The impact of the 2008 crisis on the volatility of industries within Europe Master thesis International Business and Management

1

The impact of the 2008 crisis on the volatility of

industries within Europe

Master thesis International Business and

Management

Author:

H. van der Weide (s1929089)

Supervisor:

Dr. K. van Veen

2

3

Index

1 Introduction . . . 5

2 Financial crisis . . . 7

. . . 8

. . . 8

. . . .

3 Europe and the crisis . . . 10

3.1 European banking sector

. . . 10

4 How volatility influences different industries . . . 12

4.1 Spillover effects

. . . 12

4.2 Volatility and country factors

. . . 13

4.3 Volatility during crises

. . . 14

4.4 Portfolio management implications

. . . 14

5

Hypotheses . . . 16

6

Data and methodology . . . 18

4

Appendix 1: Bivariate correlations . . . 34

Appendix II: Auto correlations . . . 38

Appendix III: Volatility plots per industry . . . 45

Appendix IV: Descriptive statistics per industry . . . 52

Appendix V: Data and plot explaining variable, Market Value . . . 63

5

1

Introduction

Currently the western world and thus also Europe are in a financial crisis. This crisis started due to a stagnating housing market in the USA, and related to that the overvaluation of houses. This housing crisis also called mortgage crisis reflected on the financial institutions. Due to the fact that banks offered excessive mortgages that could not be repaid and underlying bonds of the mortgages were rapidly decreasing in value, banks faced a liquidity problem in the 2007-2008 period. Because it was not clear which financial institutions would face the biggest problems, banks refused to lend each other money. Banks also made it more difficult for other borrowers to get a loan which made it more difficult for businesses to keep their firms running. This liquidity decline leads to a reduction in business activity, reduced employment and deflating stock prices. This finally resulted in the financial crisis we are currently in. The equity crisis were this paper focuses on started basically in September 2008 when investment bank Lehmann Brothers filed bankruptcy and AIG was bailed out. In this third quarter of 2008 the United States entered a recession the most intense one since the 1974-1975 crisis, this current crisis involved the entire developed world (Grigor’ev and Salikhov, 2009).

The 2008 crisis can basically be described in three steps. The first stage is the period between July 2007 and August 2008. In this period the American mortgage crisis was developed, due to enormous write-downs by banks on mortgages and the first bankruptcies. During this period the international financial system incurred already reported losses of approximately $800 Billion. The second stage, which is also named the liquidity crisis, became visible in September 2008. In this period banks traded only with someone else’s money which they got from the central banks and did not trusted other banks enough to lend them money. An important indicator for this trust issue was the bankruptcy of investment bank Lehman Brothers. This bankruptcy showed citizens and businesses that any big financial institution could face liquidity problems. The third stage is the credit crisis resulting from the liquidity crisis which was visible in stage two. This crisis thus basically evolved from the distrust between banks and other financial institutions. And the longer this distrust continues the more harm there will be done to current economic activity, but especially to future capital investments (Grigor’ev and Salikhov, 2009).

6 lot of volatility. During the 2008 crisis there was a spike of volatility in many countries and volatility was high market wide.

Thus it is clear that financial service firms experienced a lot of volatility during the recent crisis but how does their volatility relate to other industries? And besides that, there are also industries that despite a crisis experience low volatility. Which industries are the industries that experience low volatility and how is it possible that they are not affected in that sense by the crisis? These questions lead to the following research question:

Which European industries show high degrees of volatility and which industries show low degrees of volatility during the recent, 2008-2010, financial crisis and how can these differing volatilities be explained?

Answering this question will give managers of companies an insight in how risk full the industry is where they are located in during a credit crisis. From a risk management perspective this can give vital information about the steps a manager has to take when a crisis arises in the future. But also for investors it is interesting to see which stocks are less volatile during a crisis. Because less volatility means less risks for investors. Stocks that experience low volatility are therefore interesting to keep in a stock portfolio, especially for differentiation purposes. The difficulty with volatilities however is that they do not move constant over time and that the behavior of volatilities varies a lot between industries (Wang, 2010). Therefore it is for investors important to know why certain industries experience less volatility in certain periods and if there are correlations between different periods of volatility. The recent 2008 crisis adds a different chapter to the already existing literature about industry volatilities, Soros (2008) indicates that this recent crisis asks for a new paradigm to understand what is going on. The current crisis marks the end of a era of credit expansion that lasted for twenty five years. The paradigm that financial markets tend towards equilibrium has to be revisited which also asks for a different look at volatility movements (Soros, 2008).

7

2

Financial crisis

In this chapter is explained how the financial crisis started and which indicators are important to explain why the financial crisis started. Because the latest crisis is not the only crisis the world has known it is also interesting to see if there are parallels with other financial crises. A lot of comparisons are already made with the great depression of the 1930’s. This chapter will paint a picture of how this recent crisis influences economic factors such as volatility.

The most recent financial crisis is part of a prolonged pattern that can be traced back to crises in 1857,1893, 1907 and 1929-1933 (Bordo, 2008). All these crises were started by triggering events in the U.S. financial system. This time the trigger was the bursting of the housing bubble in the U.S. which caused the values of securities attached to real estate to depreciate enormously. Not only financial institutions within the U.S. where hit by these depreciations but on a global scale a lot of financial institutions encountered the same depreciations. But there are also dissimilarities, Bartram and Bodnar (2009) state that this current crisis is different from earlier crises because this crisis is both severe and global. Therefore this crisis challenged investors to rethink their perceptions considering equity investments. Mainly because the crisis has driven down the stock prices of nearly all companies in every sector or industry in the world. Falling stock markets is also one of the three indicators of this financial crisis according to Chari et.al. (2008). The other two indicators are that several major financial institutions have failed and that the spreads on a variety of different types of loans over similar U.S. treasury securities have widened exponentially.

8 2.1 Mortgage crisis

To elaborate further on the three steps of the genesis of the crisis, first the mortgage crisis is described. The mortgage crisis basically begun in the year 2000 when mortgage interest rates were low and home prices increased drastically. For this reason home buyers bought high-priced houses with nontraditional mortgage products. The buyers expected that there houses would stay increasing in value but nevertheless house sales stagnated and the prices of houses decreased. A lot of people took a so called ‘subprime mortgage’ which is basically a mortgage for borrowers that have poor credit histories. A subprime mortgage is not a conventional mortgage and is offered to people that, in the view of the lender, bear more than average risks. For this reason subprime mortgage have a higher interest rate than conventional mortgages in order to compensate the lender for carrying more risk. Due to these subprime mortgages more people were able to buy houses, which increased the demand for houses in the 2003-2006 period. As a result the demand for houses dropped for the upcoming years after this period.

2.2 Liquidity crisis

A liquidity crisis is a crisis that involves a shortage of cash and/or other immediately available funds. Through this shortage repayments cannot be made and therefore banks, financial markets and businesses can and will be affected.

9 2.3 Credit crisis

The credit crisis is the last stage of the current crisis and follows after the mortgage and liquidity crisis. In this specific case the credit crisis is a direct result of the aforementioned crises. A credit crisis can be described as a crisis in which the availability of loans or credits plummets, and is often mentioned by the synonym credit crunch.

10

3

Europe and the crisis

The 2008 credit crisis started as mentioned before in the U.S. initiated by the mortgage crisis. But due to the fact that markets are becoming more and more integrated it could not last long before the crisis also started to feel in Europe. This chapter therefore describes how Europe was prepared for the crisis, which events took place and how Europe reacted on the crisis.

The European Union (EU) was overall not very well prepared to manage a crisis that also involved systematic cross-border institutions (Pisani-Ferry and Sapir, 2010). In the EU the main problems that were expected to arise concerned liquidity and solvency risks. On the liquidity side there was an absence of clear guidelines for lender-of-last-resort functions in situations when EU banks would experience serious liquidity problems (Nieto and Schinasi, 2007). The solvency side also lacked clear arrangements for, for instance fiscal burden-sharing mechanisms.

3.1 European banking sector

In the pre EU era European banks had to cope with the specific regulations of their home country. Because of tax and regulatory differences between countries it was hard for European banks to employ cross border activities. But due to the EU, which was founded in 1958, it became easier for banks to start and expand cross border activities. One of the most important institutes of the EU for banks is the European Central Bank (ECB), which is the European equivalent of the U.S. Federal Reserve.

The EU banking sector changed a lot since the liberalization of capital movements in the early 1990’s. Because after this breakthrough the euro was introduced in 1999 and new EU member states were added in 2004 and 2007. The creation of a bigger single market supplemented with one single currency inspired banks to internationalize more frequently. This more liberal way of banking made sure that Mergers and Acquisitions (M&As) occurred more often. Thus prior to the crisis the European banking sector was becoming more integrated (Pisani-Ferry and Sapir, 2010).

11 Europe tried to solve this oversight problem by highlighting the importance of the principles of decentralization, segmentation and cooperation (Pisani-Ferry and Sapir, 2010 and Nieto and Schinasi, 2007).

The first pillar decentralization is on country level, build upon prudential supervisors, central banks and treasuries. This is also the level where accountability is controlled for. Overarching institutions as the ECB and the National Central Banks (NCBs) have financial stability related responsibilities. They for instance control and contribute to national policies concerning financial stability and supervision. An important function as the lender of last resort function however is a national responsibility. The term lender of resort function refers to the possibility of an institution to extend credit to banks or other companies when no one else will, this to avoid bankruptcy. This function is only executed when the bank or other company is to systematically important for the country to let it go bankrupt. The second pillar segmentation indicates that financial stability functions are located in different sectors. And because the EU does not have created a common regulatory and supervisory framework there is a quality difference in financial supervision. This is also the reason why for instance deposit insurance schemes developed in different ways in different EU member states. The third pillar cooperation is developed in different areas. EU legislation was developed to keep EU member states on the same page concerning law. Besides that a Committee of European Banking Supervisors (CEBS) was established to advise the European Commission about regulation and supervisory practices in the banking industry. Also for potential crisis times, provisions were made for cooperation in those times of turmoil.

12

4

How volatility influences different industries

During a crisis most industries experience declining stock rates and increased stock volatility. Per industry it differs how much influence a crisis has on these factors. Certain industries will take big hits during a crisis while other industries will not experience that much negative turmoil. And it is quite possible that there are also industries that even perform better during a crisis. Therefore it is interesting to compare previous literature about which industries will probably experience the biggest hits and which industries show positive trends during a crisis. Besides that Cavaglia et al (2000) indicate that industry factors are becoming more important than country factors. One of their most important indicators is that diversification through global industries provides more risk reduction than diversification across countries. Therefore it is remarkable that there has been surprisingly little empirical research about volatility on the firm or industry level (Campbell et al, 2001).

But there is an ongoing debate about the explanatory powers of country factors compared to industry factors in relation to volatility (Cavaglia et all, 2000; Brooks and Catao, 2000; Campbell et al, 2001; Catao and Timmerman, 2003; Ferreira and Gama, 2005; Wang, 2010). In this debate authors explain why certain industries are more vulnerable for volatility than other industries. Smaller global industries for instance, that have not got a lot of variation in countries where they are active or that are concentrated in a single country are more risky and thus more susceptible for volatility. Wang (2010) for example finds that most industry volatilities are lower than the market volatility, with exception of the smaller industries tobacco, mines and coal. This indicates that those three smaller industries experience more volatility shocks. The lower volatility of larger industries can mainly be explained by the fact that large industries have established themselves all over the world while smaller industries can still be centered in certain parts of the world. Ferreira and Gama (2005) then find that the highest industry specific variance is in the mining industry followed by information technology, tobacco and the water industry. When specifically examining trends in volatility then can be found that most industries experience a linear upward industry risks trend. This upward trend is most visible for the oil, autos, steel and mines industries. Also for most of the global industries this trend is visible, however this trend is not significant. Furthermore Moore (2011) concludes that the number of stocks in a portfolio, to achieve a certain level of diversification, must be increased when the market is in a negative spiral.

4.1 Spillover effects

13 Wang (2010) found that the financial service industry and business supplies industries (paper) are the main indicators of industry volatility, Granger-causality tests prove that those industries help forecasting most other industries. This suggests that volatility shocks in those industries have the biggest effects on other industries. The business supplies industry helps forecasting 19 of the 29 industries and the finance industry helps forecasting 12 industries. This is tested with the granger causality test (a test to determine if one times series can be used to forecast another times series) at a 5% significance level. On the other side of the spectrum the automobile industry does not lead any industry and consumer goods, transportation, wholesale and tobacco products only help forecasting one other industry. It is also remarkable that oil, which is the second largest industry in the sample of Wang (2010), only predicts three other industries. Because in general industry shocks in large industries do influence other large industries and small industries, while industry shocks in small industries only influence other small industries and no large industries. The observed difference between small and big industries is determined from the contemporaneous causal relationships analysis from Wang (2010). Exceptions upon the differences between small and large industries are traditionally large industries like oil and autos which do not appear to have a significant influence on other large industries. The leading positions of the business supplies industry and the finance industry was only overtaken in the late 1990’s by the business equipment and services industry, most probably due to the fast growing information technology sector where both industries have linkages with. The fact that the oil industry has so little influence can be explained by the fact that the growth within the U.S. is more dependent on services and high tech industries. The influence of manufacturing on the economy is decreasing which is also reflected by the decreased capitalization weight of the oil sector. Moore (2011) also finds that the finance sector, especially in Anglo-Saxon stock markets, has a position as a market maker and thus as an indicator for volatility movements within other industries.

4.2 Volatility and country factors

14 In general correlations are higher among Anglo-Saxon countries (Canada, U.S. and U.K.) than between other countries. Correlations are lowest between the U.S. and much of continental Europe and Japan, with as an exception the Netherlands. Volatility within the Netherlands is highly correlated with the U.S. and the U.K. this can mainly be explained by one very large company, Shell. Because the Dutch country portfolio is pretty thin, one company can make a huge impact on the country volatility. As stated earlier there is an ongoing debate about whether portfolio diversification by countries or by industries is more efficient. In general can be said that in the 1990’s country diversification was considered as the way to go but more recently researchers consider industry diversification as important as country diversification (Ferreira and Gama, 2005).

4.3 Volatility during crises

Because the recent crisis is also called the credit crisis and therefore banks were the epicenter of the crisis it is expected that financials will take a bigger hit than non-financials. Bartram and Bodnar (2009) accordingly find that the volatility within the financial sector is nearly 50% higher than the financial sector. They also find that there can be made a comparison between financials and non-financials but also between countries. In the country analysis can be found that U.S. and developed markets financials have quite the same behavior, they perform much less than indices in for example emerging markets. Financials in emerging market most likely perform much better because they have lower exposure to mortgage securities. Mahran (2011) also found that the financial sector took one of the biggest hits in their analysis of the Egyptian stock market during the crisis. Rehman and Akbar (2011) specifically took a look on the performance and investment policies of UK public firms during the crisis. They differentiate inter alia between manufacturing firms and firms in the service sector. What can be found in this comparison is that the crisis has a bigger impact on manufacturing firms than on firms in the service sector. This can probably be explained by the fact that manufacturing firms depend more on debt, and when it becomes harder to attract credit these firms are the first to be hit. Wang (2010) also finds that all industry volatilities are significantly higher during a crisis period, except the mining industry.

4.4 Portfolio management implications

16

5

Hypotheses

The aforementioned literature leads to different assumptions concerning the performances of stocks during the crisis. During crises stocks experience more volatility, but there are industries that are expected to experience more volatility than other industries and there are industries that are expected to experience less volatility than other industries. Wang (2010) finds the following: ‘Most industry

volatilities are lower than the market’s with the exception of three small industries (tobacco products, mines, and coal), a result similar to the finding of Ferreira and Gama (2005) on global industries. Large industries tend to have low volatilities (e.g., finance and utilities) with the exception of oil’. In

general smaller industries are not present in a lot of different countries. Bigger industries are present in more countries which implies that they can level out the country risks. When for instance a smaller industry experiences a lot of turmoil in of their countries they will not be able to level this risk out within other countries. Thus considering Wang’s (2010) and Ferreira and Gama’s (2005) research smaller industries tend to experience more volatility than large industries. This leads to the following hypothesis:

H1: Because the mining and quarrying industry is the smallest industry it will experience one of the highest rates of volatility.

The financial sector experienced a lot of uncertainty during the crisis which started with the bankruptcy of Lehmann Brothers and Icesave. Furthermore AIG was nationalized which also caused a lot of uncertainty in the financial sector. Uncertainty in industries usually leads to high levels of volatility, thus the financial sector is also expected to experience a lot of volatility. Mahran (2011) found the following: ‘First, sectors showed a significant difference in both price and liquidity before

and after the crisis. These sectors are most affected by the global crisis, such as building and construction sector, cement, spinning and weaving sector financial sector and service sector’. Bartram

and Bodnar (2009) found that the volatility within the financial sector is almost 50% higher than the volatility within the non-financial sector. This leads to the following hypothesis:

H2: Due to the high uncertainty levels in the financial industry (financial and insurance industry), this industry will experience the highest rates of volatility.

Besides industries that experience more volatility during crises there are also industries that are expected to experience less volatility during crises. Mahran (2011) states the following about those industries: ‘While some sectors showed a significant difference in the average price but no significant

17

group of sectors, which showed a significant difference in both price and liquidity’. Besides that it is

widely known that industries that deliver consumer goods which are indispensable, such as food and utilities, are less affected by a crisis. Consumers rather economize on luxury goods such as jewelry and holidays instead of economizing on their daily needed products. This leads to the following hypothesis:

H3: Due to the indispensability of consumer goods (Accommodation and food service activities), this industry will experience one of the lowest rates of volatility.

The volatility of industries is influenced by different factors like the performance of the companies in the specific industry, the state of the world economy but also by the volatility of other industries. This so called spillover effect indicates which industries influence other industries concerning their volatilities. Wang (2010) states the following: ‘Interestingly, the business supplies industry (Paper) is

the most important lead indicator of industry volatilities as it helps forecast most other industry volatilities (19, or about two thirds of the other 29 industries). Finance is the second most important volatility indicator. It Granger-causes 12 other industries (including many large sectors) and the market’. Wang (2010) used the granger causality test to compute how much industries can be

forecasted by another industry. He build a model where he performed a regression of the actual time t on three lags (t-1, t-2 and t-3) at a 5% significance level. Those lags are day based, thus the spillover effect is measured at a short term because over a longer period the spillover effect is mostly not visible anymore. Moore (2011) indicates that in Anglo Saxon markets the financial services industry is the main market maker. The main market maker can be considered as the industry that indicates in which economic direction the other industries will move, for instance towards high volatility levels or low volatility levels. Considering that European countries are mainly Anglo Saxon and considering that the banking sector was the initiator of the crisis it is expectable that the financial services industry is the lead indicator of volatility in Europe, which leads to the following hypothesis

H4: Because the financial service industry (financial and insurance industry) was one of the initiators of the crisis this sector will help forecast most other industry volatilities.

18

6

Data and methodology

In this section the research design is provided which explains how the data is gathered and how the research is methodologically performed. In the data part is explained how the data is gathered and why certain choices have been made concerning data delimitation. In the methodology part is explained how the data is processed and which tests have been performed.

6.1 Data

The main source for data in this thesis is Datastream. This is a database which consists of a wide variety of economic and business related data. Via Datastream the price indexes (the stock returns over a given point of time) of all the industries are downloaded. Those price indexes are used to compute the rate of returns of the companies. The rate of return, also called the total return, is a method to compute the efficiency of a portfolio. This is done by subtracting the old stock price value from the new stock price value and then dividing it by the old stock price value (New – Old / Old). The data that are extracted from Datastream are the daily closing stock prices.

Campbell et al (2001) and Wang (2010) specifically researched industry volatility without directly comparing it with country factors. So their researches more or less relate to the research that has been constructed and conducted in this thesis. But those researches focus on the U.S. to explore the importance of industry factors. This research uses Europe as a starting point to measure the influence of industry factors. Brooks and Catao (2000) and Wang (2010) already researched volatility movements in the U.S. and Cavaglia et al (2000) and Catao and Timmerman (2003) took a more worldwide portfolio to research volatility movements. Specific research concerning the European market has not been conducted before and therefore a focus on the five biggest European stock exchanges will add a new chapter to the already existing literature. Per country there is an index which include all the main companies that are listed in that specific country, all companies within one of these indexes are used for the analysis. The five biggest exchange indexes of Europe include:

- CAC40 (France) - DAX (Germany)

- FTSE 100 (United Kingdom) - S&P/MIB (Italy)

- IBEX-35 (Spain)

19 an industry comparison. Therefore there has been chosen for an analysis of the biggest indexes of Europe instead of a single country analysis.

The data are gathered over the total period of 1 January 2006 – 31 December 2010. January the first of 2006 till 31 July 2008 will be considered as pre-crisis period. Grigor’ev and Salikhov (2009) describe that the credit crisis basically started in September 2008, therefore the crisis period will be the period of 1 September 2008 till 31 December 2010. When the data is gathered there first will be a check if there is a significant change in volatility when the pre-crisis period ends and the crisis period starts. If the volatility change in Europe is in another period there can be considered to change the pre-crisis and crisis dates. The data that will be gathered are stock prices per day, from every trading day in the aforementioned period. The reason to choose for daily data is because daily data give a far more representative resemblance of the volatility of stocks than for instance monthly data. When for instance choosing for monthly data a lot of peaks in the volatility will be leveled out due to the larger time frame. But also the larger data set is an important reason for choosing daily stock prices. Companies that have not available data in the five year timeframe used for the research are deleted from the sample.

The NACE (Nomenclature des Activités Économiques dans la Communauté Européenne) industry classification is used to subordinate the companies into different industries. NACE is a European classification system which is similar in function to SIC (Standard Industry Classification) and is used to compare economic activities of companies with each other.

Enumeration of the NACE industries: - Agriculture, forestry and fishing (A) - Mining and quarrying (B)

- Manufacturing (C)

- Electricity, gas, steam and air conditioning supply (D)

- Water supply; sewerage; waste management and remediation activities (E) - Construction (F)

- Wholesale and retail trade; repair of motor vehicles and motorcycles (G) - Transporting and storage (H)

- Accommodation and food service activities (I) - Information and communication (J)

- Financial and insurance activities (K) - Real estate activities (L)

20 - Public administration and defense; compulsory social security (O)

- Education (P)

- Human health and social work activities (Q) - Arts, entertainment and recreation (R) - Other services activities (S)

- Activities of households as employers; undifferentiated goods - and services - producing activities of households for own use (T)

- Activities of extraterritorial organizations and bodies (U)

6.2 Methodology

The methodology of Campbell et al (2001) is used to develop a framework to measure specific components of volatility. In Campbell’s framework volatility is divided into three components (market wide return, industry specific residual and firm specific residual) but for this specific research two components are adequate, the firm specific residual can be ignored because the analysis is not at company level. Therefore in this analysis the return of a stock can be decomposed into two components: the market wide return and an industry specific residual. The perspective of this methodology is from the point of a global investor whose returns are in Euro’s as four of the five countries use the Euro as their currency. Only the U.K. does not use the Euro as their currency, hence their Pound is converted into Euro’s.

Industries are denoted with an i, individual firms with an j. Rjit then represents the excess return of firm

j that belong to industry i in period t. And Wjit represents the weight of firm j in industry i in period t.

To compute the excess return (the rate of return that exceeds the return from a benchmark or a portfolio with a similar risk, thus the value added), first the return has to be computed by the following equation:

(1)

In this equation the return on stocks of firm j in industry i at time t is the stock price of that specific firm at that time, minus the stock price at time t-1 divided by the stock price at time t-1.

The excess return of i in period t is then given by:

(2)

This equation computes the return on stocks of industry i at time t. This return is a weighted average of the returns on firm level. When there are for example three companies (A,B,C) and three returns (Ra,

21 And the weight of industry i in the total market is denoted as Wit, following Campbell’s methodology

the market return is computed by using the same weights for the different industries. This results in the following excess market return:

(3)

This equation has the same principle as the second (2) equation the only difference is that with this equation the weighted average over all the industries returns is computed and not one industry.

According to the capital asset pricing model (CAPM) intercepts can be set to zero in the following equation:

(4)

In this equation the β of each industry is computed. This is done by performing a regression in SPSS in this regression Rit is the dependent variable and Rmt the independent variable.

With the following equation the volatility of the market return (MKTt) can be computed. The µm refers

to the mean of the market return over the whole sample. The interval at which returns are measured is indexed as s.

² (5) With this equation the volatility of the market in a specific interval, in this case a year, is computed. The volatility of the market in interval s equals the sum from the deviation between the daily market return and the average market return. The square is to not let the positive and negative returns level each other out. Simplified the last part of the equation is like this: (Daily market return – Average market return)2.

To measure the volatility per industry i the squares of the industry specific residual within period t have to be summed, this is what the next equation does:

(6)

22 To statistically test which industries experience the highest degrees of volatilities and whether volatilities in certain industries influence volatilities in other industries different tests are used.

First a Student’s T-test is used to test whether there are significant differences between the mean of the pre-crisis period and the actual crisis period. The Granger Causality test is used for determining whether a specific time series can forecast another time series. Thus can certain industries forecast the volatility movements of other industries. Besides that, also an autocorrelation test is performed to determine whether there are repeating patterns in the movement of the volatilities. With this test can be measured to what extend the volatility from a previous month influences the volatility of the next month. But also to what extend volatilities in 2006 can explain volatilities in 2010.

7

Results

This section shows and describes which results follow from the tests as described in the methodology section. For the clarity of this section most tables can be found in the appendices, only the most important and most explaining tables and charts are used in this section. In this section is described if the hypotheses are supported and how the different results can be explained.

To compute the volatilities of the different industries, first the betas of the industries must be computed. Berk et al (2008) define beta as follows ‘beta is the expected percent change in the excess

return for a security for a 1% change in the excess return of the market (or other benchmark) portfolio’. The beta of an industry already explains a lot about how the industry moves in accordance

23

Industry Beta

Mining and quarrying (B) 0,522

Manufacturing (C) 1,046

Electricity, gas, steam and air conditioning supply (D) 0,810 Water supply; sewerage; waste management and remediation activities (E) 0,903

Construction (F) 1,267

Wholesale and retail trade; repair of motor vehicles and motorcycles (G) 1,019

Transporting and storage (H) 0,945

Accommodation and food service activities (I) 1,107

Information and communication (J) 0,954

Financial and insurance activities (K) 1,365

Real estate activities (L) 1,184

Professional, scientific and technical activities (M) 1,219

Administrative and support service activities (N) 0,831

Arts, entertainment and recreation (R) 0,827

Figure 1: Estimated beta coefficients of the capital asset pricing model (equation 4) per industry.

Figure 2 and 3 show how much volatility the different industries have experienced in the recent years, where industry B till H is displayed in figure 2 and industry I till M in figure 3. Most notable are the high volatilities of industry B (Mining and quarrying) and industry L (real estate activities) especially during the 2008 crisis period. Also remarkable is the proportionally low volatility of industry K (Financial and insurance activities). The crisis started in the banking sector and the financial and insurance industry has overlaps with the banking sector thus a higher volatility rate was expected. On the other hand the real estate activities industry, which also has overlaps with the banking sector, does have high rates of volatility as expected. The high volatility of the mining and quarrying industry is accompanied with a relatively low beta for this same industry of 0,522. This basically means that although the mining and quarrying industry is not closely related to the market this industry still experiences the highest peak of volatility during the 2008 crisis period. The volatility peak of the real estate industry arises a little later than the peak from the mining and quarrying industry. The peak of the real estate industry is in the 2009 period and thus follows after the peak of the mining and quarrying industry. In contrast to the mining and quarrying industry the real estate industry has a rather high beta of 1,184. This means that in this case the beta does not explain everything about the volatility movements of the industries over time. Because in general, lower betas mean less volatility peaks and high betas more volatility peaks. In the table of appendix V the most notable peaks are shaded yellow , to give an overview of the heights and the timing of the peaks.

24

Figure 2: Volatilities per industry per year (part 1), volatilities are computed by squaring the residuals from the capital asset pricing model (equation 4) and taking the sum over a year.

As can be read from figure 2, hypothesis one is supported because the mining and quarrying industry has the highest volatility level. Also the industries displayed in figure three do not level the volatility peaks of the mining and quarrying industry during the crisis period.

Figure 3: Volatilities per industry per year (part 2), volatilities are computed by squaring the residuals from the capital asset pricing model (equation 4) and taking the sum over a year.

Figure 3 shows mainly that the finance and insurance activities industry does not experience that high rates of volatility as have been expected. This means that hypothesis two is rejected, because this industry does not have one of the highest rates of volatility. Most notable in this figure is that the finance and insurance industry does not experience high levels of volatility while the real estate activities industry does experience a high peak of volatility. The crisis started in the real estate sector and the finance and insurance sector which makes it remarkable that their volatilities are quite

25 different from each other. The difference of this research in comparison with for example Wang’s (2010) research, wherein the finance industry does have high volatility levels, is that Wang used an industry classification wherein finance was a single industry. In this research finance is coupled with insurances which could explain the aberrant outcome in comparison with other researches. In figure 3 is also shown that industry I (Accommodation and food service activities) does not experience the low rates of volatility that have been expected. Industry C (Manufacturing) has the lowest rates of volatility both in the pre-crisis period and the crisis period. This means that hypothesis three is also rejected.

Figure 4 shows how the different industries can forecast each other. The horizontal columns show by how many industries this specific industry can be forecasted and the vertical row indicates how much industries this specific industry forecasts. This data is gathered by performing a regression per industry on all other industries at a 5% significance level. When industries are significantly related this means that they have forecasting abilities towards each other. As an addition appendix I shows the bivariate correlations between the industries. Those bivariate correlations are the correlations that an industry has with the other industries over the whole sample period. This gives important information about which industries can be considered good crisis forecasters. In figure 4 can be found that Industry C (Manufacturing) and industry D (Electricity, gas, steam and air conditioning supply) can forecast most other industries and also can be forecasted by most other industries. Another notable result is that the mining and quarrying industry only has a relation with the real estate activities industries and besides that cannot predict nor cannot be predicted by any other industries.

X B C D E F G H I J K L M N R Total B X I 1 C X I I I I I I I I I I I 11 D I X I I I I I I I I I 10 E I I X I 3 F I I X I I I I 6 G I I X I I I 5 H I I X I I I 5 I I I I I X I I 6 J I I I I X I I I 7 K I I I I I X I I 7 L I I I I I I X I 7 M I I I I I X 5 N I I I I I X 5 R I I I X 3 Total 1 11 10 3 5 7 5 6 7 7 7 5 5 2 X

26 In figure 5 is shown which movements the industries make per month. Figure 4 already shows which industries have correlations with each other, figure 5 gives an overview in which directions the industries move per month, thus do they have a negative or positive return per month. A positive monthly return of an industry is indicated with the value 1 and a negative return with -1. Since there are 14 industries the maximum length of a bar can be 14 and the minimum -14. If the bar has a length of 14 this means that all the industries have a positive return in this month (example: January 2006). In February 2006 it can be seen that 13 industries experience positive returns and 1 industry a negative return. In May 2012, 8 industries show positive returns and 7 industries negative returns.

Figure 5: Industry return movements per month, where a positive monthly return of an industry is indicated with the value 1 and a negative return with -1. Since there are 14 industries the maximum length of a bar can be 14 and the minimum -14. If the bar has a length of 14 this means that all the industries have a positive return in this month.

Most striking is that most industries experience the same kind of return movements per month. This implies that the spillover between industries happens in an earlier phase. Appendix VI shows a linear regression between the dependent variable which are the returns of an industry at time t, and the independent variables which are the returns of all industries at time t-1. Where figure 5 does not indicate that there is a lead industry, appendix VI shows that industry J (Information and communication) forecasts 8 of 14 industries and is thus the main indicator of industry volatility. Industries B (Mining and quarrying), C (Manufacturing), G (Wholesale and retail trade; repair of

27 motor vehicles and motorcycles) and L (Real estate activities) forecast 3 of 14 industries and therefore also forecast the industry volatilities for a bit. Thus on a month basis a lead industry cannot be identified but when looking at day base than can be concluded that the information and communication industry leads the most industries.

The hypothesized high forecasting abilities of the financial and insurances industry do not reflect in figure 4 and appendix I. Figure 4 reports that the manufacturing industry and the electricity, gas, steam and air conditioning supply industry do not really outrun each other. The manufacturing industry only predicts one industry more than the electricity, gas, steam and air conditioning supply industry. Appendix I shows that the manufacturing industry not only has the most forecasting abilities but also has the highest correlation with the other industries. Appendix V also shows that when performing a regression with a one day lag the financial and insurances industry has not got the highest forecasting abilities. Figure 6 gives an overview of the averages of the bivariate correlations as displayed in appendix I. In appendix I the correlations between the industries over the whole period are displayed. Because all the correlations are positive it is possible to take the average correlation that an industry has with the other industries, and this is what figure 6 reflects.

The closer the number is to one the more correlated the industry is with the other industries. This figure shows that the manufacturing industry and the information and communication industry have the highest correlations with the other industries and not the financial and insurances activities industry. Therefore hypothesis 4 has to be rejected.

Industry Average bivariate

correlations

Mining and quarrying (B) 0,266

Manufacturing (C) 0,754

Electricity, gas, steam and air conditioning supply (D) 0,631 Water supply; sewerage; waste management and remediation activities (E) 0,561

Construction (F) 0,730

Wholesale and retail trade; repair of motor vehicles and motorcycles (G) 0,670

Transporting and storage (H) 0,709

Accommodation and food service activities (I) 0,652

Information and communication (J) 0,728

Financial and insurance activities (K) 0,704

Real estate activities (L) 0,548

Professional, scientific and technical activities (M) 0,676

Administrative and support service activities (N) 0,573

Arts, entertainment and recreation (R) 0,393

29

8

Discussion

Volatilities of stock markets are hard to predict because several factors (size of the industry, state of the economy) can influence the volatility of these stocks. The external shock of a crisis is one of the biggest shocks a stock exchange market can get. But while it is quite easy to predict that the volatility of stocks will rise during a crisis it is hard to predict which industries will take the hardest hits. Several articles are written about how and why certain industries will face the biggest hits but it is still hard to predict which industries will take the biggest hits. In general stock movements are hard to explain, because stock movements for a large part cannot be explained by previous movements. This research shows that stocks cannot forecast each other on month basis because this period is too long and all movements are leveled out. Only when a lag of one day is taken the results show that industry J (Information and communication) forecasts 8 of 14 industries and is thus the industry with the highest forecasting abilities. Thus can be concluded that forecasting abilities of industries can only be controlled for in a short period of time (days) because longer periods do not show significant results.

The most noteworthy results from the research are the low beta, closest one to zero, of the mining and quarrying industry and with that the high degree of volatility of this same industry. Besides the mining and quarrying industry also the real estate activities industry has high degrees of volatility, mainly during the crisis period. On the other hand the manufacturing industry experienced low degrees of volatility both in the pre-crisis period and the crisis period. The high degrees of volatility of the mining industry fits with the literature because the mining and quarrying industry is a small industry. Also the high volatilities within the real estate sector are expected because the crisis started in the financial sector. On the other hand the financial and insurance activities did not experience proportionally high volatilities. High volatilities were expected in the financial and insurances sector because the financial sector is the sector were the crisis started. An possible explanation for the relatively low volatility in the financial and insurances sector is that this sector is broader than only the banking sector. As can be seen in appendix IV the financial and insurances companies industry involves a lot of companies, perhaps the not banking companies did not experience that much volatility which could have leveled the volatility in the banking sector out.

30 used and that a single company point of view would change the results. Another possibility can be to control for the type of products that an industry offers, for instance differentiate between service and manufacturing firms.

31

9

Conclusions

This paper describes how the 2008 financial crisis influenced the stock volatility of companies from different industries. All these companies are listed on one of the five biggest exchange indexes of Europe. Different aspects are analyzed such as the volatility in the pre-crisis period and in the crisis period but also the correlations between the different industries. There is also examined if the already existing literature matches with the results found in the analysis of the data.

The main purpose of this conclusion is to answer the research question, which reads as follows: Which European industries show high degrees of volatility and which industries show low degrees of volatility during the recent, 2008-2010, financial crisis and how can these differing volatilities be explained?

The different industry volatilities evidently rose during the 2008 crisis period, although industries differed a lot from each other concerning the degrees of volatilities they experienced. The highest peaks of volatilities during the crisis were felt by the mining and quarrying industry and the real estate industry. The high degree of volatility that the real estate industry experienced can be explained by the fact that the housing market was heavily hit during the crisis. A lot of people were not able anymore to repay their mortgages and housing prices declined. This of course created a lot of turmoil which correspondingly translated into high degrees of volatility. The high degrees of volatility of the mining and quarrying industry are most surprising and have at first sight no direct relations with the crisis. The beta of the mining and quarrying industry, which is closest to zero of all industries, also indicates that this industry has little relation to the state of the economy.

On the other hand the manufacturing industry experienced the lowest degrees of volatility. This contradicts Rehman and Akbar (2011) findings who found that the crisis would have a bigger impact on manufacturing firms than on service firms. They expected that manufacturing firms would take a harder hit because they depend more on debt, and during a crisis it becomes much harder to attract capital. This debt financing problem did not cause high degrees of volatility in the sample that is taken in this thesis.

9.1 Limitations

32 References

Bartram, S. and Bodnar, G. (2009). “No place to hide: The global crisis in equity markets in 2008/2009”. Journal of International Money and Finance, Vol. 28, No. 8

Berk,J. DeMarzo,P. and Harford,J. (2008). “Fundamentals of corporate finance”. Prentice

Hall.

Bordo, M. D. (2008). “An Historical Perspective on the Crisis of 2007–2008”. NBER Working Papers No 14569.

Brooks, R. and Catão,L. ( 2000). “ The New Economy and Global Stock Returns”. IMF

working paper 00/216, Washington: International Monetary Fund.

Campbell, J.Y., Lettau,M., Malkiel,B.G. and Xu,Y. (2001) “Have individual

stocks become more volatile? An empirical exploration of idiosyncratic risk”. Journal of Finance 1–43.

Catao, L. and Timmerman, A.(2005) “Country and Industry Dynamics in Stock Returns”. CEPR Discussion Papers, 4368, C.E.P.R. Discussion Papers

Cavaglia, S., Brightman, C. and M. Aked. (2000). “The increasing importance of industry factors”. Financial Analysts Journal (September/October), 41–54.

Chari, V.V., Christiano, L.J. and Kehoe,P.J. (2008). “Facts and Myths about the Financial Crisis of 2008”. Working Paper, Federal Reserve Bank of Minneapolis.

Ferreira, M.A., Gama, P.M. (2005). “Have world, country and industry risks changed over time? An investigation of the developed stock markets volatility”. Journal of Financial and Quantitative Analysis 40, 195–222.

Grigor'ev, L. and Salikhov, M. (2009). “Financial Crisis 2008: Entering Global Recession”. Problems of Economic Transition. 51(10). 35-62.

33 Mahran, S.M.R. (2011). “Study the impact of the global crisis on stock prices and liquidity in the stock market”. 2011 International Conference on E-business, Management and Economics

IPEDR Vol.25

Moore,T. (2011). “The volatility spillover from the market to disaggregated industry stocks: The case for the US and UK”. International Journal of Business and Economics, Vol. 10, No. 1, pp. 61-68, 2011

Nieto, M.J. and Schinasi,G. (2007). “EU framework for safeguarding financial stability: towards an analytical benchmark for assessing its effectiveness”. IMF Working Paper WP/07/260, IMF, Washington, DC.

Pisani-Ferry, J. and Sapir,A. (2010). “Banking crisis management in the EU: an early assessment”. Economic Policy April 2010 pp. 341–373

Rehman,S. and Akbar,S. (2011). “The effect of credit crisis on performance and investment policies of the UK public firms”. Working paper at: http://ssrn.com/abstract=1966528

Schoenmaker, D. and Oosterloo,S. (2007). “Cross-border issues in European financial Supervision”. In D.G. Mayes and G. Wood (eds.), The Structure of Financial Regulation, Routledge, London.

Schwert, G. W (2002). “Stock volatility in the new millennium: How wacky is Nasdaq?”. Journal of monetary economics, Vol 49, 2002, pp. 3-26.

Schwert, G. W (2011). “Stock volatility during the recent financial crisis”. European Financial Management, Vol. 17, No. 5, 2011, 789–805.

Soros, G. (2008). “ The new paradigm for financial markets: The credit crisis of 2008 and what it means ”. New York: Public Affairs.

34 Appendix 1: Bivariate correlations (of one industry with the other industries)

Correlations

Industry B Industry C Industry D Industry E Industry F

35 Correlations

Industry G Industry H Industry I Industry J Industry K

36 Correlations

Industry L Industry M Industry N Industry R

37 Correlations

Industry B Industry C Industry D Industry E Industry F

Industry N Pearson Correlation ,267 ,728** ,550** ,489** ,670** Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 N 1667 1667 1667 1667 1667 Industry R Pearson Correlation ,156** ,476 ,418** ,314** ,470** Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 N 1667 1667 1667 1667 1667 Correlations

Industry G Industry H Industry I Industry J Industry K

Industry N Pearson Correlation ,651 ,653** ,624** ,668** ,613** Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 N 1667 1667 1667 1667 1667 Industry R Pearson Correlation ,390** ,463 ,422** ,478** ,445** Sig. (2-tailed) ,000 ,000 ,000 ,000 ,000 N 1667 1667 1667 1667 1667 Correlations

Industry L Industry M Industry N Industry R

Industry N Pearson Correlation ,528 ,644** 1** ,365** Sig. (2-tailed) ,000 ,000 ,000 N 1667 1667 1667 1667 Industry R Pearson Correlation ,322** ,399 ,365** 1** Sig. (2-tailed) ,000 ,000 ,000 N 1667 1667 1667 1667

38 Appendix II: Auto correlations (one industry with the other industries)

40

45 Appendix III: Volatility plots per industry

52 Appendix IV: Descriptive statistics per industry (All companies and their main statistics)

58 Industry E

59 Industry G

Industry H

63 Appendix V: Data and plot explaining variable, Market Value

Average volatility per Industry per year (Yellow = high volatility level)

B C D E F G H I J K L M N R 2006 0,0299 0,0020 0,0064 0,0175 0,0063 0,0061 0,0052 0,0128 0,0046 0,0039 0,0283 0,0143 0,0186 0,0411 2007 0,0376 0,0021 0,0080 0,0145 0,0081 0,0089 0,0058 0,0140 0,0043 0,0069 0,0416 0,0139 0,0271 0,0361 2008 0,1665 0,0075 0,0440 0,0693 0,0198 0,0458 0,0155 0,0444 0,0131 0,0274 0,0861 0,0417 0,0367 0,1002 2009 0,0862 0,0059 0,0210 0,0468 0,0122 0,0189 0,0123 0,0319 0,0097 0,0358 0,1299 0,0309 0,0285 0,0951 2010 0,0326 0,0038 0,0053 0,0164 0,0088 0,0076 0,0059 0,0138 0,0033 0,0139 0,0170 0,0134 0,0164 0,0433 2011 0,0535 0,0041 0,0119 0,0268 0,0074 0,0099 0,0101 0,0179 0,0053 0,0217 0,0329 0,0164 0,0274 0,1054

Average market value per industry per year

64 Appendix VI: Regressions per industry with other industries with one lag

Model Summary

Model R R Square Adjusted R

Square Std. Error of the Estimate Change Statistics R Square Change F Change 1 ,819a ,671 ,668 ,0099699 ,671 240,091 ANOVAa

Model Sum of Squares df Mean Square F Sig.

1

Regression ,334 14 ,024 240,091 ,000b

Residual ,164 1651 ,000

Total ,498 1665

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients t Sig. B Std. Error Beta 1 (Constant) 6,351E-005 ,000 ,258 ,796 Industry B LAG ,052 ,015 ,052 3,435 ,001 Industry C LAG 1,082 ,067 ,905 16,089 ,000 Industry D LAG ,078 ,037 ,060 2,093 ,037 Industry E LAG -,058 ,024 -,055 -2,443 ,015 Industry F LAG ,142 ,040 ,148 3,573 ,000 Industry G LAG -,104 ,033 -,094 -3,163 ,002 Industry H LAG -,096 ,044 -,077 -2,194 ,028 Industry I LAG -,030 ,027 -,030 -1,110 ,267 Industry J LAG -,396 ,051 -,313 -7,703 ,000 Industry K LAG ,140 ,031 ,162 4,566 ,000 Industry L LAG -,044 ,016 -,054 -2,668 ,008 Industry M LAG ,050 ,027 ,054 1,855 ,064 Industry N LAG ,069 ,025 ,059 2,823 ,005 Industry R LAG -,009 ,014 -,010 -,610 ,542

65 Model Summary

Model R R Square Adjusted R

Square Std. Error of the Estimate Change Statistics R Square Change F Change 1 ,104a ,011 ,002 ,0144669 ,011 1,294 ANOVAa

Model Sum of Squares df Mean Square F Sig.

1

Regression ,004 14 ,000 1,294 ,203b

Residual ,346 1651 ,000

Total ,349 1665

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients t Sig. B Std. Error Beta 1 (Constant) ,000 ,000 ,645 ,519 Industry B LAG ,000 ,022 ,000 ,014 ,989 Industry C LAG ,175 ,098 ,175 1,792 ,073 Industry D LAG -,008 ,054 -,007 -,141 ,888 Industry E LAG ,033 ,034 ,037 ,966 ,334 Industry F LAG -,070 ,058 -,088 -1,221 ,222 Industry G LAG ,050 ,048 ,054 1,043 ,297 Industry H LAG ,048 ,063 ,046 ,763 ,445 Industry I LAG ,066 ,039 ,079 1,688 ,092 Industry J LAG -,186 ,075 -,176 -2,496 ,013 Industry K LAG -,004 ,045 -,005 -,080 ,936 Industry L LAG -,039 ,024 -,057 -1,613 ,107 Industry M LAG -,034 ,039 -,044 -,877 ,380 Industry N LAG -,009 ,036 -,009 -,251 ,802 Industry R LAG -,007 ,021 -,009 -,322 ,747

66 Model Summary

Model R R Square Adjusted R

Square Std. Error of the Estimate Change Statistics R Square Change F Change 1 ,106a ,011 ,003 ,0133060 ,011 1,328 ANOVAa

Model Sum of Squares df Mean Square F Sig.

1

Regression ,003 14 ,000 1,328 ,183b

Residual ,292 1651 ,000

Total ,296 1665

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients t Sig. B Std. Error Beta 1 (Constant) -6,006E-005 ,000 -,183 ,855 Industry B LAG ,012 ,020 ,015 ,591 ,554 Industry C LAG ,111 ,090 ,121 1,241 ,215 Industry D LAG ,046 ,050 ,046 ,912 ,362 Industry E LAG ,044 ,031 ,054 1,389 ,165 Industry F LAG -,040 ,053 -,054 -,754 ,451 Industry G LAG -,057 ,044 -,067 -1,313 ,189 Industry H LAG ,062 ,058 ,065 1,070 ,285 Industry I LAG ,015 ,036 ,020 ,421 ,674 Industry J LAG -,040 ,069 -,041 -,577 ,564 Industry K LAG -,050 ,041 -,075 -1,213 ,225 Industry L LAG -,017 ,022 -,027 -,766 ,444 Industry M LAG -,048 ,036 -,067 -1,327 ,185 Industry N LAG -,007 ,033 -,007 -,200 ,841 Industry R LAG -,011 ,019 -,016 -,557 ,578

67 Model Summary

Model R R Square Adjusted R

Square Std. Error of the Estimate Change Statistics R Square Change F Change 1 ,093a ,009 ,000 ,0164009 ,009 1,036 ANOVAa

Model Sum of Squares df Mean Square F Sig.

1

Regression ,004 14 ,000 1,036 ,413b

Residual ,444 1651 ,000

Total ,448 1665

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients t Sig. B Std. Error Beta 1 (Constant) ,000 ,000 -,372 ,710 Industry B LAG ,019 ,025 ,020 ,783 ,434 Industry C LAG ,191 ,111 ,168 1,723 ,085 Industry D LAG -,087 ,062 -,071 -1,411 ,158 Industry E LAG ,089 ,039 ,089 2,286 ,022 Industry F LAG -,052 ,065 -,057 -,790 ,429 Industry G LAG -,056 ,054 -,054 -1,042 ,297 Industry H LAG ,037 ,072 ,031 ,511 ,609 Industry I LAG ,050 ,044 ,053 1,131 ,258 Industry J LAG -,115 ,085 -,095 -1,353 ,176 Industry K LAG ,008 ,051 ,010 ,158 ,875 Industry L LAG -,006 ,027 -,008 -,221 ,825 Industry M LAG -,040 ,045 -,045 -,897 ,370 Industry N LAG -,030 ,040 -,027 -,755 ,451 Industry R LAG ,014 ,024 ,017 ,588 ,557

68 Model Summary

Model R R Square Adjusted R

Square Std. Error of the Estimate Change Statistics R Square Change F Change 1 ,099a ,010 ,001 ,0180468 ,010 1,161 ANOVAa

Model Sum of Squares df Mean Square F Sig.

1

Regression ,005 14 ,000 1,161 ,300b

Residual ,538 1651 ,000

Total ,543 1665

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients t Sig. B Std. Error Beta 1 (Constant) -2,514E-005 ,000 -,056 ,955 Industry B LAG -,003 ,027 -,003 -,110 ,912 Industry C LAG ,262 ,122 ,210 2,150 ,032 Industry D LAG ,043 ,068 ,032 ,639 ,523 Industry E LAG ,001 ,043 ,000 ,012 ,990 Industry F LAG -,063 ,072 -,063 -,874 ,382 Industry G LAG ,067 ,059 ,058 1,121 ,262 Industry H LAG ,046 ,079 ,035 ,581 ,562 Industry I LAG ,052 ,049 ,050 1,055 ,292 Industry J LAG -,191 ,093 -,145 -2,055 ,040 Industry K LAG -,039 ,056 -,043 -,697 ,486 Industry L LAG -,054 ,030 -,063 -1,799 ,072 Industry M LAG -,029 ,049 -,030 -,592 ,554 Industry N LAG -,046 ,044 -,038 -1,032 ,302 Industry R LAG -,005 ,026 -,006 -,195 ,846

69 Model Summary

Model R R Square Adjusted R

Square Std. Error of the Estimate Change Statistics R Square Change F Change 1 ,121a ,015 ,006 ,0156063 ,015 1,753 ANOVAa

Model Sum of Squares df Mean Square F Sig.

1

Regression ,006 14 ,000 1,753 ,040b

Residual ,402 1651 ,000

Total ,408 1665

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients t Sig. B Std. Error Beta 1 (Constant) 4,906E-005 ,000 ,127 ,899 Industry B LAG -,053 ,024 -,059 -2,250 ,025 Industry C LAG ,101 ,105 ,093 ,958 ,338 Industry D LAG ,060 ,059 ,051 1,025 ,305 Industry E LAG -,001 ,037 -,001 -,035 ,972 Industry F LAG -,024 ,062 -,027 -,383 ,702 Industry G LAG ,064 ,051 ,064 1,241 ,215 Industry H LAG ,030 ,068 ,027 ,441 ,659 Industry I LAG ,055 ,042 ,061 1,305 ,192 Industry J LAG -,268 ,081 -,234 -3,324 ,001 Industry K LAG ,035 ,048 ,045 ,731 ,465 Industry L LAG -,046 ,026 -,063 -1,782 ,075 Industry M LAG -,027 ,042 -,032 -,638 ,524 Industry N LAG ,038 ,038 ,036 1,000 ,317 Industry R LAG ,021 ,022 ,026 ,928 ,354