• Exchange rate

The study is an extension of the approach suggested by Karolyi [20], Longin, and Solnik [31] to examine the future contracts (such as foreign exchange rates,

treasury bond, and index of stock prices). To this end, we propose the following hypothesis:

H3: The exchange rate impact is positive on the stock market volatility

• Market index:

A hint of what is designed to measure price changes of a set of markets, such as the stock market or the bond market. This variable was also considered by Zhian et al. [32] and Koulakiotis et al. [33].

H4: The impact of the market index on the stock market volatility is positive

*3.3.2 Model*

#### *3.3.2.1 GARCH model*

Eqs. (8) and (9) show the return and volatility equations, respectively, which have been used in the investigation of the impact of corporate governance variables on the volatility persistence and error terms. Accordingly, the corporate governance variables are embedded in the model below to detect the effect on volatility and error:

$$\begin{aligned} \pi\_t &= \beta\_0 + \beta\_1 s\_1 + \beta\_2 s\_2 + \varepsilon\_t \\ \varepsilon\_t / \pi\_{t-1} &\sim T(\mathbf{0}, h\_t) \end{aligned} \tag{8}$$

variables. It may be noted that the lowest coefficients of negative skewness are recorded for boards of directors, while the highest skewness is recorded in the case

different variables studied are not normally distributed.

SBF120 1.000000 0.0392

*The correlation matrix of corporate governance and risk management.*

SBF120 9.8018

Exchange rate 17.4655

Institutional investors 25.5105

Board size 1.1e + 02

*Unit root test based on levels of variables for all four panels.*

Exchange rate 1.000000 0.0569

**Variables First level**

The coefficients of kurtosis variables are significantly more than three (SBF 120). This shows that for these series, which have a flatter distribution than the normal distribution, all other distributions are leptokurtic. According to the test Jarque-Bera (JB), the null hypothesis (*H*0) of normality is not rejected, and the

**Table 11** shows the correlations of all the variables. We observe any high correlations among the independent variables that might affect our regression results. This table shows the correlations between all the variables. We observe high correlations among the independent variables that might affect our regression results. **Table 12** shows the results of the panel unit root tests for the levels of the variables. It can be seen from **Table 12** that all the variables in first difference are statistically significant under the LLC and HLM tests, indicating that all variables are integrated of order one, I(1). Furthermore, the results shown in the table indicate that all the series that display values LLC and HLM are below the critical values. Therefore, we accept hypothesis H1. The variables of this study are stationary and integrated of order zero because there is no differentiation for the first

In **Table 13**, we can observe that the results uncover non-normality since both the Ljung-Box (Q(10)) and the Breusch-Godfrey LM statistics point to the absence of autocorrelation in the residual series, which reveals that the chosen AR (1) specification seems sufficient to eliminate any serial correlation present in the data. Our results showed the stationarity constraint of the model is verified (α + β < 1) for all the equations, which supports a weak presence of effect ARCH and GARCH

**Variables SBF 120 Exchange rate Board size INST INV**

Board size 1.000000 0.0539

Institutional investors 1.00000

0.4686

(0.0000)

(0.0000)

(0.0000)

(0.0000)

0.0569 0.2930

0.2930

**LLC HLM**

0.0809 0.1345

0.0108 0.8420

0.3188

1.3191 0.0936

6.0866 (0.0000)

2.9574 (0.0016)

7.0130 (0.0000)

of returns SBF120.

*The Primary Origin of the Financial Crisis DOI: http://dx.doi.org/10.5772/intechopen.86173*

stationary.

**Table 11.**

**Table 12.**

**25**

*s*<sup>1</sup> denotes the variable of corporate governance of the average board size. The second corporate governance variable *s*<sup>2</sup> controls the share of employee representatives. The sample period is from 2006 to 2013. A symmetric response to shocks is made from Bollerslev's univariate GARCH model:

$$h\_t = a\_0 + a\_1 h\_1 + a\_2 \varepsilon\_{t-1}^2 \tag{9}$$

#### *3.3.3 Empirical results and discussions*

**Table 10** reports the summary statistics and the diagnostic tests of AR (1) residuals. We can observe that the results uncover non-normality since the Jarque-Bera test rejects the null hypothesis of Gaussianity at 1% level. The series also displays a negative skewness and leptokurtic behavior, symptomatic of a heavier tailed distribution than the standard.

The descriptive statistics of the different variables for the panel are given in **Table 1**.

From **Table 1**, we find that the coefficients of skewness are positive in some cases and negative in others; it is to that the distribution of the variables is shifted left asymmetric for some variables (board of administration) and right for other


**Table 10.**

*Summary statistics of corporate governance and risk management.*

### *The Primary Origin of the Financial Crisis DOI: http://dx.doi.org/10.5772/intechopen.86173*

treasury bond, and index of stock prices). To this end, we propose the following

A hint of what is designed to measure price changes of a set of markets, such as the stock market or the bond market. This variable was also considered by Zhian

H4: The impact of the market index on the stock market volatility is positive

Eqs. (8) and (9) show the return and volatility equations, respectively, which have been used in the investigation of the impact of corporate governance variables on the volatility persistence and error terms. Accordingly, the corporate governance variables are embedded in the model below to detect the effect on volatility and

*rt* ¼ *β*<sup>0</sup> þ *β*1*s*<sup>1</sup> þ *β*2*s*<sup>2</sup> þ *ε<sup>t</sup>*

*s*<sup>1</sup> denotes the variable of corporate governance of the average board size. The second corporate governance variable *s*<sup>2</sup> controls the share of employee representatives. The sample period is from 2006 to 2013. A symmetric response to shocks is

*ht* <sup>¼</sup> *<sup>α</sup>*<sup>0</sup> <sup>þ</sup> *<sup>α</sup>*1*h*<sup>1</sup> <sup>þ</sup> *<sup>α</sup>*2*ε*<sup>2</sup>

**Table 10** reports the summary statistics and the diagnostic tests of AR (1) residuals. We can observe that the results uncover non-normality since the Jarque-Bera test rejects the null hypothesis of Gaussianity at 1% level. The series also displays a negative skewness and leptokurtic behavior, symptomatic of a heavier

The descriptive statistics of the different variables for the panel are given in

From **Table 1**, we find that the coefficients of skewness are positive in some cases and negative in others; it is to that the distribution of the variables is shifted left asymmetric for some variables (board of administration) and right for other

SBF120 0.169785 0.851913 6.300151 61.426594 56358.580960 Institutional investors 0.377604 0.085123 0.484719 �0.907466 25.274010 Exchange rate 2.007108 0.000049 0.454862 �0.623050 17.426292 Board size 1.090817 0.020644 �0.772521 1.339174 59.921052

**Mean Variance Skewness Kurtosis Jarque-Bera**

(8)

*<sup>t</sup>*�<sup>1</sup> (9)

*εt=π<sup>t</sup>*�<sup>1</sup> � *T*ð Þ 0*; ht*

made from Bollerslev's univariate GARCH model:

*Summary statistics of corporate governance and risk management.*

*3.3.3 Empirical results and discussions*

tailed distribution than the standard.

H3: The exchange rate impact is positive on the stock market volatility

hypothesis:

*3.3.2 Model*

error:

**Table 1**.

**Table 10.**

**24**

• Market index:

*3.3.2.1 GARCH model*

et al. [32] and Koulakiotis et al. [33].

*Financial Crises - A Selection of Readings*

variables. It may be noted that the lowest coefficients of negative skewness are recorded for boards of directors, while the highest skewness is recorded in the case of returns SBF120.

The coefficients of kurtosis variables are significantly more than three (SBF 120). This shows that for these series, which have a flatter distribution than the normal distribution, all other distributions are leptokurtic. According to the test Jarque-Bera (JB), the null hypothesis (*H*0) of normality is not rejected, and the different variables studied are not normally distributed.

**Table 11** shows the correlations of all the variables. We observe any high correlations among the independent variables that might affect our regression results. This table shows the correlations between all the variables. We observe high correlations among the independent variables that might affect our regression results.

**Table 12** shows the results of the panel unit root tests for the levels of the variables. It can be seen from **Table 12** that all the variables in first difference are statistically significant under the LLC and HLM tests, indicating that all variables are integrated of order one, I(1). Furthermore, the results shown in the table indicate that all the series that display values LLC and HLM are below the critical values. Therefore, we accept hypothesis H1. The variables of this study are stationary and integrated of order zero because there is no differentiation for the first stationary.

In **Table 13**, we can observe that the results uncover non-normality since both the Ljung-Box (Q(10)) and the Breusch-Godfrey LM statistics point to the absence of autocorrelation in the residual series, which reveals that the chosen AR (1) specification seems sufficient to eliminate any serial correlation present in the data. Our results showed the stationarity constraint of the model is verified (α + β < 1) for all the equations, which supports a weak presence of effect ARCH and GARCH


#### **Table 11.**

*The correlation matrix of corporate governance and risk management.*


#### **Table 12.**

*Unit root test based on levels of variables for all four panels.*


institutional investors in firms reduce their stock price volatility, and hence, they become less stock market's volatility. Secondly, a clear understanding of the stock market volatility and effects of institutional investors is important for policy makers in making relevant policies on foreign capital restrictions, especially policies in

response to shocks during the financial crisis.

*The Primary Origin of the Financial Crisis DOI: http://dx.doi.org/10.5772/intechopen.86173*

**Author details**

\* and Jarboui Anis<sup>2</sup>

and Management of Sfax, Tunisia

Business Administration of Sfax, Tunisia

provided the original work is properly cited.

\*Address all correspondence to: mounafba@yahoo.fr

1 Department of Financial and Accounting, University of Faculty Economics

2 Department of Financial and Accounting, Universities Higher Institute of

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,

Aloui Mouna1

**27**

**Table 13.**

*Univariate GARCH effects with and without the impact of corporate governance variables.*

in all the cases (exchange rate and institutional investor), except for (board size, SBF120), i.e., (α + β) ≥ 1 has a high persistence of volatility shocks. So, in this we can see that the institutional investors reduce their stock price volatility.
