Measuring the Impact of Financial Crisis on International Markets: An Application of the Financial Stress Index

The scope of paper is to examine whether the recent financial crisis has had any impact on international capital markets and more precisely on the 4 primary international stock markets of England, France, Japan, the United States and Greece. The research is based on the use of the Financial Stress Index (FSI) from July 2005 until December 2008 and August 2009. Research results showed that the recent financial crisis has had a negative impact on all examined markets, with the Tokyo stock exchange being the one mostly affected. It was, also, found increased variability of performances following the start of the financial crisis, a fact that is indicative of the presence of conditional heteroscedasticity. As far as the Greek market is concerned, the recent financial crisis has not affected in general the credit expansion towards enterprises and households; however, it has affected the credit expansion to enterprises and households on a case-to-case basis.


Introduction
As is widely known, the recent financial crisis was initially manifested in the US real estate market in August 2007 (Reinhart & Rogoff, 2008) and had as primary causes the drop of interest rates, the broad expansion of high-risk housing loans (subprime lending), the transfer of credit risk from the banks' balance statements to investors through securitisation (Ephraim & Kassimatis, 2004) and the strong demand in terms of hedge funds for securitisation bonds (Petroff, 2008, Boston, 2007.
The purpose of this paper is to investigate the impact of the recent financial crisis on the 4 primary international stock markets of England, France, Japan, the United States, as well as, on the Greek market. To this purpose, it is initially attempted to measure the impact on the four primary stock markets and the Greek Stock Exchange.

The Financial Stress Index
In order for a disturbing financial incident to be regarded as an event of financial crisis, Illing and Liu (2006) have suggested the use of a Financial Stress Index (FSI), the values of which would mark the coming of a financial crisis when exceeding certain limits (Cardarelli et all, 2004).
 Three indices that relate to the Banking System: the first is the beta index of the banking stocks (the beta index of banking stocks measures the covariance between the total returns of the banking sector index and the general capital market index); the second is the spread between interbank interest rates and the performance of Treasury Bills (namely the so-called TED spread that measures the premium charged by banking institutions in addition to the T-Bill interest rates); the third is the direction coefficient of the banking sector performance curve.  Three indices that relate to the entire securities market: Firstly, the spreads of corporate bonds, secondly, the performance of the capital market and thirdly, the volatility of the capital market performance against time.  One index that involves the foreign exchange market: The volatility of foreign exchange rates against time.
The FSI index is the mean of the above seven indices. In order to calculate this mean index, each of the above seven individual indices is awarded a statistical weight, which is the reverse of the variance of each respective index; this aims at reducing the impact of the most volatile of the seven constituent indices on the resulting value of the FSI index (Longin, 2000).

Research Methodology
In order to evaluate the effect of the financial crisis on international money markets, it is attempted to evaluate the following model using dummy variables (Illing & Liu (2006), Green, 2000).
Where tj y : j is the performance of the Stock Index during month t . tj d : Is a dummy variable which is characterised by: It is therefore supposed that the "key" date for our analysis is the month August of 2009, when the crisis started manifesting itself. The conditional mean of the above econometric model is as follows: From the above two equations it becomes evident that in the econometric model (1.1), the 0 j  factor represents the mean value of the Stock Index j prior to the manifestation of the financial crisis, while the result indicates the mean value of the above index after the manifestation of the crisis. Therefore, the j  factor records the difference between the Stock Index mean value j before and after the manifestation of the crisis.
If the above factor is statistically important, it is then ascertained whether the crisis manifestation has had a positive or negative impact (depending on the positive or negative sign of the ˆj  estimate) on the Stock Index j .
Then it is examined the zero hypothesis 0 : 0 j H   , which indicates that the manifestation of the financial crisis has zero impact on the Stock Index j -against the alternative hypothesis of 1 : 0 j H   , which indicates that the manifestation of the financial crisis has some impact, either positive or negative, on the Stock Index.
The data used for this analysis have been drawn from the database of the Naftemporiki economic newspaper and involve the performance of the four main stock indices: a) the English FTSE, b) the French CAC 40, c) the Japanese Nikkei 225, and d) the USA Dow Jones Industrial Average.

The impact of the crisis on international capital markets
Based on the above econometric model, it is attempted to examine whether the recent financial crisis has had any impact on international capital markets. It is assumed that the "key" date for our analysis is the month August of 2009, when the crisis started manifesting itself. The time period before the manifestation of the crisis ends in July 2005, while the time period after the crisis ends in December 2008. Based on the above time limits, the results from the evaluation of the econometric model (Annex) are shown in the Table 1. The data presented in the Table 1 clearly show that the recent financial crisis has had a negative impact on all examined markets, with the Tokyo stock exchange being the one mostly affected.
More specifically, prior to the start of the crisis, the average monthly performance of the Nikkei 225 index was 1.47%, while after the crisis the same was decreased by 4.97 %. France followed the same pattern, as its CAC 40 average monthly performance prior to the start of the crisis was 1.1%, while after the crisis the same was reduced by 4.47 percentage points. The least affected was the London index. Diagram 1 illustrates the progress of performances across time, as well as, the FTSE closing prices starting from July 2005 until December 2008.
The Diagram 1 shows increased variability of performances following the start of the financial crisis, a fact that is indicative of the presence of conditional heteroscedasticity. Namely, the variability of the FTSE Stock Index performances changes depending on the time period examined. More specifically, it is observed that the variability is greater during the time of market decline in the period of 2007:11-2008:12 and lower during the time of rising markets. Table 2 examines certain results of descriptive statistics with regard to the previous index.
During the rising market phase, the average monthly performance of the FTSE Stock Index was 0.903% with a standard deviation (variability) of 2.28%. However, during the declining market phase, the variability of performances of the FTSE Stock Index has more than doubled reaching 5.55%. Similar events have, also, been recorded with the remaining three Stock Indices.

The impact of the crisis on the credit expansion of Greek Credit Institutions
The profitability of a financial institution is directly associated with the size (and the quality) of its lending portfolio, since a bank profits from the margin between its (average) lending and borrowing interest rates. Therefore, the ratio of its loans to its total assets is indicative of a bank's credit expansion. Since 2003, there has been significant increase in the activities of Greek banking institutions, with their loans accounting for almost half of their total activities (Diagram 5).
The question is whether the recent financial crisis has affected the credit expansion of domestic banking institutions. The answer to that question comes through the estimate of the econometric model used. The results of such estimate are presented in Table 3. Table 3 data clearly shows that the recent financial crisis has not affected in general the credit expansion towards enterprises and households; however, it has affected the credit expansion to enterprises and households on a case-to-case basis. More specifically, prior to the start of the financial crisis, the average loans to enterprises as a percentage of the total assets of domestic credit institutions were 26.96%; the same percentage after the manifestation of the crisis was reduced by 3.28%.
To the contrary, as far as households are concerned, prior to the start of the crisis the average loans to households were 22.82% of the total assets of credit institutions; the same percentage after the manifestation of the crisis was increased by 3.11%. Indeed, the crisis has neither reduced housing credit nor consumer credit. Diagram 6 shows the progress of housing and consumer credit as a percentage of the assets of domestic credit institutions.
From the econometric model used, it emerges that the average housing and consumer credit values as a percentage of the total assets of domestic credit institutions were 15.06% and 7.14%, respectively; the same percentages after the manifestation of the crisis increased by 2.19% for housing credit and 0.89% for consumer credit. As evidenced, what the crisis created was in fact a delay in the (average monthly) rate of growth of credit expansion (Table 4).
Even though lending constitutes the primary activity of banking institutions, it cannot be regarded as a reliable source of liquidity for banks. On the other hand, banking institutions, the assets of which contain a generous number of short-term securities find themselves in a more favourable state when compared with banking institutions the assets of which are dominated by loans, even if such loans have been issued in the most prudent manner (Georgiadis, 2005).
Those opposing to the above allegations assert that what is important in loan approval is not the liquidity loss it entails for the bank (usually due to long-term nature), but rather the financial ability of the borrower to repay his liabilities towards the bank in due time.

The impact of the crisis on the interest rates of Greek banks
Domestic interest rates prior to the accession of Greece to the Euro Zone had a steadily declining course. Diagram 7 illustrates the historical progress of the lending weighted average interest rate to households during the 1998-2004 period (Lekkos, 2006), in the three countries with the highest respective interest rates (Greece, France and Portugal), as well as, the EU-10 average.
As shown in Diagram 7, during the period prior to the accession of Greece to the Euro Zone, the lending interest rates of Greek banks were significantly higher than the European average (by approximately six percentage points). After that, the accession of Greece to the Euro Zone in 2001 resulted in a severe decline of lending interest rates, which by the end of 2004 were higher than the European average by approximately 1.25%. Diagram 8 shows the historical progress of consumer credit interest rates from January 2003 until January 2010.
In order to examine whether the recent financial crisis has affected domestic consumer credit interest rates, an assessment is performed using the main econometric model used based on the above scheme. The data of Table  6 show that the recent financial crisis has not affected consumer credit interest rates. Only credit card interest rates have presented a statistically significant, yet only slight increase of 97 baseline points. Therefore, the Greek banking sector continues to aim at gaining profits from the great margin that exists between lending and borrowing interest rates (Diagram 9).
As emerges from the Diagram 9, the margin between lending and borrowing interest rates, namely the interest rate margin, is higher in Greece as compared with the Euro Zone equivalent by 1.17%. According to Malliaropoulos (2006), such difference in interest rate profit margins may be attributed to Greece's higher inflation rates as compared with the countries of the Euro Zone.

Conclusions
One of the main reasons of the recent financial crisis was the increasing use of loan securitisation, a practice that has already started since the 1980s and has enabled banking institutions to significantly loosen their financing criteria, due to the fact they can transfer the credit risks entailed in lending via securitisation (Goodhart, 2006). Based on this process, financial institutions sell claims to customers (i.e. loans), namely to Special Purpose Vehicles, companies which issue bonds commonly known as asset backed securities (ABS), which they later sell to investors (Driffill et all, 2006).
However, the practice of securitisation has gained great significance during the latest years, resulting in the issue of CDO bonds, (collaterised debt obligations), which are based on ABS bonds, namely the initial bonds issued as part of the securitisation practice. However, CDOs, also known as toxic bonds, were not usually the subject of negotiations in secondary markets and that is why prices not always existed for these. Moreover, the assessment of the credit capacity of toxic bond issuers (namely the CDOs) was not always accurate. The buyers of ABS and CDO bonds received their interest coupon on a periodical basis, the payment of which depended on cash flows produced by securitised loans (Tully, 2007a,b).
When borrowers of securitised housing loans in the United States became unable to repay their obligations, CDO buyers found themselves in a position where they held bonds with extremely low rating. The owners of such bonds included certain banks (as well as subsidiaries of bank's that were awarded guarantees for buying securities), which became subsequently exposed to serious problems both in terms of solvency (as the losses from such bonds needed to be recorded) and liquidity (owed to their subsidiaries). This has created conditions of uncertainty in the market, as it was not known from the beginning (summer of 2007) which banking institutions had problems or the extent of such problems. As a result, interest rates of the interbank market started in turn to rise (Baur & Niels, 2008).
This survey has shown that the financial crisis has had no effect on the overall credit expansion of Greek banking institutions towards enterprises and households. Yet, on an individual basis, the crisis has resulted in a decline in bank loans issued to enterprises as a percentage of the total assets of domestic credit institutions by 328 baseline units. With regard to the increase in household lending, it was found that the average housing credit as a percentage of the total assets of domestic credit institutions increased from 15.06% by 2.19%, while consumer credit increased to a lesser extent from 7.14% by 89 baseline points.
As a result, the financial crisis has had an indirect impact on the Greek economy by increasing the cost of money due to the reluctance of banks to lend one another. This cost also affects the cost of lending both for individual and corporate customers. Source: Bank of Greece, following data processing