0 Bewertungen0% fanden dieses Dokument nützlich (0 Abstimmungen)

6 Ansichten8 SeitenEconometric Methods Phase 2 of Work

Oct 07, 2016

© © All Rights Reserved

DOCX, PDF, TXT oder online auf Scribd lesen

Econometric Methods Phase 2 of Work

© All Rights Reserved

Als DOCX, PDF, TXT **herunterladen** oder online auf Scribd lesen

0 Bewertungen0% fanden dieses Dokument nützlich (0 Abstimmungen)

6 Ansichten8 SeitenEconometric Methods Phase 2 of Work

© All Rights Reserved

Als DOCX, PDF, TXT **herunterladen** oder online auf Scribd lesen

Sie sind auf Seite 1von 8

Phase 1

X

450,000

400,000

350,000

300,000

250,000

200,000

150,000

1980

1985

1990

1995

2000

2005

2010

2015

2000

2005

2010

2015

Y

160,000

140,000

120,000

100,000

80,000

60,000

40,000

20,000

1980

1985

1990

1995

Both X (UK GDP) and Y (UK Imports) have a positive linear relationship over time, with

expected anomalies due to economic events. Both immediately see a decline, GDP falls

3.45% between 1980 and 1981 and Imports fell 17.4% in the same period. This was due to

the early 1980s recession, with GDP not reaching its 1980 level until the fourth quarter 1982.

Both variables generally continue with a steady increase, until the UK economy was affected

by another global recession in 1990. Variable X decreased for 5 straight quarters (1990 Q2-

1991 Q3) whilst Y was decreasing for 4 (1990-Q1-1991-Q1). For a long period of time

following this, a steady incline of both X and Y against time took place. A significant rise of

UK imports occurred in 2005, which wasnt as noticeable in the UK GDP graph. However

the global financial crisis of 2008 is very visible on both graphs, in which the greatest period

of fall in X and Y is recorded. Upon recovery from the recession, both graphs display a

general return to the steady increase in values up to the present day.

Phase 2

If the expected disturbance is zero, then Yt will equal to the result of the constant, B0, plus a

value of B1 for each additional unit of Xt. Thus, starting with B0, for every additional 1 of Xt,

you can expect Yt to increase by B1.

Dependent Variable: Y

Method: Least Squares

Date: 04/19/16 Time: 16:46

Sample: 1980Q1 2015Q1

Included observations: 141

Variable

Coefficient

Std. Error

t-Statistic

Prob.

C

X

-80086.58

0.519456

1605.405

0.004993

-49.88558

104.0330

0.0000

0.0000

R-squared

Adjusted R-squared

S.E. of regression

Sum squared resid

Log likelihood

F-statistic

Prob(F-statistic)

0.987320

0.987228

4409.363

2.70E+09

-1382.263

10822.88

0.000000

S.D. dependent var

Akaike info criterion

Schwarz criterion

Hannan-Quinn criter.

Durbin-Watson stat

82399.46

39017.04

19.63493

19.67676

19.65193

0.274235

From studying this output, we are able to see that the OLS estimate of B1 is:

B1 = 0.519 (3 d.p.)

Also, that the OLS estimate of B0 is:

B0 = -80086.58

Thus, the sample regression suggests that as a result of a 1million increase in GDP, Imports

of Goods and Services will go up by 519,456 on average. This aligns with our expectations

as mentioned earlier.

Also, the value of B0 implies that when GDP is equal to zero, then our imports of goods and

services will be -80086.58. For this number to be positive, there is a certain amount of GDP

we must have. This is found by dividing -B0 by B1, which is 154,309.40 (2d.p.).

From the output, we also find the Durbin-Watson statistic, which is 0.274235 for this model.

This is a 1 variable model, with 141 observations. The corresponding set of d statistics have a

lower bound of 1.720, and an upper bound of 1.746. The given d statistic lies very well below

this range; thus we must accept that they are subject to first-order autocorrelation.

Phase 3

If the expected disturbance is zero, B0 represents the average value of Y when X and Z are

equal to zero. B1 represents the change in the mean value of Y for each unit change of X,

when Z is held constant. B2 represents the change in the mean value of Y for each unit change

of Z when X is held constant. Therefore, when GDP and the exchange rate are zero, B0

should roughly be the value of imports of goods and services. When GDP goes up by one

unit, imports of goods and services increases by B1, with the exchange rate held constant.

When the exchange rate goes up by one unit, imports of goods and services goes up by B2,

with GDP held constant.

Z

130

120

110

100

90

80

70

1980

1985

1990

1995

2000

2005

2010

2015

Across the period shown, from 1980, to 2015, the exchange rate has decreased a significant

amount. The graph shows a peak of approximately 124 in the first quarter of 1981. Also in

the period shown, the graph shows a trough of approximately 78 in the first quarter of 2009.

This means that the exchange rate fell by 46 index points in the 28 years from 1981 to 2009.

These are index values, with 2005 as the base year.

Dependent Variable: Y

Method: Least Squares

Date: 04/19/16 Time: 20:31

Sample: 1980Q1 2015Q1

Included observations: 141

Variable

Coefficient

Std. Error

t-Statistic

Prob.

C

X

Z

-110167.6

0.535512

263.4654

4495.280

0.004868

37.46399

-24.50740

110.0005

7.032499

0.0000

0.0000

0.0000

R-squared

Adjusted R-squared

S.E. of regression

Sum squared resid

Log likelihood

F-statistic

Prob(F-statistic)

0.990665

0.990530

3796.939

1.99E+09

-1360.669

7322.618

0.000000

S.D. dependent var

Akaike info criterion

Schwarz criterion

Hannan-Quinn criter.

Durbin-Watson stat

82399.46

39017.04

19.34282

19.40556

19.36832

0.403730

From studying the output, we are able to see that the OLS estimate of B2 is:

B2 = 263.47 (2 d.p.)

Also, that the OLS estimate of B1 is:

B1 = 0.536 (3 d.p.)

And finally, that the OLS estimate of B0 is:

B0 = -110167.6

Thus, the sample regression suggests that as a result of a 1million increase in GDP, with the

exchange rate being held constant, our imports of goods and services will go up by 535,512.

Also, that if the exchange rate goes up by 1 index point, with GDP being held constant, then

our imports of goods and services increase by 263.47. Also, that when the exchange rate

index is set to zero, and GDP is zero, our imports of goods and services will be -110167.60.

a) Ho: B1 = 0 against Ha: B1 0

Using C(2) = 0. We get the following output:

Wald Test:

Equation: Untitled

Test Statistic

t-statistic

F-statistic

Chi-square

Value

df

Probability

110.0005

12100.11

12100.11

138

(1, 138)

1

0.0000

0.0000

0.0000

Value

Std. Err.

0.535512

0.004868

Null Hypothesis Summary:

Normalized Restriction (= 0)

C(2)

Studying this output, we see the t-statistic probability is less than 0.05, therefore we reject

our null hypothesis in favour of our alternate hypothesis.

b) Ho: B2 = 0 against Ha: B2 0

Wald Test:

Equation: Untitled

Test Statistic

t-statistic

F-statistic

Chi-square

Value

df

Probability

7.032499

49.45604

49.45604

138

(1, 138)

1

0.0000

0.0000

0.0000

Value

Std. Err.

263.4654

37.46399

Null Hypothesis Summary:

Normalized Restriction (= 0)

C(3)

Studying this, we see our t-statistic probability is less than 0.05, thus we reject the null

hypothesis in favour of our alternate hypothesis.

c) Ho: B2 = 0, B1 = 0 against Ha: at least one of B1 0, B2 0.

Using C(2) = 0, C(3) = 0. We get the following output:

Wald Test:

Equation: Untitled

Test Statistic

F-statistic

Chi-square

Value

df

Probability

7322.618

14645.24

(2, 138)

2

0.0000

0.0000

Value

Std. Err.

0.535512

263.4654

0.004868

37.46399

Null Hypothesis Summary:

Normalized Restriction (= 0)

C(2)

C(3)

Studying this, we can see the F-statistic probability is less than 0.05, thus we can reject

the null hypothesis, in favour of the alternate hypothesis, meaning at least one of B1 or B2

does not equate to zero.

There are four measures for the goodness of fit, they include the following:

Residual Sum of Squares (RSS)

Coefficient of Determination (R2)

Standard Error of the Regression ( ^

2

The Adjusted R-squared Statistic ( R

For the first regression model, the goodness of fit measures are as follows:

RSS = 2702505...- This value indicates the amount of variance that is unexplained by the

regression model. Typically, the smaller this value is, the better.

R2 = 98.7320% - This indicates that the model explains almost all the variability of the

response data around the mean.

^ = 4409.363 - This shows us how wrong our model is, in terms of the units of Y. This is

the average distance the observed values fall from the regression line. Typically, the smaller

the number, the better.

2 = 98.7228% - This is still a very high value, thus it indicates the model has a good fit.

R

For the second regression model, the goodness of fit measures are as follows:

RSS = 1989510...- This value is smaller than that of its counterpart for the 1st regression

model, thus the 2nd model has a better goodness of fit.

R2 = 99.0665% - This indicates that the model explains almost all of the variability of the

response data around the mean, even more so than the 1st model. Thus based on this, the 2nd

model has a better goodness of fit as opposed to the 1st model.

^ = 3796.939 This value is smaller than the one for the 1st regression model, thus the

second has a better goodness of fit.

2 = 99.0530% - After adjustment it is not as high, but is still higher than that of the 1st

R

model, thus the 2nd has a better goodness of fit.

From the output, we also find the Durbin-Watson statistic, which is 0.403730 for this model.

This is a 2 variable model, with 141 observations. The corresponding set of d statistics have a

lower bound of 1.706, and an upper bound of 1.760. The given d statistic lies very well below

this range; thus we must accept that there is first order autocorrelation.

Phase 4

X

log .( t)+ B2 log .(Z t )+ut

log . ( Y t ) =B0 +B 1

This model uses the standardised form, the coefficients will determine how much Imports

responds to a change in its variables (GDP and Exchange Rate).

B1, is the (constant) partial elasticity of Imports with respect to GDP, whilst B2 is the

(constant) partial elasticity of Imports with respect to UK Effective Exchange Rate. They

provide the sensitivity of a change in these variables to a change in Imports.

Variable

Coefficient

Std. Error

t-Statistic

Prob.

C

LOG(X)

LOG(Z)

-16.60486

2.142638

0.164697

0.288884

0.014166

0.034264

-57.47927

151.2518

4.806740

0.0000

0.0000

0.0000

R-squared

Adjusted R-squared

S.E. of regression

Sum squared resid

Log likelihood

F-statistic

Prob(F-statistic)

0.995280

0.995211

0.036206

0.180904

269.3573

14548.46

0.000000

S.D. dependent var

Akaike info criterion

Schwarz criterion

Hannan-Quinn criter.

Durbin-Watson stat

11.19252

0.523205

-3.778118

-3.715379

-3.752623

0.521050

An Ordinary Least squares regression show estimates of B1 and B2, as 2.1426 and 0.1647

respectively. B0 is given by the coefficient of C, which is -16.6049. The estimates show that

Imports are more sensitive to changes in GDP than it is to Effective Exchange Rate, seen by

the difference between 2.1426 and 0.1647.

For the second regression model, the goodness of fit measures are as follows:

RSS = 1989510...

R2 = 99.0665%

^ = 3796.939

2

R

= 99.0530%

For the third regression model, the goodness of fit measures are as follows:

RSS = 0.180904 This is a lot smaller in comparison to the 2nd model, thus we must accept

the 3rd model has a better goodness of fit.

R2 = 99.5280% - This is only marginally higher than that of the 2nd model, but it is higher

nonetheless. Hence, the 3rd model has a better goodness of fit.

^ = 0.036206 This is a considerable amount smaller than that of the 2nd model, thus the

3rd model has a better goodness of fit.

2 = 99.5211% - The difference compared to the unadjusted R2 is negligible, but it is still

R

higher compared to the second model. Thus we must accept that the 3rd model has a better

goodness of fit in comparison.

The Durbin-Watson statistic is displayed in the regression as 0.5211, which is comfortably

under 2. This provides evidence of positive serial correlation, so are subject to first-order

autocorrelation.

Phase 5

There is less standard error in the first model as it contains only one explanatory variable.

This will give a more accurate correlation between Imports and GDP, however this is not a

sufficient amount of variables for a reliable model. Imports are dependent on more than just

this one variable, so variable Z (Exchange Rate) being added in the second model will help

eliminate the under specificity issues from the first. Although these are not the only two

variables affecting imports, it is preferable over the first which is even more underspecified.

An issue with the second model, is that the standard error of B0 has increased greatly, with the

addition of a second variable. All three regression models follow a fitted regression line, with

their high R-squared values. The third model gives the highest R-squared value, being only

0.0047 away from 1. The third model also uses standardised values, making it easier to

compare the two variables used. Lagged variable could instead be used, to distinguish

between short-run and long-run effects. A dynamic variable like this will help factor in timelags for import changes in a response to these variables used. All three models suffer from

first order autocorrelation, as distinguished from the Durbin-Watson tests. This could

potentially be rectified by transforming the data used. With only 2 variables, it is unlikely all

effect on Import changes will be covered.

## Viel mehr als nur Dokumente.

Entdecken, was Scribd alles zu bieten hat, inklusive Bücher und Hörbücher von großen Verlagen.

Jederzeit kündbar.