Econometrics  data on expenditure on food and total expenditure

Question 1

Foodexp.xls  gives  data on expenditure on food and total expenditure measured in  $, for a
sample of 55 households in U SA  for a recent period. Using this data, perform the following
tasks (using EVIEWS)

(a)  Import these data into an EVIEWS work file. Plot the data on food  expenditure and total
expenditure, and sketch a regression line through the scatter plots.
( 20 marks)

(b)  Regress food  expenditure  on  an  intercept  and  total  expenditure.  Comment  on the  R
square and the adjusted R square of this regression.
(20 marks)

(c)  Comment on the F statistic of this regression as well as the t ratio for the significance of
the coefficient of total expenditure.
(20 marks)

(d)  Create the logarithmic transformation of the two variables (food expenditure and total
expenditure)  (call   the  resulting  variable  lnfood   and  lntotal   respectively).Graph  lnfood   and
lntotal   together and comment on the graph.
(20 marks)

(e)   Regress  lnfood   on intercept and lntotal . Test whether the coefficient of  lntotal   is 1 using a
t test.

(20 marks)

You m ust provide EVIEWS output incorporated in your MS-Word document and fully explain
your actions and findings by commenting on the provided EVIEWS  output directly.

Question 2

Consider the following regression results (estimated standard errors in  parentheses) based
on data from a local cinema obtained at a monthly frequency over a two year period (n=24).

75. 0
) 10. 0( ) 16. 0( ) 03. 1(
ln 30. 0 ln 76. 0 54. 1
ˆ
ln
2
3 2
?
? ? ?
R
X X Y
t t t

where  Yt
= quantity of ‘hot- dogs’ demanded;
X2t
= price of ‘hot- dogs’;
X3t
=price of nachos;
R
2
=the coefficient of determination; and
ln = the natural logarithm of a variable.

(a) Comment on the economic interpretation that can be given to the estimates attached to
lnX
2t
and  lnX
3t
and  consider  whether  these  estimat es  suggest  hot – dogs  and  nachos  are
substitutes or complements.
3

(40 marks)
(b) Comment on the statistical significance of each of the estimated coefficients at the 5 and
10 per cent levels.                  (30 marks)

(c )   Use an F- test to test   the statistical significance of the regression model, i.e., test whether
the population parameters attached to lnX 2t   and  lnX 3t   are jointly equal to zero.