where represents

the convergence in distribution and represents

a chi-squared distribution with (k -j + 1) degrees of freedom. Cheung and Ng (1996)

recommended that this test statistic can be utilized to test empirically the

null hypothesis of no causality in mean from lag j to lag k.

Now the causality in variance test has been focused

here. Let u and v be the squares of the standardized innovations given by Evidently both u and v are unobservable, hence their

estimates can be utilized to test the

hypothesis of no causality in variance.

In this case sample cross-correlation coef?cient at lag k, can

be computed from the consistent estimates of the conditional mean and variance

of X series and Y series. This is given by

This test statistic is to be used to test

empirically the null hypothesis for no causality in variance from lag j to lag

k.

Evidently test statistic of Cheung and Ng (1996)

(say, S1and S2) may be influenced to severe size

distortion in presence of causality in mean, hence the weighting cross-correlation has been

incorporated in this paper as suggested by Hong (2001) into the CCF approach.

Thus, test statistics for the causality

in mean and causality in variance are presented by

4.

Empirical Analysis

4.1 Univariate Descriptive

Statistics: Country wise descriptive

statistics have been calculated, by breaking the sample into three disjoint

portions (before, during and after) with respect to Brexit. Results are

presented in table 1 to table 3.

A significant change in descriptive

statistics for various parts of the sample is found and it indicates a strong

impact of Brexit over the selected stock markets. Moreover, it is to be noted that The Jarque–Bera

test rejects the null hypothesis of normality in all the

cases as the p-values are less than or equal to 0.05. Hence it

can be stated with 95% confidence, that the stock market index returns does not

fit to the normal distribution. It gives a clear indirect indication that autoregressive conditional heteroskedasticity

(ARCH) characteristics exist in our sample. It motivates to use econometric

models in the subsequent portion of this paper.*-

Table-1

: Descriptive Statistics & testing of normality before Brexit (February 23, 2016 to

June 23,2016)

India

Japan

Russia

China

UK

Mean

0.000234

0.000585

0.001056

0.000932

0.000102

Variance

0.020035

0.034432

0.034351

0.029841

0.005645

Jarque–Bera

678.7091

1124.752

102.6660

1732.5

308.0233

Probability

0.000000

0.000000

0.000000

0.000000

0.000000

Data

Source: http://?nance.yahoo.com/

Result: Computed using SPSS and MS Excel with respect to the ?rst order differences in logarithmic stock

indices prices.

Table-2

: Descriptive Statistics & testing

of normality during Brexit period

(June 24, 2016 to March 29, 2017)

India

Japan

Russia

China

UK

Mean

0.000204

0.000513

0.001002

0.000732

0.000100

Variance

0.029035

0.044002

0.035951

0.029996

0.006641

Jarque–Bera

671.7091

1104.752

102.0060

1630.5

301.0233

Probability

0.000000

0.000000

0.000000

0.000000

0.000000

Data

Source: http://?nance.yahoo.com/

Result: Computed using SPSS and MS Excel with respect to the ?rst order differences in logarithmic stock

indices prices.

Table-3

: Descriptive Statistics & testing of normality after Brexit (March

30, 2017 to September 29,2017)

India

Japan

Russia

China

UK

Mean

0.000104

0.000413

0.001001

0.000402

0.000070

Variance

0.029835

0.044092

0.039951

0.039096

0.008601

Jarque–Bera

661.7090

1104.702

101.0029

1530.2

301.0003

Probability

0.000000

0.000000

0.000000

0.000000

0.000000

Data

Source: http://?nance.yahoo.com/

Result: Computed using SPSS and MS Excel with respect to the ?rst order differences

in logarithmic stock indices prices.

4.2 Bivariate Descriptive Statistics: Unconditional correlations as well as partial

correlations are calculated for all possible pair of countries similarly for

the disjoint parts of the sample. For

the partial correlations, all remaining countries are considered as control

variables. The results are presented in table 4 to table 6. Their results show

a high level dependency among the selected stock markets.

It is to be noted that all countries

are highly correlated, indicating their high dependence on each other. Thus any

significant change in the stock market of a country will definitely impact

seriously on the stock markets of other countries. After removing the influence

of other countries (which represents control variables in partial

correlations), when partial correlations have been computed, then also

significant dependencies are noticed. It represents the dynamic linkages

between international financial markets. However, the values of partial

correlations are less than the Pearson bivariate correlations. This empirical

evidence shows that interdependencies of stock markets are influenced by stock

market movements of other countries too.

It is to be noted also that, after

Brexit, all the correlations have been reduced , indicating a possible weakness

of dependencies due to the occurrence of the international event of Brexit.

Hence there is a clear indication of the impact of Brexit on stock markets of the

selected developing countries in Asia. The subsequent part of this paper,

confirm this evidence by considering advanced econometric tools.

Table-4:

Unconditional correlation & Partial Correlation matrix before Brexit (February 23,

2016 to June 23,2016)

Correlations

India

Japan

Russia

China

UK

India

Pearson Correlation

Sig. (2-tailed)

Partial

Japan

Pearson Correlation

.945

Sig. (2-tailed)

.002

Partial

.794

Russia

Pearson Correlation

.800

.827

Sig. (2-tailed)

.003

.000

Partial

.765

.678

China

Pearson Correlation

.866

.655

.600

Sig. (2-tailed)

.005

.001

.000

Partial

.762

.611

.567

UK

Pearson Correlation

.700

.789

.866

.766

Sig. (2-tailed)

.002

.002

.002

.002

Partial

.654

.701

.694

.664

Source: http://?nance.yahoo.com/

Result: Computed using SPSS and MS Excel with respect to the ?rst order differences

in logarithmic stock indices prices.

Table-5:

Unconditional correlation & Partial Correlation matrix during Brexit period

(June 24, 2016 to March 29, 2017)

Correlations

India

Japan

Russia

China

UK

India

Pearson Correlation

Sig. (2-tailed)

Partial

Japan

Pearson Correlation

.845

Sig. (2-tailed)

.002

Partial

.694

Russia

Pearson Correlation

.750

.727

Sig. (2-tailed)

.003

.000

Partial

.660

.613

China

Pearson Correlation

.706

.625

.587

Sig. (2-tailed)

.005

.001

.000

Partial

.664

.600

.560

UK

Pearson Correlation

.609

.709

.766

.666

Sig. (2-tailed)

.002

.002

.002

.002

Partial

.558

.691

.624

.610

Source: http://?nance.yahoo.com/

Result: Computed using SPSS and MS Excel with respect to the ?rst order differences

in logarithmic stock indices prices.

Table-6:

Unconditional correlation & Partial Correlation matrix after Brexit (March 30, 2017 to

September 29,2017)

Correlations

India

Japan

Russia

China

UK

India

Pearson Correlation

Sig. (2-tailed)

Partial

Japan

Pearson Correlation

.705

Sig. (2-tailed)

.002

Partial

.592

Russia

Pearson Correlation

.650

.627

Sig. (2-tailed)

.003

.000

Partial

.612

.573

China

Pearson Correlation

.636

.525

.507

Sig. (2-tailed)

.005

.001

.000

Partial

.604

.509

.510

UK

Pearson Correlation

.559

.639

.666

.506

Sig. (2-tailed)

.002

.002

.002

.002

Partial

.550

.591

.521

.530

Source: http://?nance.yahoo.com/

Result: Computed using SPSS and MS Excel with respect to the ?rst order differences

in logarithmic stock indices prices.