Normality Test Heteroscedasticity Test

43 Table 3.1 The Difference between Cross section and Time series Data Year Country X 1 X 2 Y 2009 A 4.3 9.2 26 2010 A 4.1 10.4 40 2011 A 4.1 9.3 53 2009 B 5.2 4.6 18 2010 B 4.3 6.2 36 2011 B 3.4 6.5 22 2009 C 7.5 2.8 41 2010 C 7.3 1.6 27 2011 C 7 2.8 10 According to Gujarati 2004:640, the advantages of using panel data are: a. Panel data give more informative data, more variability, more degrees of freedom and more efficiency. b. Panel data are better suited to study the dynamics of change. c. Panel data can better detect and measure effects that simply cannot be observed in pure cross-section or pure time series data. d. By making data available for several thousand units, panel data can minimize the bias that might result if we aggregate individuals or firms into broad aggregates. Time dimension Cross section 44 There are three models on data panel analysis, they are: Pooled Ordinary Least Square PLS, Fixed Effect Model FEM, and Random Effect Model REM. The alternative model in panel data is selected after complete the Chow test, Hausman test.

a. Pooled Ordinary Least Square PLS

According to Nachrowi and Usman 2006:311 generally, the technique used in PLS is similar with the commonly used regression. The difference between PLS and commonly used regression is, in panel data, we have to combine the cross section and time series pool data before we define the regression model. The following equation shows the combination of cross section and time series or pool data: Where: N = the number of cross section individual. T = the number of time period. α = alpha = beta With the assumption of error component, we can separately estimate each unit of individual cross section. The regression individual cross section equation is: Y it = α + X it + it ; i = 1, β, ….., N ; t = 1,β, ……, T 45 For t = 1 the equation is: Source: Nachrowi and Usman 2006:312

b. Fixed Effect Model FEM

In FEM there is a possibility of change in α on each i and t Nachrowi and Usman, 2006:313 FEM is consider on the differences in parametric value on both cross-section and time series through add the dummy variable Fadhliyah, 2008. The equation for FEM is: Nachrowi and Usman, 2006:313 Where: Y it = dependent variable for individual i and t period X it = independent variable for individual i and t period α = alpha = beta Y it = α + X it + it ; i = 1, β, ….., N Y i1 = α + X i1 + i1 ; i = 1, β, ….., N Y it = α + X it + 2 W 2t + …. N W Nt + 2 Z i2 + …. + T Z iT + it