Professional Documents
Culture Documents
Lecture 9
Lecture 9
ECONOMETRICS
DR ABDUL WAHEED
PhD Econometrics
FUNDAMENTALS OF
ECONOMETRICS
Week 5
Lecture 9
Log-Lin or growth
models
A topic of great interest to economists, the government,
the business sector, and policy makers is the rate of
growth of key economic variables.
Such as GDP, money supply, population, employment,
productivity and interest rates, etc.
models
Measuring growth rates
To see how the growth rate of an economic variable
can be measured or estimated.
We will be measure the rate of growth of real GDP
(RGDP).
For this purpose we will use the following model
models
where RGDP stands for real GDP,
r is the rate of growth, and
t is time measured chronologically.
It is the well-known compound interest formula from basic
finance.
Dependent variable = ,
Independent Variable = time (t)
Save the results for discussion
Interpret the results of Log-Lin model
models
Results Dependent Variable: LNRGDP
Method: Least Squares
Date: 02/27/23 Time: 15:20
Sample: 1 48
Included observations: 48
Variable Coefficient Std. Error t-Statistic Prob.
C 7.875662 0.009759 807.0072 0.0000
TIME 0.031490 0.000347 90.81657 0.0000
R-squared 0.994454 Mean dependent var 8.647156
Adjusted R-squared 0.994333 S.D. dependent var 0.442081
S.E. of regression 0.033280 Akaike info criterion -3.926969
Sum squared resid 0.050947 Schwarz criterion -3.849003
Log likelihood 96.24727 Hannan-Quinn criter. -3.897506
F-statistic 8247.650 Durbin-Watson stat 0.347740
Prob(F-statistic) 0.000000
models
Interpretation of results
➢ Statistical Significance?
➢ Overall Statistical significance?
➢ R2 Value?
➢ Interpretation of Intercept & Slope?
Intercept:
anti-log (7.8756) = 2632.27
which is the beginning value of real GDP, that is, the value at the beginning
of 1960, our starting point.
The actual value of RGDP for 1960 was about $2501.8 billion.
models
Slope:
The USA’s real GDP had been increasing at the rate of 3.15% per
year over the period of 1960–2007. This growth rate is
statistically significant, for the estimated t value of about 90.82
is highly significant.
A technical note:
The coefficient gives the instantaneous rate of growth (at a
point in time) and not the compound rate of growth (over a
period of time), r.
models
Compound growth rate can be calculate from time parameter.
models
The linear trend model
A linear trend model can also be calculated of the form:
This is known as the linear trend model and the time variable is
known as the trend variable.
The slope coefficient in this model gives the absolute change
(not relative or percentage) in RGDP per unit time period.
If is positive, there is an upward trend in RGDP, but if it is
negative, there is a downward trend in RGDP or any regressand.
models
Class Activity-5.2
Download data file Table 2.5 from LMS
Import data to EViews
Estimate the following regression model
Dependent variable = ,
Independent Variable = time (t)
Save the results for discussion
Interpret the results of model
models
Results Dependent Variable: RGDP
Method: Least Squares
Date: 02/27/23 Time: 15:30
Sample: 1 48
Included observations: 48
Variable Coefficient Std. Error t-Statistic Prob.
C 1664.218 131.9990 12.60781 0.0000
TIME 186.9939 4.689886 39.87174 0.0000
R-squared 0.971878 Mean dependent var 6245.569
Adjusted R-squared 0.971267 S.D. dependent var 2655.520
S.E. of regression 450.1314 Akaike info criterion 15.09773
Sum squared resid 9320440. Schwarz criterion 15.17570
Log likelihood -360.3455 Hannan-Quinn criter. 15.12719
F-statistic 1589.756 Durbin-Watson stat 0.069409
Prob(F-statistic) 0.000000
models
Interpretation of results
➢ Statistical Significance?
➢ Overall Statistical significance?
➢ R2 Value?
➢ Interpretation of parameters?
These results show that over the period 1960–2007, real GDP
in the USA increased by about $187 billion per year, showing
an upward trend.
models
Comparison
Dependent Variable: LNRGDP Dependent Variable: RGDP
Method: Least Squares Method: Least Squares
Date: 02/27/23 Time: 15:20 Date: 02/27/23 Time: 15:30
Sample: 1 48 Sample: 1 48
Included observations: 48 Included observations: 48
Variable Coefficient Std. Error t-Statistic Prob. Variable Coefficient Std. Error t-Statistic Prob.
C 7.875662 0.009759 807.0072 0.0000 C 1664.218 131.9990 12.60781 0.0000
TIME 0.031490 0.000347 90.81657 0.0000 TIME 186.9939 4.689886 39.87174 0.0000
R-squared 0.994454 Mean dependent var 8.647156 R-squared 0.971878 Mean dependent var 6245.569
Adjusted R-squared 0.994333 S.D. dependent var 0.442081 Adjusted R-squared 0.971267 S.D. dependent var 2655.520
S.E. of regression 0.033280 Akaike info criterion -3.926969 S.E. of regression 450.1314 Akaike info criterion 15.09773
Sum squared resid 0.050947 Schwarz criterion -3.849003 Sum squared resid 9320440. Schwarz criterion 15.17570
Log likelihood 96.24727 Hannan-Quinn criter. -3.897506 Log likelihood -360.3455 Hannan-Quinn criter. 15.12719
F-statistic 8247.650 Durbin-Watson stat 0.347740 F-statistic 1589.756 Durbin-Watson stat 0.069409
Prob(F-statistic) 0.000000 Prob(F-statistic) 0.000000
models
Autocorrelation
Since we are dealing with time series data, the Durbin–Watson
statistic, which is a measure of autocorrelation in the error
term, is an important statistic.
Later we will see how interpret this statistic.