ECS4863/201/0/2024

Tutorial letter 201/0/2024

Advanced Econometrics

ECS4863 (Honours)

Year Module

MARKING GUIDE

Department of Economics

Answers to Assignment 01
 ECS4863/201

Question 1: (11 marks)

1.1 Give your view on whether econometrics should be studied separately from
economics. (2)

Econometrics may be defined as the quantitative analysis of actual economic phenomena based on
the concurrent development of theory and observation, related by appropriate methods of inference
[1 mark]

This implies that the economic theory underlying your modelling remains as important as its
statistical properties [1 mark]

However also important is that econometric techniques are constantly being improved and new
methods developed [1 mark]

[max 2 marks]

1.2 Explain in your words how you test serial correlation with strictly exogenous
variables (3)

(i) Run the OLS regression of yt on xt1, …, xtk and obtain the OLS residuals, 𝑢 ̂𝑡
, for all t = 1, 2, …, n……… [1 mark]
(ii) Run the regression of 𝑢 ̂ 𝑡 on 𝑢̂ 𝑡−1 , for all t = 2, …, n obtaining the coefficient 𝜌 on 𝑢
̂ 𝑡−1
and its t statistic, TR. (This regression may or may not contain an intercept; the t statistic for
r ˆ will be slightly affected, but it is asymptotically valid either way.) [1 mark]
(iii) Use TR to test H0: 𝜌 = 0 against H1: r ≠ 0 in the usual way. (Actually, since 𝜌=0 is often
expected a priori, the alternative can be H1: 𝜌= 0.) Typically, we conclude that serial
correlation is a problem to be dealt with only if H0 is rejected at the 5% level. As always, it is
best to report the p-value for the test. [1 mark]

[ Students should explain these steps]

1.3 Explain in your own words, the concept of serial correlation and implications for
inferences in econometrics (2)

Serial Correlation happens when error terms from different (usually adjacent) time periods
(or cross-section observations) are correlated, we say that the error term is serially
correlated Serial correlation occurs in time-series studies when the errors associated with a
given time period carry over into future time periods. (1 mark)
or
Conditional on X, the errors in two different time periods are correlated when Corr(ut,us) ≠ 0,
for all t ≠ s. (1 mark)
Properties of OLS with serially correlated errors
OLS is still unbiased and consistent if errors are serially correlated. (0.5 marks)
The correctness of R-squared also does not depend on serial correlation (0.5 marks)
OLS standard errors and tests will be invalid if there is a serial correlation (0.5 marks)
OLS will not be efficient anymore if there is a serial correlation (0.5 marks)
[ max 1 mark for any two properties)

Open Rubric
 ECS4863/201

1.4 Comment on the following theorem and assumption underlying the method of OLS:
(2)
(a) Gauss Markov theorem

Under Time series assumptions 1 through 5, the OLS estimators are the best linear unbiased
estimators conditional on X. This theorem is to ensure, OLS estimates are BLUE amongst a
set of other estimators (1 mark)

(b) Linear in parameters

The stochastic process {(xt1, xt2, …, xtk, yt ): t = 1, 2, …, n} follows the linear model yt = b0+
b1xt1 + b2xt2 + … + bkxtk + ut , where {ut: t =1, 2, …, n} is the sequence of errors or
disturbances. Here, n is the number of observations or time periods). This suggests that
parameters to be estimated should not have a power exceeding one whereas explanatory
variables can have a power greater than one 1 mark)

1.5 Explain in your own words what is meant by the following: (2)

(a) Covariance stationary process

A stochastic process {xt : t = 1, 2, …} with a finite second moment [E(xt2) < ∞] is covariance
stationary if (i) E(xt ) is constant; (ii) Var(xt ) is constant; and (iii) for any t, h ≥ 1, Cov(xt , xt+h)
depends only on h and not on t. (1 mark)
(b) Dynamically complete model

A time series model where no further lags of either the dependent variable or the
explanatory variables help to explain the mean of the dependent variable. (1 mark)
or
Written in terms of yt, E(yt ,xt , yt-1, xt-1, …) = E(yt ,xt ). In other words, whatever is in xt ,
enough lags have been included so that further lags of y and the explanatory variables do
not matter for explaining yt (1 mark)

[ Students should explain the above in their own words]

Question 2: (3 marks)

In this question you need to gather and analyze time series data for a country (other than
South Africa!) (3)

Select any country which starts with the same letter as your surname (if you cannot
find one, use the first letter of your name)
Now choose any macroeconomic variable from that country (e.g. inflation, GDP,
imports/exports, etc.)
Data source: you can use any data source (e.g. World Bank, IMF, country specific central
banks, etc.)

Your time series must have at least 60 observations.

You may use any interval, (e.g. quarterly, monthly)
It can be nominal or real data.

Make sure to provide (at least) the following:

- Your surname (or name)
- The name of the country and time series you chose.
3
 ECS4863/201
- A graph of the data
- A stationarity test

Below is an example of what is expected in your answer:

The name of the time series chosen (name plus ‘KBP’ code)

Total loan debt of national government: Total gross loan debt (KBP4114M). Source is the South
African Reserve Bank: https://www.resbank.co.za/

A graph of the data

GDEBT
4,000,000
3,500,000
3,000,000
2,500,000
2,000,000
1,500,000
1,000,000
500,000
0
00 02 04 06 08 10 12 14 16 18 20

(Please allocate 1 mark for either graph)

A graph of the log of the data (only where applicable!)

LGDEBT
15.2

14.8

14.4

14.0

13.6

13.2

12.8
00 02 04 06 08 10 12 14 16 18 20

Stationarity test

Series Model ADF ADF

Lags   
 
LGDEBT(0.5 mark) Trend 0 -1.608932
and
intercept
Intercept 0 3.086562
None 0 11.59674
D(LGDEBT) (0.5 mark) Trend 0 -14.23463***
and
intercept
Intercept 0 -13.66813***

4
 ECS4863/201

When tested in levels, we could not reject the null hypothesis (0.5 mark)
Only when differenced once could the null hypothesis be rejected. Therefore we conclude that
GDEBT is I(1) (0.5 mark)

Question 3: (36 marks)

Please give each student an additional 6 marks

Dataset 1 (2000M1 to 2023M12)

In this question, you must estimate a time series model for producer prices in South Africa.
You are provided with the following data:

Variable names and description:

CPI = Headline consumer inflation, all urban areas (Index)

NEER = Nominal effective exchange rate Exchange rates . NB increase implies
appreciation
OIL =Brent crude oil in South African rand per barrel
Dum =1 from 2008M1 to end of sample and zero otherwise. Assumes it captures period
of global financial crisis

DATASETS: For this question you must use the correct dataset, depending on the first letter of your
surname.

DATASET 1: Students whose surname start with A-L;

DATASET 2: Students whose surname start with M-N;
DATASET 3: Students whose surname start with O-Z.

Please make sure to download the correct datafile (Excel file), depending on your surname.
The Excel files are clearly marked ‘DATASET 1’, ‘DATASET 2’ and ‘DATASET 3’. You will be
panelised for not adhering to thus rule.

3.0 Please provide the following information:

My Surname is:
DATASET used: (indicate number)

3.1 Provide a description of the data (type, interval, etc.) (2)

Monthly, time series data (1 mark)

For the period January 2000 to December 2023 (1 mark)

5
 ECS4863/201
3.2 Use the data to calculate both the consumer price inflation and exchange rate growth
in South Africa. Use the month-on-same-month of previous year method to calculate the
inflation rates and growth. Plot both series in a scatterplot and comment on
relationship. Perform ganger causality test and interpret the results. Find the correlation
coefficient between the series and determine if its statistically significant (5)

Formula: Inflation = (CPI/CPI (-12) – 1) x 100 or 100*( log (CPI/CPI(-12)))

Scatterplot of CPI and NEER 12-month growth rates

16

12

8 (1 mark)
LCP _12

-4
-40 -30 -20 -10 0 10 20 30 40

LNEER_12

A negative relationship, indicating a NEER appreciation lowers the consumer price index (1 mark)

Granger causality test of CPI and NEER 12-month growth rates

Hypothesis 1: LCPI12 does not Granger Cause LNEER12 (0.5 marks)

Hypothesis 2: LNEER12 does not Granger Cause LCPI12 (0.5 marks)

Decision. NEER changes Granger causes consumer price inflation and it is statistically
significant (0.5 mark)

6
 ECS4863/201

Correlation of CPI and NEER 12-month growth rates

The correlation coefficient is significant at 1% and negative (0.5 marks)

3.3 Test the variables LCPI and LNEER for stationarity (transform all applicable variables).
Also comment on their respective orders of integration. Provide your results in the table
below (please add rows where necessary): (5)

‘L’ indicates the logarithmic function is used.

Variable Model Lags ADF test statistic

   

LCPI Trend and Intercept 1 -2.084607

Intercept 1 -0.490985 (1 mark)
None 1 8.038566
LNEER Trend and Intercept 1 -3.038764
Intercept 1 -1.262610 (1 mark)
None 1 -1.594252
DLCPI Trend and Intercept 0 -11.58264***
Intercept 0 -11.56569***(1 mark)
None 3 -3.457285***
DLNEER Trend and Intercept 0 -13.39098***
Intercept 0 -13.41359***(1 mark)
None 0 -13.29744***

Statistically significant at the: 10% level (*), 5% level (**), 1% level (***)

[1 mark per variable or ‘row’ in the above table]

CPI is non-stationary, integrated of order 1, or I(1) (0.5 mark )

NEER is non-stationary, integrated of order 1, or I(1) and becomes (0.5 mark )

3.4 Test for possible cointegration between variables:

(i) Estimate the following long run cointegration equation using OLS. Copy and paste
your EViews results window in your answer sheet. Test the hypothesis of full
exchange rate passthrough to consumer price inflation i.e 𝜷𝟐 = 𝟏 (2)
7
 ECS4863/201

𝑳𝑪𝑷𝑰𝒕 = 𝜷𝟎 + 𝜷𝟐 𝑳𝑵𝑬𝑬𝑹𝒕 + 𝜷𝟑 𝑳𝟎𝑰𝑳𝒕 + 𝝁𝒕

Testing null hypothesis of full passthrough (0.5 marks)

Decision: Reject null hypothesis at 1%, 5% and 10% which indicates that there is no full
passthrough (0.5 marks)

(ii) What are your a priori expectations in terms of sign? Interpret the estimated
coefficients. (2)

𝜷𝟐 < 𝟎, which suggests that NEER appreciation reduces consumer price index
(0.5 marks)
𝜷𝟑 > 𝟎 which suggests that an increase in oil prices raises the consumer price index
(0.5 marks)

LNEER -- implies 1 percent NEER appreciation leads to 0.748081 percent decline in

consumer price index. The impact is not statistically significant (0.5 marks)

LOIL ---- implies 1 percent Oil price increase raises consumer price index by
0.214465 percent. The impact is not statistically significant (0.5 marks)

8
 ECS4863/201

(iii) Generate the residual series for your long run equation and determine if
cointegration exists between the variables in the long run model. (2)

Null hypothesis : Residuals have a unit root. (0.5 marks)

Decision. Reject null hypothesis at 5% (0.5 marks)

This suggests that residuals are stationary indicating cointegration (0.5 marks)

3.5 Construct a short run (Error Correction) component for your model

∆𝑳𝑪𝑷𝑰𝒕 = 𝜷𝟎 + 𝜷𝟏 ∆𝑳𝑪𝑷𝑰𝒕−𝟏 + 𝜷𝟐 ∆𝑳𝑵𝑬𝑬𝑹𝒕 + 𝜷𝟑 ∆𝑳𝟎𝑰𝑳𝒕 +

𝜷𝟒 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒕−𝟏 + 𝜷𝟓 𝑫𝑼𝑴 ∗ 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒕−𝟏 + 𝝐𝒕

(i) Estimate the short run cointegration equation. Copy and paste your EViews
results window in your answer sheet. (1)

9
 ECS4863/201
(ii) Interpret the coefficients 𝜷𝟏 to 𝜷𝟓 of the short run model. (5)

DLSA_CPI(-1)

A 1 percentage point increase in previous consumer price inflation raises current

consumer price inflation by 0.204195 percentage points (0.5 marks)

The impact is statistically significant (0.5 marks)

DLNEER

A 1 percentage point increase NEER lowers current consumer price inflation by -0.00018
percentage points (0.5 marks)

The impact is not statistically significant (0.5 marks)

DLOIL

A 1 percentage point in oil prices raises current consumer price inflation by 0.008610
percentage points (0.5 marks)

The impact is statistically significant (0.5 marks)

DUMRESIDS(-1)

A positive sign implies the rate of correcting towards equilibrium in post-2008 is slower by
1.3354 % per month compared to the period before 2008 (0.5 marks)

The impact is statistically significant (0.5 marks)

RESID01(-1)

A negative sign implies disequilibrium in consumer price inflation is corrected at a rate of

2.4646% in each month before 2008 (0.5 marks)

The impact is statistically significant (0.5 marks)

(iii) Determine the speed of error correction from equilibrium since the global
recession in January 2008. Test whether it is different from zero (3)
The speed of error correction is given by 𝛽4 + 𝛽5 = −0.011293. This suggests the
disequilibrium is corrected at a rate of 1,1293 % per month post-2008 (1 mark)

Estimation Equation:
=========================
DLSA_CPI = B(1)*DLSA_CPI(-1) + B(2)*DLNEER + B(3)*DLOIL
+ B(4)*DUMRESIDS + B(5)*RESID01(-1) + C(6)

10
 ECS4863/201

Decision: Reject the Null Hypothesis that B(4) + B(5)=0 at 1%. The speed of correction is
statistically significant using t-statistics, F-statistics and Chi-square statistic (1 mark)

(iv) Explain in point format how you will determine whether positive and negative oil
price inflation have differential effects on consumer price inflation (3)

Generate the two series based on partial decompositions of the negative and positive
changes in oil price inflation. (1 mark)

Enter two series in the consumer price inflation equation. (1 mark)

Apply the Wald test to determine whether their impacts on consumer price inflation are
the same (1 mark)

[Total = 50 marks]

Question 3: (36 marks)

Please give each student an additional 6 marks

Dataset 2 (2000M1 to 2019m12)

3.0 Please provide the following information:

My Surname is:
DATASET used: (indicate number)

3.1 Provide a description of the data (type, interval, etc.) (2)

Monthly, time series data (1 mark)

For the period January 2000 to December 2019 (1 mark)

3.2 Use the data to calculate both the consumer price inflation and exchange rate growth
in South Africa. Use the month-on-same-month of previous year method to calculate
the inflation rates and growth. Plot both series in a scatterplot and comment on
relationship. Perform granger causality test and interpret the results. Find the
11
 ECS4863/201
correlation coefficient between the series and determine if its statistically significant (5)

Formula: Inflation = (CPI/CPI (-12) – 1) x 100 or 100*( log (CPI/CPI(-12)))

Scatterplot of CPI and NEER 12-month growth rates

16

12
(1 mark)

8
LCPI_12

-4
-40 -30 -20 -10 0 10 20 30 40

LNEER_12

A negative relationship, indicating a NEER appreciation lowers the consumer price

inflation (1 mark)

Granger causality test of CPI and NEER 12-month growth rates

Hypothesis 1: LCPI12 does not Granger Cause LNEER12 (0.5 marks)

Hypothesis 2: LNEER12 does not Granger Cause LCPI12 (0.5 marks)

Decision . NEER Granger causes consumer price inflation and it is statistically significant
(0.5 mark)

12
 ECS4863/201

Correlation of CPI and NEER 12-month growth rates

The correlation coefficient is significant at 1% and negative (0.5 marks)

3.3 Test the variables LCPI and LNEER for stationarity (transform all applicable variables).
Also comment on their respective orders of integration. Provide your results in the table
below (please add rows where necessary): (5)

‘L’ indicates the logarithmic function is used.

Variable Model Lags ADF test statistic

   

LCPI Trend and Intercept 1 -0.685281

Intercept 1 -1,799742 (1 mark)
None 1 7,110146
LNEER Trend and Intercept 1 -2,612709
Intercept 1 -1.442695 (1 mark)
None 1 -1,392906
DLCPI Trend and Intercept 0 -10,25821***
Intercept 0 -10,25219***(1 mark)
None 2 -3,753151***
DLNEER Trend and Intercept 0 -12,20604***
Intercept 0 -12.22749***(1 mark)
None 0 -12,13801***

Statistically significant at the: 10% level (*), 5% level (**), 1% level (***)

[1 marks per variable or ‘row’ in the above table]

CPI and NEER: non-stationary, integrated of order 1, or I(1) (0.5 mark each)

NEER non-stationary, integrated of order 1, or I(1) and becomes (0.5 mark each)

3.4 Test for possible cointegration between variables:

(iv) Estimate the following long run cointegration equation using OLS. Copy and paste
your EViews results window in your answer sheet. Test the hypothesis of full
exchange rate passthrough to consumer price inflation i.e 𝜷𝟐 = 𝟏 (2)

𝑳𝑪𝑷𝑰𝒕 = 𝜷𝟎 + 𝜷𝟐 𝑳𝑵𝑬𝑬𝑹𝒕 + 𝜷𝟑 𝑳𝟎𝑰𝑳𝒕 + 𝝁𝒕

13
 ECS4863/201

Testing null hypothesis of full passthrough (0.5 marks)

( 0.5 marks )

Decision: Reject null hypothesis at 1%, 5% and 10% which indicates that there is no full
passthrough (0.5 marks)

(v) What are your a priori expectations in terms of sign? Interpret the estimated
coefficients. (2)

𝜷𝟐 < 𝟎, which suggests that exchange rate appreciation reduces consumer price index
(0.5 marks)
𝜷𝟑 > 𝟎, which suggests that an increase in oil prices raises the consumer price index
(0.5 marks)

LNEER-- 1 percent NEER appreciation leads to -0.679678 percent decline in consumer price
index . The impact is statistically significant (0.5 marks)

LOIL-- 1 percent Oil price increase raises consumer price index by 0.219643 percent. The
impact is statistically significant (0.5 marks)

14
 ECS4863/201

(vi) Generate the residual series for your long run equation and determine if
cointegration exists between the variables in the long run model. (2)

Null hypothesis: Residuals have a unit root. ……………(0.5 marks)

Decision. Reject null hypothesis at 1%....................(0.5 marks)

This suggests that residuals are stationary indicating cointegration…….(0.5 marks)

3.5 Construct a short run (Error Correction) component for your model

∆𝑳𝑪𝑷𝑰𝒕 = 𝜷𝟎 + 𝜷𝟏 ∆𝑳𝑪𝑷𝑰𝒕−𝟏 + 𝜷𝟐 ∆𝑳𝑵𝑬𝑬𝑹𝒕 + 𝜷𝟑 ∆𝑳𝟎𝑰𝑳𝒕

+ 𝜷𝟒 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒕−𝟏 + 𝜷𝟓 𝑫𝑼𝑴 ∗ 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒕−𝟏 + 𝝐𝒕

(i) Estimate the short run cointegration equation. Copy and paste your EViews results
window in your answer sheet. (1)

Interpret the coefficients 𝛽1 to 𝛽5 of the short-run model. (5)

DLSA_CPI(-1)

A 1 percentage point increase in previous consumer price inflation raises current consumer
price inflation by 0.178214 percentage points (0.5 marks)

The impact is statistically significant (0.5 marks)

(1 mark)
15
 ECS4863/201
DLNEER

A 1 percentage point increase in NEER raises current consumer price inflation by 0.002898
percentage points (0.5 marks)

The impact is not statistically significant (0.5 marks)

DLOIL

A 1 percentage point in oil prices raises current consumer price inflation by 0.009936
percentage points (0.5 marks)

The impact is statistically significant (0.5 marks)

DUMRESIDS2019(-1)

A positive sign implies the rate of correcting towards equilibrium in post-2008 is slower by
0.015803 % per month compared to the period before 2008 (0.5 marks)

The impact is statistically significant (0.5 marks)

RESID012019M12(-1)

A negative sign implies disequilibrium in consumer price inflation is corrected at a

rate of 2.7257% in each month before 2008 (0.5 marks)

The impact is statistically significant (0.5 marks)

(ii) Determine the speed of error correction from equilibrium since the global
recession in January 2008. Test whether it is different from zero (3)

The speed of error correction is given by 𝛽4 + 𝛽5 = −0.010561, suggesting that

disequilibrium is corrected at a rate of 1,1293 percent per month (1 mark)

Estimation Equation:
=========================
DLSA_CPI = B(1)*DLSA_CPI(-1) + B(2)*DLNEER + B(3)*DLOIL + B(4)*DUMRESIDS2019(-1) +
B(5)*RESID012019M12(-1) + B(6)

16
 ECS4863/201

Reject the Null Hypothesis that C(4) + C(5)=0 at 1%. The speed of correction is statistically
significant using t-statistics, F-statistics and Chi-square statistic (1 mark)

(iii) Explain in point format how you will determine whether positive and negative oil
price inflation have differential effects on consumer price inflation (3)

Generate the two series based on partial decompositions of the negative and positive
changes in oil price inflation. (1 mark)

Enter two series in the consumer price inflation equation. (1 mark)

Apply the Wald test to determine whether their impacts on consumer price inflation are
the same (1 mark)

[Total = 50 marks]

Question 3: (36 marks)

Please give each student an additional 6 marks

Dataset 3 (2005M1 to 2023m12)

3.0 Please provide the following information:

My Surname is:
DATASET used: (indicate number)

3.1 Provide a description of the data (type, interval, etc.) (2)

Monthly, time series data (1 mark)

For the period January 2005 to December 2023 (1 mark)

3.2 Use the data to calculate both the consumer price inflation and exchange rate growth
in South Africa. Use the month-on-same-month of previous year method to calculate
the inflation rates and growth. Plot both series in a scatterplot and comment on

17
 ECS4863/201
relationship. Perform granger causality test and interpret the results. Find the correlation
coefficient between the series and determine if it's statistically significant (5)

Formula: Inflation = (CPI/CPI (-12) – 1) x 100 or 100*( log (CPI/CPI(-12)))

Scatterplot of CPI and NEER 12-month growth rates

12

10

8 (1 mark)
LCPI_12

0
-40 -30 -20 -10 0 10 20 30

LNEER_12

A negative relationship, indicating a NEER appreciation lowers the consumer price inflation
(1 mark)

Granger causality test of CPI and NEER 12-month growth rates

Hypothesis 1: LCPI12 does not Granger Cause LNEER12 (0.5 marks)

Hypothesis 2: LNEER12 does not Granger Cause LCPI12 (0.5 marks)

(0.5 marks)

Decision. NEER does not Granger cause consumer price inflation and it’s not statistically
significant (0.5 mark)

18
 ECS4863/201

Correlation of CPI and NEER 12-month growth rates

(0.5 marks)

The correlation coefficient is significant at 1% and negative (0.5 marks)

3.3 Test the variables LCPI and LNEER for stationarity (transform all applicable variables).
Also comment on their respective orders of integration. Provide your results in the
table below (please add rows where necessary): (5)

‘L’ indicates the logarithmic function is used.

Variable Model Lags ADF test statistic

   

LCPI Trend and Intercept 2 -3.268632

Intercept 13 -0.595323 (1 mark)
None 1 8.297440
LNEER Trend and Intercept 1 -3.142412
Intercept 1 -1.365301 (1 mark)
None 1 -1.712332
DLCPI Trend and Intercept 1 -10.29351***
Intercept 1 -10.28590***(1 mark)
None 12 -0.601057
DLNEER Trend and Intercept 0 -12.13796***
Intercept 0 -12.15702***(1 mark)
None 0 -12.00238***

Statistically significant at the: 10% level (*), 5% level (**), 1% level (***)

[1 mark per variable or ‘row’ in the above table]

CPI is non-stationary, integrated of order 1, or I(1) in two instances (0.5 mark each)

NEER is non-stationary, integrated of order 1, or I(1) and becomes (0.5 mark each)

19
 ECS4863/201
3.4 Test for possible cointegration between variables:

(vii) Estimate the following long run cointegration equation using OLS. Copy and paste
your EViews results window in your answer sheet. Test the hypothesis of full
exchange rate passthrough to consumer price inflation i.e 𝜷𝟐 = 𝟏 (2)

𝑳𝑪𝑷𝑰𝒕 = 𝜷𝟎 + 𝜷𝟐 𝑳𝑵𝑬𝑬𝑹𝒕 + 𝜷𝟑 𝑳𝟎𝑰𝑳𝒕 + 𝝁𝒕

( 0.5 marks )

Testing null hypothesis of full passthrough (0.5 marks)

( 0.5 marks )

Decision : Reject null hypothesis at 10% significance level which indicates that there is no
full passthrough (0.5 marks)

20
 ECS4863/201

(viii) What are your a priori expectations in terms of sign? Interpret the estimated
coefficients. (2)

𝜷𝟐 < 𝟎, which suggests that exchange rate appreciation reduces consumer price
index (0.5 marks)
𝜷𝟑 > 𝟎, which suggests that an increase in oil prices raises the consumer price index
(0.5 marks)

LNEER-- implies 1 percent NEER appreciation leads to -0.828394 percent decline in

consumer price index. The impact is statistically significant (0.5 marks)

LOIL-- implies 1 percent Oil price increase raises consumer price index by 0.161325
percent The impact is statistically significant (0.5 marks)

(ix) Generate the residual series for your long run equation and determine if
cointegration exists between the variables in the long run model. (2)

Null hypothesis: Residuals have a unit root. 0.5 marks)

Decision. Reject null hypothesis at 10% (0.5 marks)

This suggests that residuals are stationary at 10% significance level indicating
cointegration relationship (0.5 marks)

3.5 Construct a short run (Error Correction) component for your model

∆𝑳𝑪𝑷𝑰𝒕 = 𝜷𝟎 + 𝜷𝟏 ∆𝑳𝑪𝑷𝑰𝒕−𝟏 + 𝜷𝟐 ∆𝑳𝑵𝑬𝑬𝑹𝒕 + 𝜷𝟑 ∆𝑳𝟎𝑰𝑳𝒕

+ 𝜷𝟒 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒕−𝟏 + 𝜷𝟓 𝑫𝑼𝑴 ∗ 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒕−𝟏 + 𝝐𝒕

(x) Estimate the short run cointegration equation. Copy and paste your EViews
results window in your answer sheet. (1)

21
 ECS4863/201

Interpret the coefficients 𝜷𝟏 to 𝜷𝟓 of the short-run model. (5)

DLSA_CPI(-1)

A 1 percentage point increase in previous consumer price inflation raises current consumer
price inflation by 0.229414 percentage points (0.5 marks)

The impact is not statistically significant (0.5 marks)

DLNEER

A 1 percentage point increase in NEER raises current consumer price inflation by 0.001530
percentage points (0.5 marks)

The impact is statistically significant (0.5 marks)

DLOIL

A 1 percentage point increase in oil prices raises current consumer price inflation by
0.009936 percentage points (0.5 marks)

The impact is statistically significant (0.5 marks)

DUMRESIDS2005(-1)

A positive sign implies the rate of correcting towards equilibrium in post-2008 is slower by
1,2424% per month compared to the period before 2008 (0.5 marks)

The impact is not statistically significant (0.5 marks)

22
 ECS4863/201

RESID012005M12(-1)

A negative sign implies disequilibrium in consumer price inflation is corrected at a rate of

2,2985% in each month before 2008 (0.5 marks)

(iv) Determine the speed of error correction from equilibrium since the global
recession in January 2008. Test whether it is different from zero (3)

The speed of error correction is given by 𝛽4 + 𝛽5 = −0.008823.The disequilibrium is

corrected at a rate of 1,1293 percent per month (1 mark)

Estimation Equation:
=========================
DLSA_CPI = B(1)*DLSA_CPI(-1) + B(2)*DLNEER + B(3)*DLOIL + B(4)*DUMRESIDS + B(5)*RESID01(-1) +
C(6)

Reject the Null Hypothesis that B(4) + B(5)=0 at 1%. The speed of correction is statistical
significant using t-statistics, F-statistics and Chi-square statistic (1 mark)

(v) Explain in point format how you will determine whether positive and negative oil
price inflation have differential effects on consumer price inflation (3)

Generate the two series based on partial decompositions of the negative and positive
changes in oil price inflation. 1 mark)

Enter two series in the consumer price inflation. Equation (1 mark)

Apply the Wald test to determine whether their impacts on consumer price inflation
are the same (1 mark)

[Total = 50 marks]

23