Date post: | 19-Dec-2015 |
Category: |
Documents |
View: | 212 times |
Download: | 0 times |
1
Marietta College
Spring 2011
Econ 420: Applied Regression Analysis
Dr. Jacqueline Khorassani
Week 14
2
Tuesday, April 12
Exam 3: Monday, April 25, 12- 2:30PMBring your laptops to class on Thursday too
Collect Asst 21Use the data set FISH in Chapter 8 (P 274) torun the following regression equation:F = f (PF, PB, Yd, P, N)1) Conduct all 3 tests of imperfect
multicollinearity problem and report your results.
2) If you find an evidence for imperfect multicollinearity problem, suggest and implement a reasonable solution.
3
4
Use EViews
• Open FISH in Chapter 8 • Run P = f (PF, PB, Yd, N)• Click on view on regression output • Click on actual, fitted, residual• Click on residual graph• Do you suspect the residuals to be
autocorrealted?
5
This is what you should have got
-.5
-.4
-.3
-.2
-.1
.0
.1
.2
.3
.4
46 48 50 52 54 56 58 60 62 64 66 68 70
P Residuals
Positive residual is followed by positive residual possible positive autocorrelation
6
Causes of Impure Serial Correlation
1. Wrong functional form– Example: effect of age of the house on its price
2. Omitted variables– Example: not including wealth in the consumption
equation
3. Data error
7
Cause of Pure Serial Correlation
• Lingering shock over time– War– Natural disaster– Stock market crash
8
Consequences of Pure Autocorrelation
• Unbiased estimates but wrong standard errors–In case of positive autocorrelation
standard error of the estimated coefficients drops
–Consequences on the t-test of significance?
9
Consequences of Impure Autocorrelation
• Biased estimates• Plus wrong standard errors
10
Let’s look at first order serial correlation
єt = ρ єt-1 + ut
ρ (row) is first order autocorrelation coefficient It takes a value between -1 to +1
u2 is a normally distributed error with the mean of zero and constant variance
11
A Formal Test For First Order Autocorrelation
• Durbin-Watson test• Estimate the regression equation• Save the residuals, e• Then calculate the Durbin -Watson Stat (d stat)
• d stat ~ 2 (1- ρ)• What is dstat under perfect positive correlation?
ρ = +1 d = 0• What is dstat under perfect negative correlation?
ρ = -1 d = 4• What is dstat under no autocorrelation?
ρ = 0 d = 2• What is the range of values for dstat?
0 to 4
12
dstat=0
Perfect positive autocorrelation
dstat=4
Perfect negative autocorrelation
dstat=2
No autocorrelation
If 2>dstat>0 then suspect (test for) positive autocorrelation
If 4>dstat>2 then suspect (test for) negative autocorrelation
13
EViews calculates d-stat automatically
• It is included in your regression output• Run P = f (PF, PB, Yd, N)• Do you see the d-stat?
14
Dependent Variable: PMethod: Least SquaresDate: 04/12/11 Time: 08:59Sample: 1946 1970Included observations: 25
Variable Coefficient Std. Error t-Statistic Prob. C -2.083188 0.271658 -7.668417 0.0000PF 0.027143 0.017355 1.563934 0.1335PB -0.012571 0.011620 -1.081865 0.2922YD 0.001597 0.000387 4.132263 0.0005N - 5.54E-05 1.27E-05 -4.376214 0.0003
R-squared 0.801154 Mean dependent var 0.160000Adjusted R-squared 0.761384 S.D. dependent var 0.374166S.E. of regression 0.182774 Akaike info criterion -0.384281Sum squared resid 0.668123 Schwarz criterion -0.140506Log likelihood 9.803514 Hannan-Quinn criter. -0.316668F-statistic 20.14505 Durbin-Watson stat 1.498086Prob(F-statistic) 0.000001
What type of serial correlation shall we test for?Positive
15
• If d stat<2, test for positive autocorrelation.• Null and alternative hypotheses
– H0: ρ≤0 (no positive auto)– HA: ρ>0 (positive auto)
• Choose the level of significance (say 5%)• Critical dstat (PP 591- 593)• Decision rule
– If dstat< dL reject H0 there is significant positive first order autocorrelation
– If dstat> dU don’t reject H0 there is no evidence of a significant autocorrelation
– if dstat is between dL and du the test is inconclusive.
16
Dependent Variable: PMethod: Least SquaresDate: 04/12/11 Time: 08:59Sample: 1946 1970Included observations: 25
Variable Coefficient Std. Error t-Statistic Prob. C -2.083188 0.271658 -7.668417 0.0000PF 0.027143 0.017355 1.563934 0.1335PB -0.012571 0.011620 -1.081865 0.2922YD 0.001597 0.000387 4.132263 0.0005N - 5.54E-05 1.27E-05 -4.376214 0.0003
R-squared 0.801154 Mean dependent var 0.160000Adjusted R-squared 0.761384 S.D. dependent var 0.374166S.E. of regression 0.182774 Akaike info criterion -0.384281Sum squared resid 0.668123 Schwarz criterion -0.140506Log likelihood 9.803514 Hannan-Quinn criter. -0.316668F-statistic 20.14505 Durbin-Watson stat 1.498086Prob(F-statistic) 0.000001
N = 25, K = 4At 5% leveldL= 1.04, dU =1.77dstat is between dL and du the test is inconclusive
17
DWstat=0
Perfect positive autocorrelation
DWstat=4
Perfect negative autocorrelation
DWstat=2
No autocorrelation
H0: ρ≤0 (no positive auto)HA: ρ>0 (positive auto)level of significance = 5% Critical d-stat
dL =1.04dU = 1.77
Decision dstat is between dL and du the test is inconclusive
1.771.04
Fail to reject H0Reject H0 inconclusive
1.5
18
• If dstat >2, you will to test for negative autocorrelation.
• Null and alternative hypotheses– H0: ρ≥0 (no negative auto)– HA: ρ<0 (negative auto)
• Choose the level of significance (1% or 5%)• Critical dstat (page 591- 593)• Decision rule
– If dstat>4-dL reject H0 there is significant negative first order autocorrelation
– If dstat< 4-dU don’t reject H0 there is no evidence of a significant autocorrelation
– if dstat is between 4 – dL and 4 – du the test is inconclusive.
19
ExampleDependent Variable: CONSUMPTIONMethod: Least SquaresDate: 11/09/08 Time: 20:11Sample: 1 30Included observations: 30
Variable Coefficient Std. Error t-Statistic Prob.
C 16222.97 5436.061 2.984324 0.0060
INCOME 0.641166 0.166878 3.842131 0.0007
WEALTH 0.148788 0.041327 3.600281 0.0013
R-squared 0.847738 Mean dependent var 52347.37Adjusted R-squared 0.836459 S.D. dependent var 31306.54S.E. of regression 12660.43 Akaike info criterion 21.82499Sum squared resid 4.33E+09 Schwarz criterion 21.96511Log likelihood -324.3748 Hannan-Quinn criter. 21.86982F-statistic 75.16274 Durbin-Watson stat 2.211726Prob(F-statistic) 0.000000
d-sta >2 test for negative autocorrelation
20
Let’s test for autocorrelation at 1% level in our example
H0: ρ≥0 (no negative auto)HA: ρ<0 (negative auto)
• 1% level of significance, k=2, n=30• dL=1.07, du= 1.34• 4- dL=2.93, 4- du= 2.66• dstat < 4- du, don’t reject H0
21
Asst 22: Due Thursday
• Use the data on Soviet Defense spending (Page 335– Data set: DEFEND Chapter 9) to regress SDH on SDL, UDS and NR only.
1. Conduct a Durbin-Watson test for serial correlation at 5% level of significance
2. If you find an evidence for autocorrelation, is it more likely to be pure or impure autocorrelation? Why?
22
Thursday April 15• Exam 3: Monday, April 25, 12- 2:30PM• Bring your laptops to class next Tuesday
23
Collect Asst 22
• Use the data on Soviet Defense spending (Page 335– Data set: DEFEND Chapter 9) to regress SDH on SDL, USD and NR only.
1. Conduct a Durbin-Watson test for serial correlation at 5% level of significance
2. If you find an evidence for autocorrelation, is it more likely to be pure or impure autocorrelation? Why?
24
Solutions for Autocorrelation Problem
• If the D-W test indicates autocorrelation problem
• What should you do?
25
1. Adjust the functional form• Sometimes autocorrelation is because we use a linear form while we
should have used a non-linear form
revenue
Price
*
*
**
*
With a linear line, errors have formed a pattern
The first 3 observations have positive errors
The last 2 observations have negative errors
Revenue curve is not linear (It is bell shaped)
What should we use?1
2
3
4
5
26
2. Add other relevant (missing) variables
• Sometimes autocorrelation is caused by omitted variables.
consumption
Income
**
**
*
1
2
3 45
We forget to include wealth in our model
In year one (obs. 1) wealth goes up drastically big positive error
The effect of the increase in wealth in year 1 lingers for 3 years
Errors form a pattern
We should include wealth in our model
27
3. Examine the data
• Any systematic error in the collection or recording of data may result in autocorrelation.
28
After you make adjustments 1, 2 and 3
• Test for autocorrelation again• If autocorrelation is still a problem then
suspect pure autocorrelation– Follow the Cochrane-Orcutt procedure– Say what?????
29
Suppose our model is Yt = β0 + β1 Xt + єt (1)
And the error terms in Equation 1 are correlated
Let’s lag Equation 1
Yt-1 = β0 + β1 Xt-1 + єt-1 (4)
Where ut is not auto-correlated. Rearranging 2 we get 3
єt - ρ є t-1 = ut (3)
єt = ρ є t-1 + ut (2)
30
Now multiply Equation 4 by ρ
ρ Yt-1 = ρ β0 + ρ β1 Xt-1 + ρ єt-1 (5)
Now subtract 5 from 1 to get 6
Yt = β0 + β1 Xt + єt
- ρ Yt-1 = - ( ρ β0 + ρ β1 Xt-1 + ρ єt-1)
___________________________________
Yt - ρ Yt-1 = β0 - ρ β0 + β1 Xt - ρ β1 Xt-1 + єt - ρ єt-1 (6)
Note that the last two terms in Equation 6 are equal to Ut
So 6 becomes
Yt - ρ Yt-1 = β0 - ρ β0 + β1 (Xt - ρ Xt-1 ) + ut (7)
Yt - ρ Yt-1 = β0 - ρ β0 + β1 (Xt - ρ Xt-1 ) + ut (7)
• What is so special about the error term in Equation 7?It is not auto-correlated
• So, instead of equation 1 we can estimate equation 7
31
Define Zt = Yt – ρYt-1 & Wt = Xt – ρXt-1
Then 7 becomes
Zt = M + β1 Wt + ut (8)
Where M is a constant = β0 (1- ρ)
Notice that the slope coefficient of Equation 8 is the same as the slope coefficient of our original equation 1.
32
The Cochrane-Orcutt Method:So our job will be
Step 1: Apply OLS to the original model (Equation 1) and find the residuals et
Step 2: Use ets to estimate Equation 2 and find ρ^ (Note: this equation does not have an intercept.)
Step 3: Multiply ρ^ by Yt-1 and Xt-1 & find Zt & Wt
Step 4: Estimate Equation 8
33
• Luckily• EViews does this (steps 1- 4) automatically• All you need to do is to add AR(1) to the set
of your independent variables.• The estimated coefficient of AR(1) is ρ^• Let’s apply this procedure to Asst 22
34
Dependent Variable: SDH
Variable Coefficient Std. Error t-Statistic Prob.
C 8.83 2.50 3.520.0020
SDL 0.97 0.04 22.180.0000
USD -0.005 0.008 -0.600.5553
NR 0.002 0.0002 9.300.0000
R-squared 0.996792 Adjusted R-squared 0.996334
Durbin-Watson stat 1.076364
Dependent Variable: SDH
Variable Coefficient Std. Error t-StatisticProb.
C -9.11 8.4 -1.08 0.2940SDL 1.38 0.17 8.10 0.0000USD 6.71E-05 0.013 0.005
0.9959NR 0.0005 0.0004 1.46 0.1608
AR(1) 0.82 0.10 8.002 0.0000
R-squared 0.997927 Adjusted R-squared 0.997490 Durbin-Watson stat 2.463339What is this?
It is ρ^
What happened to standard errors as we corrected for serial correlation?They went upPositive autocorrelation standard error
Return and discuss Asst 21Use the data set FISH in Chapter 8 (P 274) torun the following regression equation:F = f (PF, PB, Yd, P, N)1) Conduct all 3 tests of imperfect
multicollinearity problem and report your results.
2) If you find an evidence for imperfect multicollinearity problem, suggest and implement a reasonable solution.
35
36
Correlation Matrix
F P PB PF YD NF 1 0.58 0.82 0.85 0.79 0.74P 1 0.66 0.73 0.78 0.57PB 1 0.96 0.82 0.78PF 1 0.92 0.88YD 1 0.93N 1
First testPF is more correlated with PB than with F PF is a problemYd is more correlated with PB and PF than with F Yd is a problemN is more correlated with PB, PF and Yd than with F N is a problemPB is more correlated with PF than with F PB is a problemP is more correlated with everything else than with F P is a problem
37
Correlation Matrix
F P PB PF YD NF 1 0.58 0.82 0.85 0.79 0.74P 1 0.66 0.73 0.78 0.57PB 1 0.96 0.82 0.78PF 1 0.92 0.88YD 1 0.93N 1
Second test:problem areas:
PF and PBPF and YdPF and NPB and YdYd and N
Note: F being highly correlated with independent variables is a good thing not a bad thing
38
Test 3
• Need 5 regression equations1. PF = f (P, Yd, PB, N)2. P = f (PF, Yd, PB, N)3. Yd = f (P, PF, PB, N)4. PB = f (PF, Yd, P, N)5. N = f (PF, Yd, PB, P)• For all find R2 then find VIF• For all VIF>5 Each independent variable is
highly correlated with the rest
39
Solutions1. Increase sample size
– Note: we want at least a df= 30, we have df=19
2. Do we have an irrelevant variable?– Seth argued N is not needed?– What is N? (P 273)– Seth, what was your argument?
3. Generate a new variable that measures the ratio of prices– Makes sense but doesn’t solve the high correlation between
Yd and N– Note: make sure your transformed variable makes sense
• That is the estimated coefficient has a meaning that people can understand
– The ratio PF/Yd makes no sense