• Goldfeld-Quandt test

GQ test checks if model is the homoscedastic or heteroschedastic according to the Gauss-Markov theorem’s condition. Model is homoschedastic if:

But,

Fcrit > GQ and < 1/GQ ==> the variance isn’t constant in different time periods. Then our model is heteroschedastic. Then we must reestimate our model and should divide data by the variance of the disturbance term.

• Durbin-Watson test

The test assumes that d lies between 4 and 0. The critical value of d at any significance level depends on the number of explanatory variables in the regression equation and the number of observations in the sample. Our estimations show that d critical is 0,71 and it lies between dl (1,08) and du (1,53). It means that there is no information about autocorrelation of residuals.

Testing new econometric model (data is divided by the variance of the disturbance term)

As I mentioned before our model didn’t pass GQ test and to solve this problem we should divide initial data by the variance of the disturbance term. So, we get new initial data:

	Net national disposable income	Final consumption expenditures
Year	X	Y
1991	0,00000000011	0,00000000010
1992	0,00000000011	0,00000000011
1993	0,00000000012	0,00000000012
1994	0,00000000013	0,00000000012
1995	0,00000000013	0,00000000013
1996	0,00000000014	0,00000000013
1997	0,00000000015	0,00000000014
1998	0,00000000016	0,00000000015
1999	0,00000000017	0,00000000016
2000	0,00000000018	0,00000000017
2001	0,00000000019	0,00000000018
2002	0,00000000019	0,00000000019
2003	0,00000000020	0,00000000020
2004	0,00000000022	0,00000000021
2005	0,00000000023	0,00000000022
2006	0,00000000025	0,00000000024
2007	0,00000000025	0,00000000025
2008	0,00000000026	0,00000000026
2009	0,00000000025	0,00000000026
2010	0,00000000026	0,00000000027

New specification of the estimated model is:

Table 1:

Regression Statistics
Multiple R	0,995963339
R Square	0,991942973
Adjusted R Square	0,99146903
Standard Error	4,81896E-12
Observations	19

Table 2:

Table 3:

All tests have the same results then we can move to checking adequacy. You can see it in excel file accompanying this document.

Adequacy of the model

Now I’m going to check the adequacy of this model and I’ll use confidence interval for it.