Ans 1.

)
A.) Graphical analysis and descriptive stats:

1.) SPOT returns: Variable name = SPOT

i) Graphical analysis:
SPOT
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
250

ii)

500

750

1000

1250

1500

1750

2000

Descriptive Statistics

500

Series: SPOT
Sample 1 2171
Observations 2171

400

300

200

100

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

14101.57
13278.35
36542.50
5826.150
7071.339
1.260478
4.322716

Jarque-Bera
Probability

733.1466
0.000000

0
8000 12000 16000 20000 24000 28000 32000 36000

2.) Future returns : Variable name = close_price

i)

Graphical analysis:

Close Price
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
250

ii)

500

750

1000

1250

1500

1750

2000

Descriptive Statistics:

500

Series: CLOSE_PRICE
Sample 1 2171
Observations 2171

400

300

200

100

0
5000

B.) Run a regression:

Dependent Variable: SPOT

10000

15000

20000

25000

30000

35000

Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis

14105.95
12998.00
38570.00
5444.000
7254.491
1.317467
4.420478

Jarque-Bera
Probability

810.5643
0.000000

Method: Least Squares

Date: 12/30/13 Time: 12:36
Sample: 1 2171
Included observations: 2171
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
CLOSE_PRICE

368.1838
0.973588

16.22785
0.001023

22.68839
951.5976

0.0000
0.0000

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.997610
0.997609
345.7472
2.59E+08
-15770.55
905538.0
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

14101.57
7071.339
14.53021
14.53545
14.53213
0.412105

C.) Check presence of heteroscedasticity and non-stationarity:

i.)

SPOT:
a.) LEVEL Unit root test

Null Hypothesis: SPOT has a unit root

Exogenous: Constant
Lag Length: 1 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

2.718054
-3.433168
-2.862671
-2.567418

1.0000

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(SPOT)
Method: Least Squares
Date: 12/30/13 Time: 12:43
Sample (adjusted): 3 2171
Included observations: 2169 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

SPOT(-1)
D(SPOT(-1))
C

0.001333
0.179913
-7.864034

0.000491
0.021219
7.703043

2.718054
8.478701
-1.020900

0.0066
0.0000
0.3074

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood

0.037729
0.036840
160.3433
55687802
-14088.88

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.

13.36517
163.3810
12.99389
13.00175
12.99677

F-statistic
Prob(F-statistic)

42.46241
0.000000

Durbin-Watson stat

2.015346

st

1 difference unit root test:

Null Hypothesis: D(SPOT) has a unit root

Exogenous: Constant
Lag Length: 0 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

-38.51651
-3.433168
-2.862671
-2.567418

0.0000

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(SPOT,2)
Method: Least Squares
Date: 12/30/13 Time: 12:54
Sample (adjusted): 3 2171
Included observations: 2169 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

D(SPOT(-1))
C

-0.814146
10.84950

0.021138
3.459794

-38.51651
3.135881

0.0000
0.0017

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.406386
0.406112
160.5795
55877742
-14092.57
1483.521
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

-0.170562
208.3712
12.99638
13.00162
12.99829
2.018313

b.) White test - heteroscedasticity

Heteroskedasticity Test: White

F-statistic
Obs*R-squared
Scaled explained SS

192.6396
327.5949
546.6771

Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 12/30/13 Time: 12:52
Sample: 1 2171

Prob. F(2,2168)
Prob. Chi-Square(2)
Prob. Chi-Square(2)

0.0000
0.0000
0.0000

Included observations: 2171

Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
CLOSE_PRICE
CLOSE_PRICE^2

99597.08
-7.016397
0.000472

20151.40
2.463782
6.33E-05

4.942439
-2.847815
7.460464

0.0000
0.0044
0.0000

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.150896
0.150113
201377.1
8.79E+13
-29593.30
192.6396
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

119431.0
218438.6
27.26513
27.27298
27.26800
0.877100

(ii) Close_price:
a.) LEVEL Unit root test

Null Hypothesis: CLOSE_PRICE has a unit root

Exogenous: Constant
Lag Length: 0 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

1.935573
-3.433166
-2.862670
-2.567418

0.9999

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(CLOSE_PRICE)
Method: Least Squares
Date: 12/30/13 Time: 12:56
Sample (adjusted): 2 2171
Included observations: 2170 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

CLOSE_PRICE(-1)
C

0.001544
-7.830548

0.000798
12.63899

1.935573
-0.619555

0.0530
0.5356

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.001725
0.001265
268.9176
1.57E+08
-15217.95
3.746442
0.053050

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

13.93226
269.0878
14.02761
14.03285
14.02952
1.892016

1st difference unit root test:

Null Hypothesis: D(CLOSE_PRICE) has a unit root

Exogenous: Constant
Lag Length: 0 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

-43.95284
-3.433168
-2.862671
-2.567418

0.0001

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(CLOSE_PRICE,2)
Method: Least Squares
Date: 12/30/13 Time: 13:02
Sample (adjusted): 3 2171
Included observations: 2169 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

D(CLOSE_PRICE(-1))
C

-0.945579
13.18580

0.021513
5.780142

-43.95284
2.281224

0.0000
0.0226

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.471315
0.471071
268.8101
1.57E+08
-15210.07
1931.852
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

-0.408483
369.6129
14.02681
14.03205
14.02873
1.993044

D.) If yes, then run the same regression by applying the proper method:
Generating new series lnspot = log(spot) and lnfut = log(close_price)
Running unit root test on lnspot and lnfut to check if it is stationary or not.
a.) Lnfut
i.)

Unit root test

Null Hypothesis: LNFUT has a unit root

Exogenous: Constant
Lag Length: 0 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level

t-Statistic

Prob.*

0.318761
-3.433166

0.9793

5% level
10% level

-2.862670
-2.567418

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(LNFUT)
Method: Least Squares
Date: 12/30/13 Time: 13:07
Sample (adjusted): 2 2171
Included observations: 2170 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

LNFUT(-1)
C

0.000253
-0.001641

0.000795
0.007512

0.318761
-0.218494

0.7499
0.8271

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.000047
-0.000414
0.017643
0.674808
5683.157
0.101608
0.749939

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

0.000750
0.017639
-5.236089
-5.230852
-5.234174
1.895254

1st difference.
Null Hypothesis: D(LNFUT) has a unit root
Exogenous: Constant
Lag Length: 0 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

-44.16466
-3.433168
-2.862671
-2.567418

0.0001

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(LNFUT,2)
Method: Least Squares
Date: 12/30/13 Time: 13:12
Sample (adjusted): 3 2171
Included observations: 2169 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

D(LNFUT(-1))
C

-0.947847
0.000715

0.021462
0.000379

-44.16466
1.887275

0.0000
0.0593

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid

0.473712
0.473469
0.017622
0.672917

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion

-7.49E-06
0.024285
-5.238434
-5.233195

Log likelihood
F-statistic
Prob(F-statistic)

5683.082
1950.517
0.000000

Hannan-Quinn criter.
Durbin-Watson stat

-5.236518
1.997740

b.) Lnspot
i.)
Unit root test

Null Hypothesis: LNSPOT has a unit root

Exogenous: Constant
Lag Length: 2 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

0.575021
-3.433169
-2.862672
-2.567418

0.9891

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(LNSPOT)
Method: Least Squares
Date: 12/30/13 Time: 13:17
Sample (adjusted): 4 2171
Included observations: 2168 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

LNSPOT(-1)
D(LNSPOT(-1))
D(LNSPOT(-2))
C

0.000275
0.196479
0.068948
-0.002056

0.000479
0.021462
0.021467
0.004524

0.575021
9.154929
3.211817
-0.454492

0.5653
0.0000
0.0013
0.6495

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.049552
0.048234
0.010468
0.237108
6810.687
37.60668
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

0.000742
0.010730
-6.279231
-6.268749
-6.275398
2.006007

1st difference unit root test

Null Hypothesis: D(LNSPOT) has a unit root
Exogenous: Constant
Lag Length: 1 (Automatic based on SIC, MAXLAG=25)
t-Statistic

Prob.*

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

-27.24667
-3.433169
-2.862672
-2.567418

0.0000

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(LNSPOT,2)
Method: Least Squares
Date: 12/30/13 Time: 13:17
Sample (adjusted): 4 2171
Included observations: 2168 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

D(LNSPOT(-1))
D(LNSPOT(-1),2)
C

-0.733556
-0.069492
0.000542

0.026923
0.021443
0.000226

-27.24667
-3.240783
2.401562

0.0000
0.0012
0.0164

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.397044
0.396487
0.010466
0.237145
6810.521
712.8207
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

-9.43E-06
0.013472
-6.280001
-6.272139
-6.277126
2.006155

Generating new series dlnspot = D(lnspot) and dlnfut = D(lnfut)

Running unit root test on lnspot and lnfut to check if it is stationary or not.
a.) dlnspot
i.)

Unit root test Level

Null Hypothesis: DLNSPOT has a unit root

Exogenous: Constant
Lag Length: 1 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level
*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(DLNSPOT)
Method: Least Squares
Date: 12/30/13 Time: 13:23

t-Statistic

Prob.*

-27.24667
-3.433169
-2.862672
-2.567418

0.0000

Sample (adjusted): 4 2171

Included observations: 2168 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

DLNSPOT(-1)
D(DLNSPOT(-1))
C

-0.733556
-0.069492
0.000542

0.026923
0.021443
0.000226

-27.24667
-3.240783
2.401562

0.0000
0.0012
0.0164

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.397044
0.396487
0.010466
0.237145
6810.521
712.8207
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

-9.43E-06
0.013472
-6.280001
-6.272139
-6.277126
2.006155

b.) dlnfut
i.) Unit root test Level

Null Hypothesis: DLNFUT has a unit root

Exogenous: Constant
Lag Length: 0 (Automatic based on SIC, MAXLAG=25)

Augmented Dickey-Fuller test statistic

Test critical values:
1% level
5% level
10% level

t-Statistic

Prob.*

-44.16466
-3.433168
-2.862671
-2.567418

0.0001

*MacKinnon (1996) one-sided p-values.

Augmented Dickey-Fuller Test Equation

Dependent Variable: D(DLNFUT)
Method: Least Squares
Date: 12/30/13 Time: 13:22
Sample (adjusted): 3 2171
Included observations: 2169 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

DLNFUT(-1)
C

-0.947847
0.000715

0.021462
0.000379

-44.16466
1.887275

0.0000
0.0593

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.473712
0.473469
0.017622
0.672917
5683.082
1950.517
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

-7.49E-06
0.024285
-5.238434
-5.233195
-5.236518
1.997740

Running regression

Dependent Variable: DLNSPOT

Method: Least Squares
Date: 12/30/13 Time: 13:25
Sample (adjusted): 2 2171
Included observations: 2170 after adjustments
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
DLNFUT

0.000508
0.315769

0.000197
0.011162

2.577789
28.29033

0.0100
0.0000

R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
F-statistic
Prob(F-statistic)

0.269626
0.269289
0.009169
0.182273
7103.338
800.3428
0.000000

Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
Hannan-Quinn criter.
Durbin-Watson stat

0.000745
0.010727
-6.545012
-6.539774
-6.543097
2.252265