Báo cáo hết môn kinh tế lượng

HANOI UNIVERSITY
Faculty of Management and Tourism

ECONMETRICS

GROUP PROJECT

UNEMPLOYMENT RATE IN
VIETNAM (1986-2015)
Tutor:

Mr Pham Van Hung

Tut number:

1

Group:

1

Students:

Đặng Thị Thanh Hương – 1304000034
Ngô Thị Hải Yến - 1204010122
Văn Thúy Diễm – 1304010011
Vũ Thị Thu Liên - 1204010050


TABLE OF CONTENTS

I.

II.

INTRODUCTION
Unemployment is one of the major problems of developed as well as developing
countries today. Unemployment affects not only person who has lost job but also
affects many aspect of the society (family, living standard, economic growth…)
It can make individual suffer bad emotion such as stress, sadness, confusion…All
these emotions can break them down. Moreover, it can go as far as affecting the
society. Without income, firstly, their family will directly hurt from this situation
because of education of children, food, condition of living.... Secondly, when
unemployment increase, it means that the society did not use efficiently source
of labor, lead to reducing in productivity. Besides, unemployment also influences
government budget. Government has to pay more for allowance, social welfare
and no income as well as decreasing income tax. Therefore, government always
tends to decrease unemployment rate. That is reason why our group choose the
topic of unemployment in Vietnam with three main factors: GDP, inflation and
population. The period we chose from 2000-2015 to see how Vietnamese
economy has changed before and after crisis 2008-2009 up to now. The model
has used available data of GDP, inflation and population to reflect more truthful
about estimated value of unemployment rate in Vietnam.
METHOLODOGY
1. Variables
According to the theory of macroeconomic, GDP and unemployment rate are
used to measure the living standard and they have negative relationship. It
means that an increase in GDP lead to decrease in unemployment rate. In
addition,in economic, Okun’lawis an empirically observed relationship
between unemployment and GDP. The "gap version" states that for every 1%

increase in the unemployment rate, a country's GDP will be roughly an
additional 2% lower than its potential GDP. Therefore, we decide to test
whether there is an actual existence between unemployment rate and GDP.
Population is also one of significant factors that affect to the change in
unemployment rate. Many reports indicate that when the number of population
of a country increases significantly, it will create the pressure about
employment problem. It means that number of people not being find out a job
will rise directly, as a result, unemployment rate also grow up significantly.
The final variable which we selected to support for this report, is inflation
variable. According to Phillip’s curve, in the long run period, a change in
inflation has not affected to the change in unemployment rate and it was a line.
However, in the short run period, unemployment rate and inflation have related
because of the change in demand.
2. Sample size
In the report, we choose 16 years from 2000 to 2015 as sample size.


3. Collection data
We choose the time series data from 2000 to 2015 to illustrate the changes
in unemployment rate are explained by changes in other economic
variables, namely, GDP, population, and Inflation rate. All of the data are
recorded annually. All data are collected from General Statistic Office.
Table 1. Unemployment rate and other factors
Year
200
0
200
1
200
2

200
3
200
4
200
5
200
6
200
7
200
8
200
9
201
0
201
1
201
2
201
3
201
4
201
5

III.

Unemployment rate

(%)

GDP(mil USD)

Population
(mil)

Inflation rate
(%)

6.42

31,200

77.63

-1.768

6.28

32,487

78.62

-0.31

6.01

35,081

79.54

4.079

5.78

39,798

80.47

3.303

5.6

45,359

81.44

7.895

5.31

52,899

82.39

8.394

4.82

60,819

83.31

7.503

4.64

71,003

84.22

8.349

2.38

89,553

85.1187

23.12

2.9

91,533

86.025

6.71

2.88

101,623

86.9474

9.2

2.22

135,540

87.8604

18.67

1.96

155,820

88.8093

9.1

2.18

171,220

89.7595

6.59

2.1

186,200

90.7289

4.1

2.31
198,638.16
91.903961
0.63
Sources: General Statistic Office and IMF data (2000-2015).

EQUATION REGRESSION
1. Functional form


It is very necessary to test for the best model illustrating the relationship between
Unemployment and the explanatory variables. We regress all the functional
models, as well as linear model to choose a least coefficient of variance (CV).
Before establishing the model, we assume that our model response to the ten
classical assumptions (The Gaussian, standard or classical linear regression model
(CLRM)):
(1) Linear regression model
(2) X values are fixed in repeated sampling
(3) Zero mean values of disturbance : E(ui\Xi) = 0
(4) Homoscedasticity or equal variance of ui: var(ui\Xi) = σ2

(5) No autocorrelation between the disturbances: cov(ui, uj\ Xi, Xj) = 0
(6) Zero covariance between ui and Xi: cov (ui,Xi) = 0
(7) The number of observations n must be greater than the number of parameters
to be estimated.
(8) Variability in X values
(9) The regression model is correctly specified.
(10) There is no perfect multicollinearity.
Our group run OLS regression on Eviews with different functional forms and then
compute coefficient of variation (C.V.) to find the most suitable functional form with
lowest C.V.


Equation semi-Log model has the smallest CV, as a result this equation will be used to
indicate the relationship between the UN and GDP, POP, INF.
The final result is :
Ln(UE^)= 7.559700- 1.637896* ln(GDP)+ 0.144371* POP- 0.008729* INF
•
•
•

UE^: Expected unemployment rate.
GPD: Gross domestic product.
POP: Population rate.
INF: Inflation rate.
2. Model interpretation
 On average, if GDP increases by 1 unit, UE will decrease by about 1.637896
million, other things remain constant.
 On average, if POPULATION increases by 1 unit (million people), UE will
rise by about 0.144371 million, other things remain constant.
 On average, if INF increases by 1 unit (million USD), UE will decrease by

about 0.008729 million, other things remain constant.
 Confidence interval: α = 5% => tα/2, n-k = t0.025, 12 = 2.179







IV.

RESULT ANALYSIS
1. Overall significant test
Step 1:

H0: β2 = β3 = β4 = 0
H1: β22+β32 β42 ≠ 0

Step 2:

(k: Number of parameters)
= 107.1504

Step 3:

With α = 5%, Fc = Fα,df,df2 = F0.05,3,12 = 3.49

Step 4:

Reject H0 if F-value > Fc

Step 5:

Compare 107.1504 > 3.49 => Reject H0

Step 6:

There is enough statistically evidence to conclude that at

least one of these variables: Log (GDP), Population, Inflation has been
significant.
2. Individual significant test
Step 1:

H0: βi = 0
H1: βi ≠ 0

Step 2:
Step 3:

α = 5%, tc = t α/2, n-k = t0.025, 12 = 2.179


Reject H0 if |t-value| > tc

Step 4:

parameter Log(GDP)
s
|t-value|

4.164961
decision
Reject
We have table

Population

Inflation

2.538879
Reject

1.962671
Not reject

There is enough information about the significant effect of log(GDP) and
Population or log(Un) otherwise Inflation. Because inflation has negative
relationship with unemployment in short term. Hence, in long term, economist
should estimate based on log(GDP) and population.
3. Chow test
During period 2000 – 2015, there is an economic crisis over the world so this
sample is divided two periods: period 1 (2000 -2007) and period 2 (2008 – 2015).
Each period has 8 observations. Two new equations will be presented below:
Step 1:

Ho: no structural change
H1: yes

•

Restricted model:


•

Unrestricted model:

Step2:

(k: number of period)
=


Step 3:

Fα,df1,df2 = F0.05,2,12 = 3.89

Step 4:

Reject Ho if F – value > Fc

Step 5:

Since 61.11873 > 3.89 => Reject Ho

There is enough statistically evidence to infer that there should have a structural
change between Log unemployment, Log GDP, Population and Inflation. It can be
explained that in 2008, Vietnam applied the first stimulus package to encourage
the demand to overcome this depression period.
4. Errors

a.Multicollinearity
 Because R2 and individual significant test
Since R2 = 96.1013% > 90% and inflation rate variable is not
significant, multicollinearity might exist.
 VIF

From the table, centered VIF of log(GDP) and pop are greater than 10,
which result in existence of multicollinearity. Nevertheless, two above
variables have less likely collinearity with inflation rate because VIF of
the inflation is only 1.245108.
 Correlation Matrix


The result from above table might be conflicted with the prediction of
VIF, which dedicates that there is no multicollinearity among independent
variables.
 Auxiliary Regression

Step 1:
Step 2:

Run OLS on auxiliary Regression
(k is number of parameters in original
model)

Step 3:

Fc = Fα,df1,df2 = F0.05, 2,13 = 3.81

Step 4:

Reject H0: Xi is collinearity with Xf & Xm if F-

value > Fc
We have 3 independent variables corresponding with 3
auxiliary regressions.

Auxiliary regression

Log(GDP)

POP

INF


F-value
Fc
Decision

652.128
3.81
Reject

637.192
3.81
Reject

1.5932
3.81

Not reject

In conclusion, there is enough evidence to infer that there is multicolinearity
between log(GDP) & Pop, otherwise Inf.
b.
Heteroscedasity
Step1:

H0: homoscedasticity,Var (ui) = σ2
H1: heteroscedasticity, Var (ui) = σi2

Step2:

Run OLS estimation on original model => obtain

Step3:

Obtain R2 from Auxiliary Regression R2 = 0.872554

We have W = n.R2 = 16 x 0.872554 = 13.960864


Step4:

Look at χ2α,df (df: number of regressors to auxiliary regression)
χ20.05,9 = 16.919

Step 5:

If W >χ2α,df=. Reject H0

Since 13.960864 < 16.919 => Not reject H0
There is not enough evidence to infer that heteroscedasticity exists.
c.Autocorrelation
 First-order correlation
Step 1:
H0: ρ = 0, no autocorrelation
H1: ρ> 0, yes, positive autocorrelation
Step 2:
Durbin-Watson test : DW = 1.455488
Step 3:
Critical value α = 5%, n = 16, k’ = 4 – 3 = 1
dL= 1.728
du= 0.857
Step 4:
Since du< DW <dL => there is not enough information
that first-order autocorrelation happens or not.
 Higher-order correlation
Step 1:
H0: no higher-order autocorrelation
H1: yes
Step 2:
Run OLS estimation, obtain R2


R2 = 0.129469
Step 3:
Breusch-Godfrey test = n.R2 = 16 x 0.129469 =
2.071504
Step 4:
Critical value: χ2α,ρ = χ20.05,2 = 5.99147

Step 5:
Reject H0 if BG test >χ2α,ρ
Since 2.071504 < 5.99147 => Not reject H0
There is no existence of higher-order autocorrelation (with the number
of lag years is 2)
We continue the number of lag years from 3 to 10 but the results are
still the same. Finally, there is only first-order autocorrelation.
5. Limitation
In our process, the data is collected from trustworthy website, however, other
sources also provided the same period but difference data. Besides, the data is
estimated in specific sample, it’s also challenging when GDP, Inflation and
population are measured by the general Department of Statistic. Therefore, the
data may not reflect all exactly data of economy. In addition, our sample size
is still small(16 years). Furthermore, we used EViews software and knowledge
which we are learnt in econometrics to perform test, interpret results to get


V.

conclusion. So the report might have some errors, nevertheless, we had done
our best to finish this project.
REFERENCE

1.

An Ngoc, 2015, “GDP năm 2015 tăng 6,68%, caonhấttrong5năm”, available at
/>
2.

Source: IMF, Available at />

3.

Source IMF, Available at />
4.

Source:Tong cuc thong ke, Available at

5.

/>usp=sharing

6.

Uyen Minh, 2015, “Tỷlệthấtnghiệp 2015 tănglên 2.31%, Available at
/>
7.

Vu Minh, 2015, “KinhtếViệt Nam: 20 nămthăngtrầm qua cácchỉsố”, Available at
/>