In old chapter have been discussed about the formation of multiple arrested development theoretical account on informations beginning, variables and the technique uses to analyse informations. Hence, this survey applies a few specific trials to seek the existent determiner which influenced economic growing.

Therefore, in this chapter the empirical consequences and findings through a scope of econometrics patterning are presented. In which, assorted trials was invoked to analyse informations important between variables. By utilizing the E-views package system, the trials that will be conducted including descriptive statistic trial, panel least squares and Pairwise Granger Causality trials. Those trial were depends on the one-year informations of four ASEAN states which cover Malaysia, Indonesia, Philippines and Vietnam from twelvemonth 2000 to 2008. Therefore, survey findings will be discussed with more elaborate in this chapter four.

## 4.1 UNIT ROOT Trial

In this survey, stationary of variables are tested by the Phillips-Perron ( PP ) unit root trial. The ground for look intoing these belongingss is to guarantee that series used is free from clip dependence. Furthermore, utilizing the non-stationary variables may take to specious arrested development consequence. The coefficient estimates on the lagged value for the PP trials for the stationarity of variables are reported in table 4.1. Besides that, each variable consists of two types of series which can be categorized as changeless without tendency considered as intercept and changeless with tendency considered as tendency and stop severally in degree and first difference for PP trial.

Refer to consequence of PP trials at table 4.1 below, which in the class invariable without tendency show that all variables except for life anticipation at flat signifier failed to reject void hypothesis ( non-stationary ) even at 10 % . Therefore, it indicates that these variables are non-stationary at their degree signifier, contain a unit root trial. However, in the class of changeless with tendency show that cyberspace users and export are undistinguished at 1 % , 5 % and 10 % important degree which the series can non be rejected because t-statistic values in absolute signifier are less than the t-critical values. Therefore, the variables are non-stationary in degree. On the other manus, the consequence of unit root trial after first differencing of series are important at 1 % , 5 % and 10 % important degree, which show that all variables are stationary in first difference, by the t-calculated values in absolute footings exceed the t-critical values. Therefore, this indicates that after the first differencing, the hypothesis of a unit root can be rejected for all variables. Therefore, all the variables viz. GDP per capital, life anticipation, cyberspace users, gross fixed capital formation and export are stationary and integrated at I ( 1 ) procedure.

Table 4.1: Unit of measurement Root Test

Degree

First DIFFERENT

Variable

Changeless Without Tendency

Changeless With Tendency

Changeless Without Tendency

Changeless With Tendency

GDP Per Capita

( lgdpc )

0.09066 ( 0 )

( 1.0000 )

23.9271 ( 1 ) **

( 0.0024 )

38.9572 ( 1 ) ***

( 0.0000 )

44.2510 ( 1 ) ***

( 0.0000 )

Life Expectancy

( llife )

15.2642 ( 0 ) *

( 0.0542 )

18.4213 ( 0 ) **

( 0.0183 )

56.5574 ( 0 ) ***

( 0.0000 )

41.2113 ( 1 ) ***

( 0.0000 )

Internet Users

( lnet )

5.88915 ( 0 )

( 0.6596 )

7.90339 ( 0 )

( 0.4430 )

14.5720 ( 1 ) *

( 0.0680 )

17.1406 ( 1 ) **

( 0.0287 )

Capital Formation

( lc )

1.47814 ( 0 )

( 0.9931 )

27.3973 ( 0 ) ***

( 0.0006 )

33.7789 ( 1 ) ***

( 0.0000 )

38.4062 ( 1 ) ***

( 0.0000 )

Export

( lexp )

5.90768 ( 0 )

( 0.6576 )

11.2141 ( 0 )

( 0.1899 )

14.5338 ( 1 ) *

( 0.0689 )

15.0337 ( 1 ) *

( 0.0585 )

Note: For PP trial, the void hypothesis is that the series is non-stationary or incorporate a unit root trial. The figure in parenthesis ( ) is mentioning to the selected slowdown length which based on the Schwarz Info Criterion ( SIC ) .While, figures in blacket exhibit t-critical values which *** , ** and * denote significance at 1 % , 5 % and 10 % significance degrees.

## 4.2 DESCRIPTIVE STATISTIC Trial

This portion reported the value mean, average, minimal, maximal, standard divergence, Skewness, Kurtosis and Jarque-bera for all variables. Mean is a step that has cardinal inclination information. It is a value that adds with intent to be predetermined place and place the value where informations centralized. Besides that, mean is one norm value from series informations that obtained through informations add-on series and so split with study figure. Therefore, computation mean are heightening all value and divided with value figure.

Median besides is one of cardinal inclination step. Median value is individual value that obtained with determine the in-between location for informations set or figure. Generally, for the distributions that skew to the right or positively skew, intending that the value of mean will ever larger from average and average value were larger from manner value. Whereas, for distribution that skew to the left ( or negatively ) , mode value were larger from average value and average value were larger from average value. Therefore, either the information set distribution is positive or negative skew ; normally median is the best location step. This is because the average value ever in the center which is between mean and manner value.

Following is looks on the value of upper limit and lower limit of consecutive in current sample. Meanwhile, the standard divergence is use to mensurate the distribution variable in consecutive. It is difference between observations with average informations. The standard divergence step is most effectual in mensurating observation distance from average informations. When there is broad informations set distribution, standard divergence value was high and if there are no distribution which all informations value are same so the standard divergence is zero. Standard divergence is more feasible because it include all value that has inside in informations set. Apart from that, Skewness is an dissymmetry distribution. In Where, the positive Skewness significance that the distribution have rear long on the right and whereas negative Skewness show that distribution have rear long on the left. Here, the Skewness readings show as follows:

• If S=0, frequence distribution were normal and symmetricalness

• If S & A ; gt ; 0, frequence distribution is slope positive.

• If S & A ; lt ; 0, frequence distribution is slope negative.

Then, Kurtosis is to mensurate consecutive degree. In extra, Kurtosis besides is a parametric quantity which reflects random variable chance distribution signifier that considers two chance maps denseness ( PDFs ) . For the Jarque-Bera trial which it is one statistical trial to prove whether consecutive informations are normal distribution. The descriptive statistic result for this survey indicated in table 4.2 as follows:

Table 4.2: Descriptive Statistic Analysis

## Variable

## Mean

## Median

## Maximum

## Minimum

## Standard Deviation

## Lopsidedness

## Kurtosis

## Jarque-Bera

## GDP Per Capital ( LGDPC )

## 7.110258

## 6.901144

## 8.547004

## 5.995269

## 0.81726

## 0.676536

## 2.144689

## 3.843545 ( 0.146347 )

## Life Expectancy ( LLIFE )

## 4.271654

## 4.275137

## 4.309183

## 4.210214

## 0.028322

## 0.429903

## 2.09509

## 2.337189

## ( 0.310803 )

## Internet Users

## ( LNET )

## 14.89725

## 5.763431

## 55.8

## 1.257614

## 17.26447

## 1.285357

## 3.282965

## 10.03296

## ( 0.006628 )

## Gross Fixed Capital Formation

## ( LC )

## 23.80928

## 23.82145

## 24.79341

## 22.91136

## 0.50067

## 0.200547

## 2.083726

## 1.50065

## ( 0.472213 )

## Export ( LEXP )

## 24.88013

## 24.90521

## 25.85341

## 23.56556

## 0.619069

## -0.246117

## 2.196854

## 1.331007

## ( 0.514015 )

Note: *** , ** and * denote significance at 1 % , 5 % and 10 % significance degrees.

## .

Based on table 4.2 above, show that the highest mean value is export that is 24.88 % and the lowest average value is life anticipation that is 4.27 % . Besides that, average value for GDP per capital is 6.90 % , life anticipation is 2.28 % , cyberspace and gross fixed capital formation is 5.76 % and 23.82 % severally and average value for export is 24.91 % .

In footings of upper limit and minimal value, show that the maximal value of GDP per capital is 8.55 % and the minimal value is 5.99 % . The maximal and minimal value in exogenic variable for life anticipation is every bit much as 4.31 % and 4.21 % severally. While, for the cyberspace show that maximal value 55.8 % and lower limit is 1.26 % . Apart from that, the maximal value for gross fixed capital formation is 24.79 % and lower limit is 22.91 % . Finally, the upper limit and minimal value for export is 25.85 % and 23.57 % . However, the standard divergence analyses show that GDP per capita is 0.82 % . While, independent variable for life anticipation is 0.03 % , cyberspace every bit much as 12.26 % , gross fixed capital formation and export is 0.50 % and 0.62 % severally.

Following, look on the Skewness trial which it shows of step for distribution dissymmetry around mean. Skewness consequence show four variable that is GDP per capital, life anticipation, cyberspace and gross fixed capital formation variable are positive inclination. This due to normal distribution is zero which if value which found more than zero so that it was positive lopsidedness. However, export variable show the negative inclination where the lopsidedness coefficient was lower from zero ( -0.25 % ) . This is because average value is lower than average value and the average value is lower than manner ( average & A ; lt ; average & A ; lt ; manner ) .

While, Kurtosis in this statistics portion show that there is one variable were exceeded value 3, viz. cyberspace users which showed that the distribution is highest degree from normal distribution. In extra, the GDP per capital, life anticipation, gross fixed capital formation and export variables is less than 3, so the distribution is horizontal compared with normal distribution.

Finally, the Jarque-Bera was proving whether informations series is the normal distribution. Jarque-Bera is mensurating the difference between Skewness and Kurtosis from normal distribution series. Therefore, GDP per capital, life anticipation, cyberspace and gross fixed capital formation in Jarque-Bera show that reject void hypothesis from normal distribution in important degree at 10 % and cyberspace reject void hypothesis in important degree at 1 % .

## 4.3 PANEL LEAST SQUARES TEST

Panel least square used to gauge the multiple arrested developments theoretical account which it has been built in informations analysis portion and theoretical account specification in chapter three. Postpone 4.1 showed the best consequences after has been done the chiefly estimated. Whereby, due to insignificant of instruction consequence in chiefly panel least squares trial hence instruction variable bead out from the estimated theoretical account. Therefore, the new Pooled OLS theoretical account can be specified as below:

Yit = ?0 + ?1X1it + ?2X2it + ?3X3it + ?4X4it + µit

In where,

Yit = Log ( Gross domestic merchandise per capital ( changeless 2000 US $ ) )

X1 = Log ( Life anticipation at birth, entire ( old ages ) )

X2 = Interner users ( per 100 people )

X3 = Log ( Gross fixed capital formation ( changeless 2000 US $ ) )

X4 = Log ( Export of goods and services ( changeless 2000 US $ ) )

µ = Error term

I = Cross sectional unit ( state )

T = Time series ( one-year 2000-2008 )

i+t = Pool informations

The panel least squares trial conducted by utilizing 9 old ages data including from twelvemonth 2000 until 2008. In extra, this end product consequence aims to see important degree on t-statistic value and analyse whether dependent variable of arrested development theoretical account are influenced by each one of independent variables. So, the hypothesis for this t-statistic can be composing as follows:

Bottom of Form

Holmium: Variable Ten does non act upon variable Y.

Hour angle: Variable Ten influence variable Yttrium.

Therefore, the appraisal of end product showed as below:

Table 4.3: Panel Least Squares Test for GDP per capital

Variable

Coefficient

t-statistic

Prob.

C

17.12996

1.650517

0.1089

LLIFE

5.275369

2.591492

0.0144**

LNET

0.016339

3.641892

0.0010***

LC

1.078806

9.082798

0.0000***

LEXP

-1.525597

-11.63816

0.0000***

R-squared ( R2 )

0.960781

Adjusted R-squared

0.955721

F-statistic

189.8599

Prob ( F-statistic )

0.000000

Durbin-Watson stat.

0.437249

N ( pool )

36

Note: *** , ** and * denote significance at 1 % , 5 % and 10 % significance degrees.

## .

In where,

LLIFE – Life Anticipation

LNET – Internet Users

LC – Gross Fixed Capital Formation

LEXP – Export

The theoretical account equation estimation:

LGDPC = 17.129+ 5.273 LLIFE + 0.016 LNET + 1.079 LC – 1.526 LEXP

Based on Table 4.1, demoing consequences that published by panel least square trial which four independent variables that is life anticipation, cyberspace users, gross fixed capital formation and export show are important relationship with GDP per capital in degree of significance at 5 % and 1 % by expression into important degree in value of chance ( p-value ) . Hence, reject a void hypothesis and concluded there is important relationship between each variables with GDP per capital.

Multiple arrested development consequence show through coefficient determiner ( R2 ) which it used to mensurate the fittingness of the theoretical account and this construct of coefficient besides used to find the strength of relationship among two variables. Therefore, the value of coefficient determiner ( R? ) is 0.9608 shows that every bit much as 96.08 % fluctuations which occurred in GDP per capital interpretable by independent variable. On the other manus, remain 3.92 % explain by other factor outside the theoretical account.

Besides that, the adjusted R? that worth every bit much as 0.9557 mean there are every bit much as 95.57 % fluctuations which occurred in GDP per capital can be explain by see the loss of freedom degree. Following, trial for the cogency of the theoretical account by utilizing the F-statistic which besides calls for the jointly trial. In which, the void hypothesis statement is non valid ( H0: & A ; szlig ; 1= & A ; szlig ; 2= & A ; szlig ; 3= & A ; szlig ; 4=0 ) and for the alternate hypothesis is valid ( H1: & A ; szlig ; 1= & A ; szlig ; 2= & A ; szlig ; 3= & A ; szlig ; 4?0 ) . Through the p-value ( prob ) show that it smaller than 1 % ( 0 & A ; lt ; 0.01 ) important degree. This mean, reject void hypothesis and conclude that articulation together life anticipation, cyberspace users, gross fixed capital formation and export can explicate GDP per capital.

Furthermore, harmonizing to the coefficient end product consequence which it indicates that if life anticipation grew by 1 % , GDP per capital will grew by 5.275 % . Besides that, increase 1 % in cyberspace and gross fixed capital formation will raise the GDP per capital by 0.016 % and 1.079 % severally. Finally, if export grew by 1 % , GPD per capital death every bit much as 1.526 % by assume other variables are changeless for each partial coefficient arrested development estimated. Besides that, the consequence besides point out that export variable is still important to act upon the GDP per capital at flat 1 % although the coefficient was negative. For the t-statistic end product, the independent variables viz. life anticipation, cyberspace users and gross fixed capital formation were positive, except export variable was found to be negative in this trial.

In extra, by expression into important degree in value of chance ( p-value ) able to see whether the independent variables consequence of dependant ( GDP per capital ) variable. Therefore, the t-statistic in life anticipation variable is 5.275 ( p=0.0144 ) , internet users is 0.016 ( p=0.0010 ) , gross fixed capital formation is 1.079 ( p=0.0000 ) and export is -1.526 ( p=0.0000 ) so concluded that all of this four variables is important at confident degree 5 % and 1 % . Hence, reject a void hypothesis demoing there is important relationship between all four variables with GDP per capital.

## 4.4 GRANGER CAUSALITY TESTS

The Granger ( 1969 ) attack goes into the inquiry whether ten ( independent variables ) causes y ( dependent variables ) which to see how much of the current Y can be explained by past values of Y and so to see whether adding lagged values of x ( independent variables ) can better the account. Besides that, Y ( dependent variable ) is said to be Granger-caused by x ( independent variable ) if x ( independent variables ) helps in the prognosis of Y ( dependent variables ) . Therefore, Granger causality can be divide into two classs there are one manner causality which x ( independent variables ) Granger cause Y and Y does non Granger cause ten ( independent variables ) and the other classs is two manner causing which the instance of x ( independent variables ) Granger causes Y ( dependent variables ) and y ( dependent variables ) Granger causes x ( independent variables ) .

The void hypothesis of this Granger causality trial can be create as ten does non Granger cause Y in the first arrested development and so Y does non Granger cause ten ( independent variable ) in the 2nd arrested development. While, for the alternate hypothesis show as x ( independent variable ) has causal consequence on Y ( dependent variable ) in the first arrested development and so y ( dependent variable ) has causal consequence on ten ( independent variables ) in the 2nd arrested development. Furthermore, this survey consequence of end product purposes to see the important degree on F-statistic value which can be clearly seen through important degree in value of chance ( p-value ) .

Table 4.3 shows the consequence of Granger causality between independent variables those are life anticipation, cyberspace, gross fixed capital formation and export with dependent variable that is GDP per capital. For the consequence of life anticipation with GDP per capital show that reject the void hypothesis in first and 2nd arrested development which indicated life anticipation has causal consequence on GDP per capital and every bit good as GDP per capital besides has causal relationship on life anticipation. Therefore, conclude that it appears bipartisan causality that is from life anticipation causal on GDP per capital and every bit good as the GDP per capital causal on life anticipation.

Besides that, granger causality between cyberspace and GDP per capital show that rejects the void hypothesis at 5 % important degree which indicates that the cyberspace has a causal relationship with GDP per capital. On the other manus, 2nd arrested development can non reject the hypothesis mean that GDP per capital has no causal consequence on cyberspace. Hence, it shows that farmer causality runs one manner from cyberspace to GDP per capital. Furthermore, the consequence of F-statistic in gross fixed capital formation is 0.53919 ( p=0.5904 ) . Therefore, concluded that fail to reject void hypothesis mean that gross fixed capital formation has no causal consequence on GDP per capital. Meanwhile, for the void hypothesis of GDP per capital does non granger cause gross fixed capital formation demo the F-statistic is 2.99167 ( p=0.0700 ) where this show reject void hypothesis at 10 % important degree and conclude that GDP per capital has a causal consequence on gross fixed capital formation. Overall, indicate that there a one manner farmer causality that is from GDP per capital to gross fixed capital formation.

Finally, consequence of export variables with GDP per capital point out export has causal relationship with GDP per capital so reject the void hypothesis in first arrested development at 5 % degree of important where the F-statistic is 6.35402 ( p=0.0086 ) and for the 2nd arrested development was fail to reject void hypothesis with F-statistic is 0.07907 ( p=0.9242 ) indicate that GDP per capital has no causal consequence on export. As a consequence, there are one manner causality which from export to GDP per capital.

Table 4.4: Granger Causality Trials

## Null Hypothesis

## F-Statistic

## Prob.

## LLIFE does non Granger Cause LGDPC

## 6.08140

## 0.0076

## LGPDC does non Granger cause LLIFE

## 20.1544

## 9.00E-06

## LNET does non Granger Cause LGDPC

## 3.56269

## 0.0449

## LGPDC does non Granger cause LNET

## 0.01269

## 0.9874

## LC does non Granger Cause LGDPC

## 0.53919

## 0.5904

## LGPDC does non Granger cause LC

## 2.99167

## 0.0700

## LEXP does non Granger Cause LGDPC

## 6.35402

## 0.0086

## LGPDC does non Granger cause LEXP

## 0.07907

## 0.9242

Note: *** , ** and * denote significance at 1 % , 5 % and 10 % significance degrees.

In where:

GDPC -GDP per capital

LLIFE – Life Anticipation

LNET – Internet Users

LC – Gross Fixed Capital Formation

LEXP – Export

## 4.5 Decision

Overall, this survey theoretical account can be interpreted through trials that were implemented as above. Therefore, the survey findings show that panel least square trial handled to analyse coefficient, t-statistic value and chance ( p-value ) independent variable on dependent variable severally, R square, adjusted R square and F-statistic. Besides that, the descriptive statistics trial is use to analyses informations distribution in studied theoretical account. Finally, Granger causality trial which it used to identity whether the independent and dependent variable have one manner or two manner consequence.