the in sample forecasts.
Base Incl FII Incl moneymrkt. Incl yieldspr. Incl stockmrkt. Incl EA FII Bulgaria 0,171 0,483 0,442 0,425 0,313 0,270 Cyprus 0,162 0,361 0,334 0,354 0,677 0,502 Czech Republic 0,317 0,548 0,359 0,306 0,543 0,647 Estonia 0,166 0,537 0,434 0,580 0,598 0,569 Euro Area 0,430 0,646 - - 0,531 -Greece 0,234 0,246 0,262 0,270 0,242 0,268 Hungary 0,617 0,680 0,642 0,659 0,677 0,710 Iceland 0,085 0,382 0,163 0,304 0,221 0,373 Latvia 0,110 0,546 0,308 0,300 0,473 0,576 Lithuania 0,135 0,196 0,135 0,180 0,557 0,571 Poland 0,065 0,127 0,224 0,110 0,072 0,080 Romania 0,264 0,490 0,271 0,787 0,307 0,398 Russia 0,229 0,275 0,232 0,526 0,286 0,604 Slovakia 0,014 0,345 0,054 0,065 0,122 0,460 Slovenia 0,268 0,355 - 0,357 0,434 0,299 Turkey 0,060 0,272 - - 0,340 0,228 Ukraine 0,093 0,300 - - 0,215 0,449 Average 0,201 0,399 0,297 0,373 0,389 0,438
Note 1: The abbreviated variables are the following: FII is the Financial Instability Index for the specific country, moneymrkt. is the spread between the specific country’s 3-month interbank rate and the 3-month euribor, yieldspr. is the spread between the specific country’s benchmark bond yield and the German benchmark bond yield, stockmrkt. is the inverted 1-month stock market index and the EA FII is the Financial Instability Index for the Euro Area. This also holds for table 4.
Note 2: Best performance in bold.
and the Euro Area FII also perform better on average and therefore the results do not confirm the Financial Instability Indices as good forecasters of growth in GDP.
Table 4 – Average mean squared errors for the out of sample forecasts.
Base Incl FII Incl moneymrkt. Incl yieldspr. Incl stockmrkt. Incl EA FII
Bulgaria - - - -Cyprus 0,00005 0,00005 0,00005 0,00005 - 0,00003 Czech Republic 0,00012 0,00009 0,00014 0,00018 0,00009 0,00007 Estonia 0,00053 0,00045 0,00061 0,00045 0,00029 0,00035 Euro Area 0,00006 0,00005 - - 0,00005 -Greece 0,00008 0,00010 0,00008 0,00010 0,00015 0,00009 Hungary 0,00007 0,00006 0,00007 0,00008 0,00006 0,00007 Iceland 0,00081 0,00057 0,00103 0,00055 0,00067 0,00051 Latvia 0,00102 0,00114 0,00168 0,00235 0,00138 0,00080 Lithuania 0,00256 0,00377 0,00491 0,00864 0,00248 0,00145 Poland 0,00012 0,00009 0,00014 0,00011 0,00015 0,00010 Romania 0,00042 0,00034 0,00049 - 0,00049 0,00043 Russia 0,00029 0,00035 0,00031 - 0,00027 0,00021 Slovakia 0,00056 0,00041 0,00059 0,00062 0,00054 0,00050 Slovenia - - - -Turkey - - - -Ukraine - - - -Average 0,00051 0,00057 0,00084 0,00131 0,00055 0,00039
Note: The short history of Bulgaria, Slovenia, Turkey and Ukraine did not allow for out of sample forecasting.
Financial Instability Indices are good turning point indicators and that they outperform the individual components at forecasting turning points.
When evaluating the three performed exercises, a mixed picture emerges. It turns out that the FIIs are a good forecaster of turning points, but that growth forecasting gives better results when the Euro Area Financial Instability Index is used. This does not necessarily mean that the FIIs are useless at growth forecasting, because in both the out of sample and the in sample forecasts it was the best performer for some countries. All in all, it can be said that the FIIs are a reasonable forecaster of GDP, but that there are some flaws to be noted in both growth and turning point forecasting and therefore the hypothesis can be rejected.
11.2 Causes of financial stability
For the last three hypotheses, possible causes of financial instability will be assessed. This will be done using the monthly data for all countries, which has been grouped in a panel. To see what the proper model is that has to be used to assess the causes, first the Hausman test will be performed. The results can be found in table 9.
Table 5 – Hausman test results.
(b) (B) (b-B)
fixed random Difference Inflation 0,12553 0,11248 0,013056 EU -0,36728 -0,30249 -0,06479 Euro 0,67993 0,45982 0,220112 chi 2 22.11 Prob. 0.0001 Coefficients
membership has a significant negative influence on financial instability, which is conform expectations about this. The Euro Area membership on the other hand has a significant positive influence on financial instability, even though the positive stabilizing benefits are obvious. The reason for this might be that most countries adopting the euro have done so right before the credit crunch, creating a causal effect that relates to the recession more than the membership. This can easily be tested for by excluding all data from after the fall of Lehman Brothers, but even without that data the euro dummy stills remains significant at a 5% level with a negative influence on financial stability. As the variable is a dummy variable and there are few variables that change over the time series, this may lead to questionable results. This means that it becomes valuable to leave the variable out of the model and this leads to the same significant results for inflation and EU membership as found in the model including the euro variable.
Table 6 – Fixed effects generalized least squares regression coefficients, t-values in parentheses under coefficient.
Full sample Sample until Aug. 2008 Full sample Sample until Aug. 2008
Inflation 0.126 0.153 0.125 0.153 (-6.98) (10.52) (6.9) (10.51) EU -0.367 -0.856 -0.286 -0.814 (-7.03) (-18.89) (-5.59) (-18.43) Euro 0.680 0.231 (6.37) (2.05) Constant -0.006 -0.025 0.032 -0.013 (-0.2) (-0.96) (1.01) (-0.51) overall r² 0.05 0.18 0.03 0.16 N 2165 2165 1841 1841
Full model Model without euro variable
12. CONCLUSIONS
The Financial Instability Indices for Central and Eastern Europe are meant to serve as a proxy for financial stability in the countries and in that purpose, they seem to serve rather well. The four markets that are covered in the indices, the equity market, the currency market, the banking market and the debt market all capture different episodes of financial stress and together they measure peaks in financial instability that coincide with events in which the financial system was under stress. The similarities with The Conference Board Financial Instability Index for the United States and the peaks in the indices that coincide with financial or economical events encourage the belief that this is a helpful tool in assessing the stability of the financial system.
The hypotheses researched use the indices as a proxy for financial stability and the findings are interesting. Convergence in growth between the CEE countries was expected, because the pattern of financial stability movement was expected to behave more similarly as the openness and integration of the systems progressed over time during the 1990s and 2000s. Evidence for this was mixed, hinting at integration of systems, but with their very own characteristics and risks. Evidence for this can currently be seen in the Euro Area, as the Greek, Portuguese and Irish financial systems are almost faltering, while the Northern European financial systems seem to be rather stable. Convergence with the Euro Area was expected for the same reasons and evidence for this is a little stronger, with the recent crisis years showing high correlations. This means that the Western- and Eastern European blocks seem to have converged in financial stability over the past years, which is in line with research on European financial integration mentioned earlier in this thesis.
financial stability in the Euro Area than financial stability in the countries itself. This could well be attributed to the fact that the banking system in a large majority of the countries is dominated by Western European banks and that therefore the financial stability in the country of origin of those banks is more important. The fact that the Euro Area is an important trade partner of the CEE countries is an important contributor to this as well.
The Financial Instability Indices for the CEE countries do serve best at signaling turning points in the economy and therefore they do seem to be valuable in forecasting recessions. This is a positive result, because there is not that much data that comes out in a timely manner across CEE.
The last three hypotheses that have been tested show interesting results. The fixed effects model shows that there is a very strong positive link between price stability and financial stability, confirming the conclusions from Schwartz (1995). The same strong positive relationship was found for the EU membership. This is conform expectations as the EU membership comes with strict financial rules and the open economic character of the Union is something that could increase financial stability across nations.
European countries has gone over in the 2000s that marked integration and stabilization until 2007 after which a period of increased financial instability has started for countries from Western Europe to Russia and Turkey.
13. CONSIDERATIONS
The Financial Instability Indices that have been created and used for research into the financial situation in Europe have given promising and interesting results. As a measure, this might be interesting to use for investors to see changes in risk, or policymakers to monitor their financial systems. Despite that fact that the index seems to perform well as an indicator of financial stability, there are some drawbacks that need to be mentioned. The first is one that has already been addressed in this thesis, being that fact that the index cannot be properly benchmarked. This is because the concept of stability is something that is not quantified and therefore it is hard to validate the indices. When an index of financial instability increases when crowds are queuing at ATM machines, one assumes a correct increase has taken place. The question remains however how high an increase should actually be. On the same note, at what level of the index will a financial system falter? This issue has merely been noted in this thesis, but further work in the field of financial instability indices will need to be addressed to increase accuracy and reliability of indices on financial systems.
A second drawback is that the indices cannot directly be compared with one another. The indices have all been normalized to make comparison between included variables possible. This implies that the mean of the indices are all zero and therefore comparison between the levels of financial stability is not possible. This means that it is impossible to tell whether the financial system in the one country is more stable than the other.
The last thing that should be kept in mind while looking at the Financial Instability Indices is that the structure of the financial system is constantly changing and that at some point some market might be more important than the other. This has not been accounted for in this study, but as the CEE economies are maturing, a shift towards commercial debt might take place as we have seen in the United States and Western Europe as well. If the right information is available for countries, it would be preferable to use a dynamic index with changing variables and weights corresponding to the changing financial system.
REFERENCES
Aghion, P. and Blanchard, O.J. 1998. On privatization methods in Eastern Europe and their implications. Economics of Transition, Vol. 6, No. 1, pp. 87-99.
Albulescu, C.T. 2010. Forecasting Romanian financial sector stability using a stochastic simulation model. Romanian Journal of Economic Forecasting. No. 1, pp. 81-98. Aspachs, O., Goodhart, C.A.E., Segoviano, M., Tsomocos, D.P. and Zicchino, L. 2006.
Searching for a metric for financial stability. LSE Financial Markets Group Special Paper Series. No. 167.
Babetskii, I., Komárek, L., Komárková, Z. 2007. Financial integration of stock markets among new EU member states and the euro area. Czech Journal of Economics and
Finance, Vol. 57, No. 7-8, pp. 341-362.
Baele, L., Ferrando, A., Hordahl, P., Krylova, E. and Monnet, C. 2004. Measuring European financial integration. Oxford Review of Economic Policy, Vol. 20, No. 4, pp. 509-530.
Barro, R.J. 1991. Economic Growth in a Cross Section of Countries. The Quarterly
Journal of Economics, Vol. 106, No. 2, pp. 407-443.
Beck, T. and Levine, R. 2002. Industry growth and capital allocation: does having a market-or bank-based system matter? Journal of Financial Economics, Vol. 64, No. 2, pp. 147-180.
Bernanke, B.S. 1990. On the predictive power of interest rates and interest rate spreads.
NBER working paper, No. w3486.
Boot, A.W.A. and Thakor, A.V. 1997. Financial system architecture. Review of
Financial Studies, Vol. 10, No. 3, pp. 693-733.
Bry, G. and Boschan, C. 1971. Cyclical analysis of time series: selected procedures and
computer programs. New York: National Bureau of Economic Research.
Caprio, G. and Klingebiel, D. 1999. Episodes of systemic and borderline financial crises.
World Bank Discussion Paper, World Bank, October.
Claessens, S., Kose, M.A. and Terrones, M.E. 2009. What happens during recessions, crunches and busts? Journal of Economic Policy, Vol. 24, No. 60, pp. 653-700.
Conference Board, The 2001. Business Cycle Indicators Handbook. New York: The Conference Board.
Crockett, A. 1997. Why is financial stability a goal of public policy? Federal Reserve
Bank of Kansas City Economic Review, No. Q IV, pp. 5-22.
D’Antonio, P. 2008. A View of the U.S. Subprime Crisis. In: EMA Special Report, Citi Group Global Markets, pp. 26-28.
Ebner, A. 2009. An empirical analysis on the determinants of CEE government bond spreads. Emerging Markets Review, Vol. 10, pp. 97-121.
Eckart, C. and Young, G. 1936. The approximation of one matrix by another of lower rank. Psychometrika, Vol. 1, pp. 211-218.
English, W. and Tsatsaronis, K. and Zoli, E. 2005. Assessing the predictive power of measures of financial conditions for macroeconomic variables. BIS Papers, No. 22, pp. 228-252.
Estrella, A. and Mishkin, F.S. 1998. Predicting US recessions: financial variables as leading indicators. Review of Economics and Statistics, Vol. 80, No. 1, pp. 45-61.
Grabowski, S. 2009. The Financial Indicators Leading Real Economic Activity – the Case of Poland. Central European Journal of Economic Modelling and Econometrics, Vol. 1, No. 4, pp. 311-332.
Guichard, S. and Haugh, D. and Turner, D. 2009. Quantifying the effect of financial conditions in the euro area, Japan, United Kingdom and United States. OECD
Economics Department Working Papers, No. 677.
Guichard, S. and Turner, D. 2008. Quantifying the Effect of Financial Conditions on US Activity. OECD Economics Department Working Papers, No. 635.
Hakkio, C.S. and Keeton, W.R. 2009. Financial stress: what is it, how can it be measured, and why does it matter? Federal Reserve Bank of Kansas City Economic Review, No. Q II, pp. 5-50.
Hardouvelis, G.A. and Malliaropulos, D. and Priestley, R. 2006. EMU and European Stock Market Integration. Journal of Business, Vol. 79, No. 1, pp. 365-392.
Hatzius, J. and Hooper, P. and Mishkin, F.S. and Schoenholtz, K.L. and Watson, M.W. 2010. Financial Conditions Indexes: A Fresh Look after the Financial Crisis. NBER
Working Paper, No. 16150.
Hekuran, N. 2009. Can Credit Default Swaps Predict Financial Crises? Empirical Study On Emerging Markets. Journal of Applied Economic Sciences, No. 7, pp. 125-136. Jolliffe, I. 1986. Principal Component Analysis. New York: Springer-Verlag.
Illing, M. and Liu, Y. 2006. Measuring financial stress in a developed country: An application to Canada. Journal of Financial Stability, Vol. 2, No. 3, pp. 243-265.
Laeven, L. and Valencia, F. 2008. Systemic Banking Crises: A New Database. IMF
Working Papers, No. 08/224.
Levine, R. 2002. Bank-Based or Market-Based Financial Systems: Which Is Better?
Journal of Financial Intermediation, Vol. 11, No. 4, pp. 398-428.
