You are on page 1of 27

National Economics University

ASSIGNMENT OF ECONOMETRICS
TOPIC: The Relationship between Economic Growth and School Enrollment

BFI 63 – GROUP 4
Lecturer:
Nguyen Hai Duong
Student Name & ID:
Pham Huu Manh - 11213747
Nguyen Thi Kim Ngan - 11214216
Do Bao Ngoc - 11214288
Nguyen Mai Vinh - 11216262
Dang Thi Ha Mi - 11219454
Nguyen Thanh Vinh - 11216267
Cao Thi Phuong Uyen - 11219463
Nguyen Phuong Linh – 11213313
ABSTRACT:.............................................................................................................3
INTRODUCTION:...................................................................................................4
A. PART 1:...............................................................................................................5
I. Basic Theory......................................................................................................5
II. Review:.............................................................................................................6
1. Research 1:.....................................................................................................6
2. Research 2:......................................................................................................7
3. Research gap:..................................................................................................8
B. PART 2:...............................................................................................................9
I. Method and data:..............................................................................................9
II. Evaluate the functions of data:......................................................................9
1. Linear function................................................................................................9
2. Logarithm function:......................................................................................10
3. Semi Logarithm function:.............................................................................11
4. Quadratic function.........................................................................................12
III. Test of economical assumptions.................................................................13
1. Linear function:.............................................................................................13
2. Logarithm function........................................................................................13
3. Semi Logarithm function..............................................................................14
4. Quadratic function.........................................................................................15
IV. Testing for the most efficient model:..........................................................16
1. RAMSEY TEST..........................................................................................16
a. Linear function:..........................................................................................16
b. Logarithm function....................................................................................16
c. Semi Logarithm function...........................................................................17
d. Quadratic function......................................................................................17
2. WHITE TEST..............................................................................................18
a. Linear function:..........................................................................................18
b. Logarithm function:...................................................................................18

2
c. Semi Logarithm function:..........................................................................19
d. Quadratic function:....................................................................................19
3. Most selection criteria:...............................................................................20
4. Interpret estimation of the most efficient..................................................20
C. PART 3..............................................................................................................21
APPENDIX............................................................................................................24

3
ABSTRACT:
This study aims to evaluate the factors that the influence of education on economic
growth in different countries around the world. The model is estimated using the
method of Ordinary Least Squares (OLS), employing data for the year 2018
available on WorldBank, and using the econometric tools of Eview12. The sample
size consisted of 142 countries. The evaluation identified that economic growth be
affected significant by education, including school enrollments: primary level,
secondary level, tertiary level. In particular, the estimated elasticities are
statistically remarkable and have the expected signs, and their values are in
accordance with the empirical literature.

INTRODUCTION:
The activities of getting or acquiring general knowledge, learning process of basics
skills such as mathematics, geography and also developing elementary
understanding of some other subjects e.g., history, natural sciences, social sciences,
and art, developing reasoning and judgmental mental power, and preparing oneself
or others intellectually for mature life. School enrollment, primary, secondary and
tertiary (% gross) has been taken as a proxy for primary, secondary and tertiary
education, respectively (Loening, 2005). The relationship between education and
the GDP is positive that shows education is a significant primary input factor for
the growth of an economy. Barro (1991) argued that there is significant and
positive association between economic growth and the education. Bils and Klenow
(2000) argued that high enrollment rate causes rapid improvement in productivity;
therefore, faster growth in per capita income (PCI) resulted in countries where
there is high rate of enrollment in schools. Hanushek and Kimko (2000) argued

4
that there is remarkable increase in productivity and national growth rates due to
the quality of the education.

This thesis aims to analyze the relationship between education and economic
growth includes 142 countries in different continents. With the following studies
will make accurate assertions about the important influence of education on the
development of countries.

5
A. PART 1:
I. Basic Theory
1. Input – output model:

GDPP = f (PRM, SEC, TER)


+ Output: GDPP: Gross Domestic Product Per Capita

+ Input : PRM: Gross School Enrollment Ratio Primary Level (%)

SEC: Gross School Enrollment Ratio Secondary Level (%)

TER: Gross School Enrollment Ratio Tertiary Level (%)

II. Review:
1. Research 1:
 Research name: The Relationship between Economic Growth and School
Enrollment Rates: Time Series  Evidence from Turkey
 Author: Sedat Gumus, Selim Kayhan
 Date: 2012
a) Purpose of research: to explore the relationship between economic
growth and educational attainment in Turkey
b) Method: The variables of GDP per capita and gross school enrollment
ratios at the primary, secondary, and tertiary levels.
c) Sample size: in Turkey
d) Model:
(1)

6
(2)

(3)

e) Result and discussion:


- There is a closer relationship between educational attainment and
economic growth at the primary school level as compared to the
secondary and higher levels of education.
2. Research 2:
 Research name: The Influence of Education on Economic Growth

 Author: ŞTEFAN CRISTIAN CIUCU, RALUCA DRAGOESCU


 Date: May 2014
a) Purpose of research: to analyze the effect of education on economic
growth in  Bulgaria and Czech Republic, during the transition period
and to compare it to the situation in a  developed country the
Netherlands.
b) Method: the multiple regression model
c) Sample size: in all three countries (Bulgaria, Czech Republic and the
Netherlands)
d) Model:

7
e) Result and discussion:
- The most important thing to be taken into consideration is that
education quantity has an effect  on economic growth after some
years. The graduates have to start their career in order to affect the 
economy of a country. For this reason a time lag will be introduced in
the model. The time lag for  primary education will be set to 10 years
and the time lag for secondary education to 5 years
- Education has an influence on economic growth in both transition
and developed countries. The level of influence varies and depends on
other factors from country to country.
3. Research gap:
a) Observation

- Research 1: 1 country: Turkey quite small

- Research 2: focus on 3 countries Bulgaria, Czech Republic and the Netherlands

Quite small also

- Our research: includes 142 countries in different continents

b) Research purpose: 3 researchs have the same purpose, that is prove the impact
of eduction on economic growth. We find some limitations of 2 researchs before:

 Research 1: The study has some limitations. In Turkey, there have been
some significant policy changes in both the economic and educational
sectors. These policy changes, as well as international economic trends, may
have impacted both economic growth and school enrollment patterns. The
data does not make it possible to take factors
 Reasearch 2: The study has limitation too. The few data are used so the
results not too much convincing.
8
Therefore, our research aims to a diverse view considering influence of
education on economic growth in several developed and developing
countries around the world which is identified by the following factors:
gross school enrollment ratio in 3 levels: primary, secondary, tertiary.

c) Methodology

 Research 1: used Toda-Yamamoto’s (1995) causality test. It is the modified


Wald (MWALD) test developed by Toda and Yamamoto
 Research 2: regression model.
 Our research: OLS method

9
B. PART 2:
I. Method and data:
1. Method: OLS method

2. Data: The data is available online on World Development Indicators (WDI)


website – the World Bank's premier compilation of cross-country comparable data
on development.

II. Evaluate the functions of data:


1. Linear function
GDPP = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3TER+ 𝑢
From the EViews result in table 2.1 (APPENDIX), we have a new linear function:

GDPP = -1914.25600286 - 247.677037229*PRM + 502.349750612*SEC +

(23269.39) (223.6879) (113.1806)

20.0977613077*TER + 𝑢

(107.5432)

n = 142, R2 = 0.312214, SSR = 5.84E+10

 Interpret estimation results:


- 𝛽0 = -1914.25600286 has shown that when other independent
variables equal 0, GDPP will equal to -1914.25600286 USD on
average.
- 𝛽1 = -247.677037229 has shown that when gross school enrollment
primary level changes by 1%, GDPP will decrease by 247.677037229
USD on average (other remain constant).

10
- 𝛽2 = 502.349750612 has shown that when gross school enrollment
secondary level changes by 1%, GDPP will increase by
502.349750612 USD on average (other remain constant).
- 𝛽3 = 20.0977613077 has shown that when gross school enrollment
tertiary level changes by 1%, GDPP will increase by 20.0977613077
USD on average (other remain constant).

2. Logarithm function:

LOG(GDPP) = 𝛽0 + 𝛽1LOG(PRM) + 𝛽2LOG(SEC) + 𝛽3LOG(TER)+ 𝑢

From the EViews result in table 2.2 (APPENDIX), we have a new linear function:

LOG(GDPP) = 11.991738184 - 3.1631547027*LOG(PRM) +

(3.836473) (0.869451)

2.25339743724*LOG(SEC) + 0.470434628864*LOG(TER) + 𝑢

(0.402793) (0.172883)

n = 142, R2 = 0.329034, SSR = 5.70E+10

3. Semi Logarithm function:


LOG (GDPP) = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3TER+ 𝑢

From the EViews result in table 2.3 (APPENDIX), we have a new linear function:

LOG(GDPP) = 8.2224560203 - 0.0257129079552*PRM +

(0.815861) (0.007843)

0.0327100528963*SEC + 0.0113444825181*TER + 𝑢

11
(0.003968) (0.003771)

n = 142, R2 = 0.329034, SSR = 5.70E+10

 Interpret estimation results:


- 𝛽0 = 8.222 has shown that when other independent variables equal 0,
GDPP will equal to 8.222 USD on average.
- 𝛽1 = -0.0257 has shown that when gross school enrollment primary
level changes by 1%, GDPP will decrease by 2.57% on average (other
remain constant).
- 𝛽2 = 0.0327 has shown that when gross school enrollment secondary
level changes by 1%, GDPP will increase by 3.27% on average (other
remain constant).
- 𝛽3 = 0.0113 has shown that when gross school enrollment tertiary
level changes by 1%, GDPP will increase by 1.13% on average (other
remain constant).

4. Quadratic function
GDPP = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3SEC2 + 𝛽4TER+ 𝑢
From the EViews result in table 2.4 (APPENDIX), we have a new linear function:

GDPP = 16942.0173004 - 209.649882769*PRM - 82.505590253*SEC +

(25211.23) (222.6880) (334.9384)

3.26898451751*(SEC)^2 + 43.4247617404*TER

(1.763946) (107.3475)
n = 142, R2 = 0.329034, SSR = 5.70E+10
 Interpret estimation results:

12
- 𝛽0 = 16942.0173004 has shown that when other independent
variables equal 0, GDPP will equal to 16942.0173004 USD on
average.
- 𝛽1 = -209.649882769 has shown that when gross school enrollment
primary level changes by 1%, GDPP will decrease by 209.649882769
USD on average (other remain constant).
- 𝛽2 = -82.505590253 has shown that when gross school enrollment
secondary level changes by 1%, GDPP will decrease first by
82.505590253 USD on average.
- 𝛽3 = 3.26898451751 has shown that when gross school enrollment
secondary level changes by 1%^2units, GDPP will increase by
3.26898451751 USD on average (other remain constant).
- 𝛽4 = 43.4247617404 has shown that when gross school enrollment
tertiary level changes by 1%, GDPP will increase by 43.4247617404
USD on average (other remain constant).

III. Test of economical assumptions


1. Linear function:
GDPP = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3TER+ 𝑢
From the EViews result in table 2.1 (APPENDIX), we have:

{ H : β =0
a. Hypothesis pair: H 0 : β1 ≠ 0
1 1

 P – value = 0.2701
 α = 0.05
 Conclusion: - P – value > α

- Accept Ho

13
- GDPP not influenced by PRM

{ H : β =0
b. Hpothesis pair: H 0 : β3 ≠ 0
1 3

 P-value = 0.8520
 α = 0.05
 Conclusion: - P-value > α
- Accept H0
- GDPP not influenced by TER

2. Logarithm function
LOG(GDPP) = 𝛽0 + 𝛽1LOG(PRM) + 𝛽2LOG(SEC) + 𝛽3LOG(TER)+ 𝑢
From the EViews result in table 2.2 (APPENDIX), we have:

{ H : β =0
a. Hypothesis pair: H 0 : β1 ≠ 0
1 1

 P-value = 0.0004
 α = 0.05
 Conclusion: - P-value < α
- Reject H0, accept H1
- GDPP be influenced by PRM

{ H : β =0
b. Hypothesis pair: H 0 : β3 ≠ 0
1 3

 P-value = 0.0073
 α = 0.05
 Conclusion: - P-value < α
- Reject H0, accept H1
- GDPP be influenced by TER

14
3. Semi Logarithm function
LOG (GDPP) = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3TER+ 𝑢

From the EViews result in table 2.3 (APPENDIX), we have:

{ H 0 : β 1=0
a. Hypothesis pair: H : β ≠ 0
1 1

 P-value = 0.0013
 α = 0.05
 Conclusion: - P-value < α
- Reject H0, accept H1
- GDPP be influenced by PRM

{ H : β =0
b. Hypothesis pair: H 0 : β3 ≠ 0
1 3

 P-value = 0.0031
 α = 0.05
 Conclusion: - P-value < α
- Reject H0, accept H1
- GDPP be influenced by TER

4. Quadratic function
GDPP = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3SEC2 + 𝛽4TER+ 𝑢

From the EViews result in table 2.4 (APPENDIX), we have:

{ H : β =0
a. Hypothesis pair: H 0 : β1 ≠ 0
1 1

 P-value = 0.3481
 α = 0.05
 Conclusion: - P-value > α

15
- Accept H0
- GDPP not influenced by PRM

{ H 0 : β 3=0
b. Hypothesis pair: H : β ≠ 0
1 3

 P-value = 0.6865
 α = 0.05
 Conclusion: - P-value > α
- Accept H0
- GDPP not influenced by TER

IV. Testing for the most efficient model:


1. RAMSEY TEST
Following results, we want to know whether our models are unbiased or not,
therefore, we use Ramsey RESET test with 1 fitted values.
a. Linear function:

Hypothesis pair: { H 0 : β 4 =0
2
H 1 : β 4 ≠0

P-value = 0.0519
α = 0.05
=>P-value > α => We accept Ho, that means (1) has correct form or it is unbiased.

16
b. Logarithm function

Hypothesis pair: { H 0 : β 4 =0
H 1 : β 24 ≠0

P-value = 0.0162
α = 0.05
=>P-value < α => We reject Ho, that means (2) has incorrect form or it is biased.

c. Semi Logarithm function

Hypothesis pair: { H 0 : β 4 =0
2
H 1 : β 4 ≠0

P-value = 0.0604
α = 0.05
=>P-value > α => We accept Ho, that means (3) has correct form or it is unbiased.

d. Quadratic function

Hypothesis pair: { H 0 : β 4 =0
2
H 1 : β 4 ≠0

17
P-value = 0.0509
α = 0.05
=>P-value > α => We accept Ho, that means (4) has correct form or it is unbiased.

2. WHITE TEST
Following results, we want to know whether the variance of the errors in a
regression model is constant or not, therefore, we use WHITE test.
a. Linear function:

{ H :homoskedasticity
Hypothesis pair: H 0:heteroskedasticity
1

Heteroskedasticity Test: White


Null hypothesis: Homoskedasticity

F-statistic 2.606309     Prob. F(9,132) 0.0084


    Prob. Chi-
Obs*R-squared 21.42629 Square(9) 0.0109
    Prob. Chi-
Scaled explained SS 187.0351 Square(9) 0.0000

Test Equation:
Dependent Variable: RESID^2
Method: Least Squares

P-value = 0.0084
α = 0.05
=> P-value < α => We reject H0, that means (1) is heteroskedasticity.

18
b. Logarithm function:

{ H : Homoskedasticity
Hypothesis pair: H 0: Heteroskedasticity
1

 P – value = 0.1543
 α = 0.05
 Conclusion:
o P-value > α
o Accept Ho
o The model is Homoskedasticity.
c. Semi Logarithm function:

{ H : Homoskedasticity
Hypothesis pair: H 0: Heteroskedasticity
1

 P – value = 0.0097
 α = 0.05
 Conclusion:
o P-value < α
o Reject Ho
o The model (2) is Heteroskedasticity

19
d. Quadratic function:

{ H :homoskedasticity
Hypothesis pair: H 0:heteroskedasticity  
1

Heteroskedasticity Test: White


Null hypothesis: Homosckedasticity

F-statistic 3.187045     Prob. F(13,128) 0.0004


Obs*R-squared 34.72367     Prob. Chi-Square(13) 0.0009
Scaled explained SS 319.8923     Prob. Chi-Square(13) 0.0000

P-value = 0.0004
α = 0.05
=> P-value < α => We reject H0, that means (3) is heteroskedasticity.

3. Most selection criteria:


Because all three models are unbiased and heteroskedasticity, we use five criteria
to select the most efficient model.

Model R2 Adjusted R2 AIC HQ SC


(1) 0.312214 0.297262 22.72921 22.76304 22.81247
(2) 0.720987 0.714922 2.212395 2.246230 2.295658
(3) 0.329034 0.309444 22.71853 22.76082 22.82261
(table 3.1)
According to the table above, the most efficient model is model (2), semi logarithm
function.

4. Interpret estimation of the most efficient


LOG(GDPP) = 8.222 - 0.026*PRM + 0.033*SEC + 0.011*TER
(0.816) (0.008) (0.004) (0.004)

n = 142, R2 = 0.720987, in SSR = 71.80880


20
From model (2) - Semi Logarithm function:
β0 = 8.222 has shown that when other independent variables equal 0, GDPP will
equal to 8.222 USD.
β1 = 0.026 has shown that when PRM increases by 1%, keeping other independent
variables constant, on average, GDPP will decrease by 2.6%.
β2 = 0.033 has shown that when SEC increases 1%, keeping other independent
variables constant, on average, SEC will increase by 3.3%.
β3 = 0.011 has shown that when TER increases 1%, keeping other independent
variables constant, on average, TER will increase by 1.1%.
R2 = 0.720987 means 72.0987% of the GDPP is explained by variations of all the
independent variables in this regression model.

21
C. PART 3

COMMUNICATION
In this study, we tested and compared 4 models:

(1) Linear function: GDPP = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3TER+ 𝑢

(2) Logarithm function: LOG(GDPP) = 𝛽0 + 𝛽1LOG(PRM) +


𝛽2LOG(SEC) + 𝛽3LOG(TER)+ 𝑢

(3) Semi logarithm function: LOG (GDPP) = 𝛽0 + 𝛽1PRM + 𝛽2SEC +


𝛽3TER+

(4) Quadratic function: GDPP = 𝛽0 + 𝛽1PRM + 𝛽2SEC + 𝛽3SEC2 +


𝛽4TER+ 𝑢

to find the most efficient model to evaluate variables that influence economic
growth.

We employed the Ramsey RESET test to see whether our models were unbiased.
The result suggested that 3 models (1), (3), (4) had its correct form and model (4)
is biased. Next, the WHITE test was used to gauge whether the variance of the
errors in a regression model is constant. Since 3 models were found to be unbiased
and heteroskedasticity, we used 5 criteria that are R2, Adjusted R2, AIC, HQ, SC
to find out the most efficient model.

According to the result was showed at table 3.1, the most efficient model was
model (2) – semi logarithm function.

22
In the econometric model used, education in primary, secondary and tertiary have
impact on economic growth. This means that these countries should be adjust
teaching methods in every schools to suitable for the student’s learning style.

The general conclusion is that policy-makers have to draw and apply a education
policy employing diligence, timely planning, responsibility, and realism.

23
APPENDIX

Table. Data Collection

24
TABLE 2.1

25
TABLE 2.2

TABLE 2.3

26
TABLE 2.4

27

You might also like