University of Westminster

 

 

Heteroskedasticity 

 

The Brusch-Pegan test is used to check for hetereskedasticity problem. The following table illustrates the 

results of the test. 

 

 

The null hypothesis is that there is no heteroscedasticity in the data. The results show that the probability value 

is highly insignificant therefore, we fail to reject the null hypothesis and conclude that the date employed is 

homoscedastic. 

 

 

 

 

 

 

 

 

 

         Prob > chi2  =    0.0080

         chi2(14)     =     29.86

         against Ha: unrestricted heteroskedasticity

White's test for Ho: homoskedasticity

---

WESTMINSTER INTERNATIONAL UNIVERSITY IN TASHKENT 

9 

001023, 00953, 000933, 001012 

 

Model specification test 

 

Based on the results of the above test hatsq is not significant so linktest failed to reject the null hypothesis and it 

can be concluded that the model is correctly specified.  

The following model was chosen for estimation: 

 

 

Y= β

0 + 

β

1

LnFin+ β

2

LnRD+β

3

Age+β

4

ManAb+β

5

 Gender+ β

6

AccFin+β

7

Exp+u 

 

 

The dependent variable is total annual sales, which is taken in logarithmic form, 

Independent variables are LnFin, which is log of expenditure on fixed assets, such as land, buildings and 

machinery; 

LnRD is log of R&D expenditure; 

Age is firms age 

ManAb is top managers number of years of experience in the firm. 

There are also control variables: 

Gender is dummy variable whether the top manager of the firm is female. It takes the value of 1 if the top 

manager is female and 0 otherwise. 

AccFin is also dummy variable, which indicates whether the firm has access to loan from financial institution.  

Exp is the direct exports which is the measured as a % of total sales.