See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320554262

The impact of macroeconomic factors on trade balance in Vietnam

Article · September 2017

CITATIONS READS

0 1,183

2 authors:

Hoang Phong Le Dang Van

University of Economics Ho Chi Minh City & Ho Chi Minh City University of Law University of Economics Ho Chi Minh City
29 PUBLICATIONS   782 CITATIONS    12 PUBLICATIONS   85 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Horizon 2020 Grant project: Institutions for Knowledge Intensive Development: Economic and Regulatory Aspects in South-East Asian Transition Economies View
project

Lecturer View project

All content following this page was uploaded by Hoang Phong Le on 18 December 2019.

The user has requested enhancement of the downloaded file.

 LE HOANG PHONG • DANG THI BACH VAN

The impact of macroeconomic

factors on trade balance in Vietnam
Le Hoang Phong(1) • Dang Thi Bach Van(2)

Received: 18 May 2017 | Revised: 12 August 2017 | Accepted: 20 August 2017

ABSTRACT: This study aims to analyze the impacts of macroeconomic

factors on trade balance in Vietnam during the period 1986-2014
with data collected from Key Indicators for Asia and the Pacific
2014 (ADB). Based on multivariable model, using Autoregressive
Distributed Lag (ARDL) model, we test for cointegration using bound
test and measure the long-run impacts of variables. In addition, we
use ECM-ARDL model to analyze the short-run impacts. The results
show that while short-run and long-run GDP and exchange rate
have the positive impacts on trade balance, money supply (M2) has
a positive impact in short-run and a negative impact in long-run.

KEYWORDS: trade balance, bound test, ARDL model, ECM model.

jel CLASSIFICATION: C15 . C22 . F13 . F14 . F17 . F32 . F41.

Le Hoang Phong - Email: [email protected].

Dang Thi Bach Van - Email: [email protected].

(1) Ho Chi Minh City University of Law;

02 Nguyen Tat Thanh Street, Ward 12, District 4, Ho Chi Minh City, Vietnam.
(2) University of Economics Ho Chi Minh City;
59C Nguyen Dinh Chieu, District 3, Ho Chi Minh City.

Volume 1: 1-148 | No.1, September 2017 | banking technology review 25

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

1. Introduction
Since the beginning of “Doi moi” in 1986, Vietnam has been implementing
a range of innovations under/following market orientation and international
economic integration in order to take advantage of economic development
opportunities. This is an essential precondition for Vietnam to achieve economic
growth and poverty reduction, bringing Vietnam from a low-income country to
lower middle-income one. During the period 1986-2011, Vietnam experienced
trade deficit owing to the high demands for foreign raw materials, equipment,
machines, and technologies. Meanwhile, there existed low capacity of domestic
production, limited funds and unbalance between exports and imports. From
2012, Vietnam began shifting trade deficit to a small-scale trade surplus. Figure
1 shows the overall trade balance in Vietnam during 1986-2014.

Figure 1: Trade balance in Vietnam during 1986-2014

4000000
Billion VND

3000000

2000000 Import
Export
1000000

0
1986
1988

1990
1992

1994
1996

1998
2000
2002

2004
2006

2008

2010

2012
2014

Year

Resource: Key Indicators for Asia and the Pacific 2014 (ADB).

According to Bui Trinh, Kiyoshi Kobayashi & Vu Trung Dien (2010), one of
the current macro instabilities in Vietnam is due to the prolonged trade deficit.
It puts pressure on Vietnam dong (VND), has negative impacts on exchange rate
and inflation; affects the balance of payment; reduces the international reserves that
could make a decrease in the effectiveness of exchange rate policy as well as the
market confidence; and leads to dollarization that puts more pressure on exchange
rate market. The high trade deficit also causes a large surplus in capital and financial
accounts, which results in the accumulation of national debt over time.
The control of trade deficits, as well as the selection of macro economic tools
for trade balance adjustment have positive effects on the economic stability
and growth in the future. Therefore, it is crucial to determine and estimate

26 banking technology review | No.1, September 2017 | Volume 1: 1-148

 LE HOANG PHONG • DANG THI BACH VAN

economic factors’ impacts on trade balance, so that policymakers can control the
fluctuation of trade balance in order to develop a suitable foreign trade policy,
supporting the economic growth in the context of the increasingly international
economic integration of Vietnam in the near future.
In recent years, there are a few studies about trade balance and factors
impacting on trade balance in Vietnam. Some notable ones are Ha Thi Thieu
Dao & Pham Thi Tuyet Trinh (2013), Phan Thanh Hoan & Nguyen Dang Hao
(2007), Tran Hong Ha (2011), Nguyen Huy Tuan (2011). However, these studies
mostly concern about one key question– the impacts of exchange rate on trade
balance.
Using ARDL model, this study adds an empirical evidence on the impacts
of macroeconomic factors on trade balance in Vietnam during the period
1986-2014. Based on the findings of this study, we recommend some solutions
to improve the trade balance in Vietnam in the future.

2. Literature review
Trade balance is the difference between the export and import values of an
economy in a period of time (Meade, 1951; Mundell, 1968). Statistic data used
to calculate trade balance is the monetary value of all products and services
exported and imported by a country (Polak, 1957). If a country has a positive
trade balance, it has a trade surplus, and in reverse, it has a trade deficit (Dash,
2003).
Trade balance is affected by numerous factors, such as exchange rate, money
supply, GDP (Duasa, 2007). The first one is the monetary approach to the balance
of payments, especially trade balance, developed by many economists such as
Mundell (1968), Magee (1976), Johnson (1976), Miles (1978), Frenkel & Razin
(1987), and so on. Trade balance is considered as a monetary phenomenon and
its imbalance is rooted in the relationship between money supply and demand. It
means that trade balance is affected by money supply and demand. If a country has
the supply of money exceeds the demand for money, the excessive money will flow
out of the country via exchange commodities, trade deficit increases as a result, and
then trade balance becomes worse. On the other hand, if domestic money demand
exceeds domestic money supply, international reserves are improved thanks to the
positive effect on the trade balance. In addition, Magee (1976) suggests that money
supply has an impact on trade balance. A decrease in the domestic money supply
will improve the trade balance, while an increase in domestic money supply will
have an adverse effect on trade balance (Magee, 1976).

Volume 1: 1-148 | No.1, September 2017 | banking technology review 27

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

The second approach to trade balance is to analyze the income effects

with the consideration that trade balance improvement requires an increase
in national income (Harberger, 1950; Meade, 1951; Alexander, 1952, 1959).
Within a country, an increase in GDP raises the domestic demand for money. In
order to rebalance the currency, the country must boost its export, which has a
positive impact on trade balance (Mundell, 1968). According to Solow growth
model, at the early stage, a less developed country which borrows foreign debt
for investments often achieve a faster economic growth rate. As a consequence,
when the economy is close to the steady-state, then the growth rate slows
down. Hence, a country with a high-growth rate is the one in the early stage of
development and tends to accept current account deficits (Solow, 1956).
The consideration of the impacts of exchange rates on the trade balance is
important and has been tested in numerous studies, such as Bickerdike (1920),
Robinson (1947), Metzler (1948), Hacker & Hatemi (2004), Chun, Chun & Chih
(2012). When exchange rate increases, domestic currency depreciates, foreign goods
are more expensive relative to domestic goods, imports decline, exports rise, trade
balance then improves (Bickerdike, 1920; Robinson, 1947; Metzler, 1948). The result
of Sugema (2005) for Indonesia in the period 1984-1997 supports the devaluation of
the exchange rate to improve the trade balance. Moreover, exchange rate also impacts
on trade balance under the J-curve effect (Ha Thi Thieu Dao et al., 2013).
Previous studies show that trade balance is affected by macroeconomic
factors such as money supply, GDP, exchange rate. A typical study is Duasa’s
(2007) using the ARDL model to test the effects of exchange rates, money supply
and GDP on the trade balance in Malaysia. The result indicates that there is a
long-term relationship among trade balance, GDP and money supply but no
relationship between trade balance and exchange rate in long-term.
In Vietnam, many researchers concern about the impacts of exchange rates on
the trade balance. But money supply is an important monetary factor that affects
trade balance. Income is a crucial factor to promote exports and imports and
affect trade balance. Therefore, we choose the following factors: money supply,
GDP and exchange rate as the variables to analyze impacts on the trade balance.

3. Data and methodology

3.1. Data
The data are collected from annual data in the period 1986-2014 and the Key
Indicators for Asia and the Pacific 2014 (ADB). The data contain gross domestic

28 banking technology review | No.1, September 2017 | Volume 1: 1-148

 LE HOANG PHONG • DANG THI BACH VAN

product (GDP), money supply (M2), exchange rate (E), trade balance variable
which is the ratio of export values (X) to import values (M). All variables take
natural logarithm form.
Table 1: Variables review
No Variable Variable definition
1 LGDP Natural logarithm of GDP
2 LM2 Natural logarithm of money supply M2
3 LE Natural logarithm of exchange rates
4 LTB Natural logarithm of trade balance variable (the ratio
of export values to import values)
3.2. Methodology
Firstly, we have to check the stationarity problem for time series data.
However, the drawback of stationarity testing is that the tests have low power.
With regards to this issue, the autoregressive distributed lag (ARDL) method
proposed by Pesaran, Shin & Smith (1996) has received lots of interests.
According to Pesaran et al. (1996), while it is necessary to implement the unit
root test in cointegration testing, it becomes unnecessary to do so in the ARDL
and the existence of the long-run relationship is tested based on two critical
value bounds. In particular, the lower bound is the critical value for all variables
which are cointegrated of order zero (I(0)) and the upper bound is for those
cointegrated of order one (I(1)).
The ARDL (p0, p1, p2, p3, ..., pn) model includes two parts: (i) The DL (distributed
lag) – lagged explanatory variables potentially affect dependent variables; (ii) The
AR (autoregressive) – dependent variable could also correlate to its previous values
(it lags). The notation (p0, p1, p2, p3, ..., pn) is the number of lags.
Assume Y is the dependent variable, the independent variables are X1, X2, ...,
Xn respectively. The ARDL (p0, p1, p2, p3, ..., pn) model could be rewritten as follows:
p0 p1
Yt = α + ∑ β 0,iYt −i + ∑ β1, j X 1,t − j
i =1 j =0

p p 2 p 3
(a) n

+ ∑ 2,k 2,t −k ∑ 3,l 3,t −l

β
X + β X
k =0
+ ... + ∑ n,m n,t −m t
β X
l =0
+ ε .
m =0

The procedure to conduct ARDL can be summarized as follows:

• Step 1: Run unit root tests
p to ensurep that no variable is integrated of order
0 1

∆Yt = α + ∑ β 0,i ∆Yt −i + ∑ β1, j ∆X 1,t − j

two. The reason is that if the i =1
variablesj =0are stationary after taking repeated
differences 2 times, wep yield spurious
2 p regressions.
3 p n

+ ∑ β 2,k ∆X 2,t − k + ∑ β3,l ∆X 3,t −l + ... + ∑ β n,m ∆X n ,t − m

k =0 l =0 m=0

+λ0Yt −1 + λ1 X 1,t −1 + λ2 X 2,t −1 + λ3 X 3,t −1 + ... + λn X n ,t −1 + ε t .

Volume 1: 1-148 | No.1, September 2017 | banking technology review 29
 The Impact of Macroeconomic Factors on trade Balance in Vietnam
p0 p1
Yt = α + ∑ β 0,iYt −i + ∑ β1, j X 1,t − j
• Step 2: The boundi =test:
1 According
j =0 to Pesaran & Pesaran (1997), the bound
test is built to test the existence of cointegrationp relationship between variables,
p 2 p 3 n

in other words, is∑

+ β 2,k X 2,t − k + ∑ β3,l X 3,t −l + ... + ∑ β n,m X n ,t − m + ε t .
kto
= 0 test the existence
l =0 of long-run m = 0 relationship between them. In

order to do the bound test, we consider the following equation:

p0 p1
∆Yt = α + ∑ β 0,i ∆Yt −i + ∑ β1, j ∆X 1,t − j
i =1 j =0
p2 p3 pn
(b)
+ β ∆X + β ∆X ∑
+ ... + β ∆X
k =0
2, k 2,t − k ∑l =0
3,l 3,t −l ∑
m=0
n, m n ,t − m

+λ0Yt −1 + λ1 X 1,t −1 + λ2 X 2,t −1 + λ3 X 3,t −1 + ... + λn X n ,t −1 + ε t .

The hypotheses for cointegration testing as follows:

Y = ϕ0 + ϕ1 X 1 + ϕ2 X 2 + ϕ3 X 3 + ... + ϕ n X n
- Null hypothesis H0: λ0 = λ1 = λ2 = λ3 = ... = λn = 0: the cointegration does not
exist, meaning no long-run relationship between variables.
p
- Alternative hypothesis H1: λ0 ≠ β1, j0, λ1 ≠ 0, λ2 ≠ 0, λ3 ≠ 0, ..., λn ≠ 0: the
1

α
∑
cointegration does ϕ0 = exist, meaning
; ϕ1 = j = 0 (some) ; long-run relationship(s) between
p p0
variables.
0

1 − ∑ β 0,i 1 − ∑ β 0,i
To test the null, we compare
i =1
the valuei =1
of computed F-statistic and the critical
p p p
values of two bounds,∑which β 2,k I(0) and I(1) are in accordance with the lower and
2 3 n

∑ β 3,l ∑ β n,m
upper bounds respectively.
ϕ2 = k =0
p 0
; ϕ3 = l =0
p
;...; ϕn = 0
m=0
p 0

∑
- If F-statistic is1 −higher
i =1
β 0,i than1 −the ∑i =1
βupper
0,i value 1− ∑ bound
i =1
β 0,i (in accordance with
I(1)), the null hypothesis is rejected. p Wep can conclude that the existence of
0 1

cointegration is valid. Yt = α + ∑ β 0,iYt −i + ∑ β1, j X 1,t − j

i =1 j =0
- If F-statistic is lowerppthan  pthe lower
0
 0
value bound (in accordance with I(0)),
∆Yt = α − ∑  ∑ β 0,i  × p∆Yt −( i −1) 
2 3 p n

+∑
the null hypothesis is accepted. β
i = 2 2, k i 2,t − k  ∑ 3,l 3,t −l
X We can
+ conclude
β X + that
... + ∑ n,misXno
there β cointegration.
n ,t − m + ε t .
- If F-statistic is in middle k =0
p of bounds,
 p 1  we cannot
l =0
1  draw a conclusion. An error
m =0

correction term +isβ1,0 ∆X 1,t − ∑to

employed ∑ β1, j  ×the
 identify ∆X 1,cointegration
t − ( j −1)  (Kremers, Ericsson &
j =2  pj  
Dolado (1992)). 0 p 1

• Step 3: Define ∆Yt = pα +∑ p

β 0,length:
i ∆Yt −i + The ∑ β1,choice
j ∆X 1,t − of
+ β 2,0the
∆X 2,optimal lag optimal lag length for
2 2

t − ∑  i∑ =1 β 2, k  × ∆X 2,
j
j =t0− ( k −1) 
all variables in our model p could be done
k = 2 
2
 k  p by choosing the  3 p minimum AIC and n

SBC criteria. + ∑ βp2,k∆Xp2,t − k +∑ β3,l ∆X 3,t −l + ... + ∑ β n,m ∆X n ,t − m

3 3

+ β3,0 ∆X 3,kt =−0 ∑  ∑ β3,l l× = 0∆X

3,t − ( l −1) 
m=0
• Step 4: Estimate the long-run
+λ0Ytl−=12+λ1 X l equation.  
1,t −1 + λ2 X 2,t −1 + λ3 X 3,t −1 + ... + λn X n ,t −1 + ε t .
The long-run equation can be written as follows:
.........................
pn
 pn  
+ β n ,0 ∆XYn ,= ϕ0∑
t − + ϕ1X∑ β × ∆X n ,t3−+( m...
1 + ϕn2, mX2 + ϕ3 X −1)+ϕ n X n (c)
m = 2  m  
p0
Of which, long-run
−(1 − ∑ βcoefficients
0,i ) ECM t −1 .
φ0, φ1, pφ2, φ3, ..., φn are defined as follows: 1
i =1
α
∑β
j =0
1, j

ϕ0 = p0
; ϕ1 = p0
;
30 banking technology review | No.1, β 0,i 1−
1 − September β 0,i
2017 | Volume 1: 1-148 ∑i =1
∑
i =1
p2 p3 pn
 +λ∆0Y
Ytt −= α + ∑ β 0,i ∆Yt −i + ∑ β1, j ∆X 1,t − j
1 + λ1 X 1,t −1 + λ2 X 2,t −1 + λ3 X 3,t −1 + ... + λn X n ,t −1 + ε t .
i =1 j =0
p2 p3 pn
+ ∑ β 2,k ∆X 2,t − k + ∑ β3,l ∆X 3,t −l + ... + ∑ β n,m ∆X n ,t − m
Y = kϕ=00 + ϕ1 X 1 + ϕ2 X 2 l+=0ϕ3 X 3 + ... + ϕLE m = 0 PHONG • DANG THI BACH VAN
HOANG
n Xn
+λ0Yt −1 + λ1 X 1,t −1 + λ2 X 2,t −1 + λ3 X 3,t −1 + ... + λn X n ,t −1 + ε t .

p1

Y = ϕ +αϕ X + ϕ
1, j ∑β
X +ϕ
X + ... + ϕ n X n
ϕ0 = 0 p 1 1; ϕ1 =2 2j =0p 0 3 3 ;
0

1 − ∑ β 0,i 1 − ∑ β 0,i
i =1 i =1
p1

∑∑
β1, j p2
(d)
p3 pn

∑ βα2,k j =β
0 3,l ∑ β n,m
ϕ = ;ϕ = ;
ϕ2 =0 k =0p p ; ϕ3 =1 l =0p p 0 ;...; ϕn = m =0p
0

1 − ∑∑ 1 − ∑∑
1 − β 0,i 0
1 − β 0,i 0 0

β
i =1 0,i
β
i =1 0,i
1 − ∑ β 0,i
i =1 i =1 i =1
p2 p3 pn

∑ 2,k coefficients
• Step 5: Estimate the short-run
β ∑ 3,l according ∑ ton,mthe unlimited error
β β
ϕ2 = k =0p p ;pϕ3 = l =0p ;...; ϕn = m =0p
correction model.  
β0,i ∑ β 0,1i − ×∑∆βY0,t −i( i −1) 
0 0 0 0

1 −−∑ 1 − ∑ β 0,i
0

(pi ∑
∆Yt = α
The ECM-ADRL with =i1=, 2p , p
0 1 i 2
, p ,
3
...,
i =1 pn
) lags in Englei =1 – Granger method
is written as follows: p 
 p 1  1 
+ β1,0 ∆X 1,t − ∑ p0  ∑
β1, j  × ∆X 1,t −( j −1) 
j = 2  j   
p0
∆Yt = α − ∑ ∑ β 0,i × ∆Yt −( i −1)  

i =p22   ip2   
+ β 2,0 ∆X 2,t − ∑p  ∑p β 2, k  × ∆X 2,t −( k −1) 
1   
+ β1,0 ∆X 1,t − ∑ ∑ β1, j × ∆X 1,t −( j −1)
1
k =2 k

p3j = 2  p3 j
   
+ β3,0 ∆X 3,t − ∑p ∑pβ3,l  × ∆X 3,t −( l −1) 
l = 2  l    
2 2

+ β 2,0 ∆X 2,t − ∑
 ∑ β 2, k  × ∆X 2,t −( k −1)  (e)
.........................
k = 2  k  
pn p3
pn p3  
+ β+nβ,03,0
∆X ∆Xn ,t 3,−
t −∑ ∑ ∑  ∑β β × ∆X ,t −t −( m( l −1) 
3,l × ∆Xn3,
n,m
m =l2= 2 m l  
p0
.........................
−(1 − ∑ β 0,i ) ECM t −1.
i =1
pn
 pn  
+ β n ,0 ∆X n ,t − ∑  ∑ β n ,m  × ∆X n ,t −( m −1) 
m = 2  m  
p0
−(1 − ∑ β 0,i ) ECM t −1.
i =1

Of which ECMt-1 is the corrected error, reflecting the adjustment speed to

long-run equilibrium.
To ensure that the ARDL is reliable and robust, we need to run relevant
diagnostic tests such as: Wald test, Ramsey’s RESET test for wrong model
identification, Larange multiplier test (LM) for serial correlation, heteroskedas-
ticity test, CUSUM (Cumulative sum of recursive residuals) and CUSUMSQ
(cumulative sum of square of recursive residuals) tests for the stability of residuals.
According to Pesaran et al. (1996) and Hamuda, Suliková, Gazda & Horváth
(2013), if we are unsure about the unit root’s property and/or the stationarity

Volume 1: 1-148 | No.1, September 2017 | banking technology review 31

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

of data, variables are not integrated of the same order I(0) or I(1), then ARDL
procedure is most appropriate for empirical studies. Besides, Pesaran et al.
(1996), Hamuda et al. (2013), Tran Nguyen Ngoc Anh Thu & Le Hoang Phong
(2014), Le Hoang Phong & Dang Thi Bach Van (2015) agree that ARDL method
has more advantages.
Firstly, ARDL model allows us to work with a small sample whereas the
Johansen’s technique requires a bigger sample in order to obtain significant
results.
Secondly, in contrast to conventional methods in finding the long-run
relationship, ARDL model does not estimate simultaneous equation. Instead, it
deals with the single equation.
Thirdly, while other methods only accept same lags for all variables in a
regression, ARDL tolerates different optimal lags.
Fourth, ARDL provides unbiased long-run estimation if there are endogenous
variables in the model.
Our ARDL model for this study can be represented as follows:
p1 p2 p3 p4
LTBt = α + ∑ β1,i LTBt −i + ∑ β 2, j LGDPt − j + ∑ β 3,k LM 2t − k + ∑ β 4,l LEt −l + ε t . (1)
i =1 j =0 k =0 l =0
p1 p2 p3 p4

According to t = α + ∑et
LTBPesaran β1,i LTB t −i + ∑ β 2,the
al. (1997), j LGDP t− j + ∑
bound β3,k LM
test ∑
2t − k +first
is the β 4,step + εt .
l LEt −l in
i =1 j =0 k =0 l =0
ARDL procedure, in p1 order to identify the existence of cointegration/long-run
∆ LTB = α
relationshipt between+ ∑
i =1
β1,i ∆LTBt −i
variables.
p1
∆LTBt = α + ∑ β1,i ∆LTBt −i
p2 p3 p4
+ ∑ β 2, j ∆LGDPt − j + ∑ iβ=13,k ∆LM 2t − k + ∑ β 4,l ∆LEt −l

j =0
p2
k =0
p3
l =0
p4
+ λ

1 LTBt −1 ∑
+ λ LGDP
β
2 2, j ∆ LGDP
t −1 + λ LM
t3− j + ∑t −1 3,k 4 t −1t −k t ∑ β 4,l ∆LEt −l
2 β + λ
∆ LE
LM 2 + ε+ . (2)
j =0 k =0 l =0

+λ1 LTBt −1 + λ2 LGDPt −1 + λ3 LM 2t −1 + λ4 LEt −1 + ε t .

The hypotheses for testing cointegration between variables as follows: Null
LTB = ϕ1 + ϕ2 LGDP + ϕ3 LM 2 + ϕ4 LE.
hypothesis H0: λ1 = λ2 = λ3 = ... = λn = 0: no cointegration, no long-run relationship
between variables. LTBAlternative
= ϕ1 + ϕ2 LGDPhypothesis + ϕ3 LM 2 + ϕH 4 LE
: λ ≠ 0, λ1 ≠ 0, λ2 ≠ 0, λ3 ≠ 0, λ4 ≠ 0:
1. 0
existed cointegration, existed long-run relationship
p 2 p p between variables.
∑
3 4

In order αto test H0, j =we 0

β 2, j
compare ∑ β
the 3, k computed ∑ β 4,lF-statistic with the critical
ϕ1 = ;ϕ2 = ; ϕ3 =p k =0p ; ϕ4 =p l =0p . p
values of boundsp
in accordance p
with standard
2
significances (lower bound and
∑
1 1 1 3 1 4

1 − ∑ β1,i
upper one are in
1 − ∑ β1,i
accordance
α with
j =
1β−2,∑
0
j
I(0)
β 1,i
and ∑ 1 β
I(1)
−3,∑
k β ∑
respectively).If
1,i
β 4,l
F-statistic is
i =1
ϕ1 = p
i =1
;ϕ2 = p
i =1
; ϕ3 = k = 0
p
i =1
;ϕ4 = l = 0
p
.
bigger than the critical 1 − ∑valueβ1,i of 1the
1

− ∑upper
1

β1,i bound
1 − ∑ β1,(in
1
accordance
1 − ∑ β1,i with I(1)), we
1

i
reject H0, concluding that i =1 the long-run i =1 relationship i =1 between i =1 variables exists. If

p1
 p1  
∆LTBt = α − ∑  ∑ β1,i  × ∆LTBt −(i −1) 
i = 2  i  
32 banking technology review | No.1,p1
 September
p1
 2017 | Volume  1: 1-148
∆LTBt = α − ∑ ∑ p2  β1,i  ×∆LTBt −(i −1)  
p2

+ β 2,0 ∆LGDPt −i = 2∑i ∑ β 2, j  × ∆LGDPt −( j −1) 

 p1 p2 p3 p4
LTBp1 t = α + ∑ β1,pi 2LTBt −i + ∑ β 2, j LGDP p3 t − j + ∑ β 3, k LM p4 2t − k + ∑ β 4,l LEt −l + ε t .
LTBt p=1 α + ∑p1β1,i LTB p2 i =1
+ ∑ p2
β LGDP jp=3 0 + ∑ p3
β LM 2
p4k = 0
+ ∑ p4
β 4,l LEtl−=l0 + ε t .
+∑
LTBt = αLTBt =βα +i =∑
1,i LTB 1 t −i 1, j =∑
∑ tβ−ti2,−i j +LGDP
β +i LTB +∑β
β2,t2,−j jj LGDP t− j
+k =∑
− kj LM
t3, ∑ β2t −4,tk−lkLE+l t=∑
β3,k k LM
02t − k3,+
β LEt −l + ε t .
−0l + ε4,t l.
l = 0 • DANG lTHI
0
i =1 i =1 j =0 j =0 k =0 LE HOANG
k = 0 PHONG = 0 BACH VAN

p1
∆LTB p1 = α +
pt1 ∑ β1,i ∆LTBt −i
F-statistic is smaller
∆LTBt =
∆LTBt than
∆LTB α +∑
=p1α +the
t =β α1,pi+∆∑ ∑
LTB
critical
β 1, i ∆LTB i =value
1t −i
βt1,−i ∆LTBt −i p 3
of lower bound (in accordance with
I(0)), we accept H0, iconcluding 2 i = 1 that there is a long-run relationship p4 between
p2
variables. p 2 + p 2β ∆LGDP
=1
+ ∑ βi =1
2, j ∆ LGDPp3
t − j + ∑ β 3, k ∆ LMp4 2
t − k + ∑ β 4,l ∆LEt −l
+ ∑ β 2,+j∑
j = 0 t − j + ∑ β 3, k ∆LM k = 0 2t − k + ∑ β 4,l ∆LE
p3 p3 p4 p4
∆=∑ ∑ ∑ k +∑ =∑
l = 0t −l
If F-statistic jlies
0 in
β2,2,j jthe
LGDP ∆ LGDP
t− j + middle tβ j k+
− 3, ∆
of LM β23,tk−∆
0bounds,
LMwe 2βt cannot +lLE
−4,k l ∆ βl 4,l ∆LE
0 t −draw any t −l conclusion, so
j =0 j =0 +λ1 LTB k = 0 t −1 + λ
k=
k =20LGDPt −1l =+0 λ3 LM l2 = 0t −1 + λ4 LEt −1 + ε t .
we need an error +λ correction
LTBt −1 + λ2 LGDP term to define
t −1 ++λλ 3 LM
the
2t −1cointegration.
+ λ4 LEt −1 + ε t .
+λ1 LTB+t −λ11 1+LTB
λ2 LGDP t −1 + λt 2−1LGDP + λ3 LM t −1 2t −13+ LMλ4 LE2t −1t −+
1 +ε
λ4 LE
t . t −1
+ εt .
The long-run equilibrium equation is written as follows:

LTB = ϕ1 + ϕ2 LGDP + ϕ3 LM 2 + ϕ4 LE. (3)

LTB = ϕ1 + ϕ2 LGDP + ϕ3 LM 2 + ϕ4 LE.
LTB = ϕ1LTB ϕ1 + ϕ+2ϕLGDP
+ ϕ2 =LGDP + ϕ LM 2 + ϕ4 LE.
3 LM 2 + ϕ34 LE.
Of which, long-run coefficients are defined as follows:
p2 p3 p4

α∑ β
∑β ∑β
p2
∑β 2,p3j p
3, k4 4,l
∑ ∑
p2 p2 p3 p4
β;ϕ = β;ϕ = p3 p4
∑ ∑;ϕβ = ∑ β ∑;ϕβ = ∑ β .
j =0
ϕ
α ∑β = ;
β ϕ =1
2, j
2,2 j
3, k3
k =0
4,l4
l =0
.
2, jp1 j = 0 p1 p p1
ϕα = α ;ϕ = β 3, k k = 0 3, k 4,l 1 l = 0 4,l

ϕ1 = ϕ = ;ϕ = ;1ϕ− ∑
1
= ;ϕ =
p1 ;1ϕ− ∑
= ;βϕ =
j =0 2 1; ϕ− ∑
=.β
j =p0
1,i1 1. − ∑ β
k = 0 3 k =p0
1,i1
l = 0 4 l =p0
1,i 1 1,i
(4)
1− ∑ β
p1 1
1− ∑ β 1− ∑ β 1− ∑ β
2p1 p1 2 3 p1 p1 3 4 p1 p1 4 p1

1 − ∑ β 1 − ∑ β1 − ∑ β 1 − ∑ 1β− ∑ β 1 − ∑1β− ∑ β 1 − ∑ β
i =1 i =1 i =1 i =1
1,i 1,i 1,i 1,i
1,i i =1 1,i 1,i i =1 1,i 1,i i =1 1,i 1,i i =1 1,i
i =1 i =1 i =1 i =1 i =1 i =1 i =1 i =1

Following Engle-Granger, our ECM-ARDL model with lags is as below:

p1
 p1  
∆LTB p1
t = pα1 −
 −p1∑1∑1β1,ii=2 ×∆LTB
p p ∑    ∑ i 
β1,i  × ∆LTBt −(i −1) 

∆LTBt p=1 α t −( i −1)  
−∑
∆LTBt =∆αLTB t =α ∑ −i =∑
β2 1,i×∑ i ∆LTB
β1,i t −×(i −∆1)LTB
 t −(pi −1)  p
i = 2  i i = 2  i   2  2  
+ β 2,0 ∆LGDP p2  p2 −
 t ∑   ∑ β 2, j  × ∆LGDP  t − ( j −1) 
t − ∑  ∑ β 2, j ×
p2  p2 p2  p2  t −( j −1)  
+ β 2,0 ∆LGDP   j = 2  ∆jLGDP 
2,0t − ∑ t ∑j β =∑ ×∑
+ β 2,0 ∆LGDP + β ∆ LGDP − 22,j j ∆LGDP
β × ∆ LGDP
j  t − ( j −1) 
2,  t − ( j −1)  (5)
 
j =2   j j = 2  j   p 3 p3   
+ β3,0 ∆ p3 LM 2
p3 p3 
 p3pt 3− ∑ ∑ β3,k  × ∆LM  2t −( k −1) 
+ β3,0 ∆LM 2t − ∑ ∑ β3,k k= 2× ∆kLM 2t−( k −1)  
p3

+ β3,0 ∆LM + β23,0t −∆∑ LM2∑ t −k =β∑ 23,k×k∑ ∆LM β3,k2t×−( k∆−1)LM  2t −( k −1) 
k = 2  k k = 2  k 4  4   
p p
+ βp4,0 4 ∆
p4 p4 
 LEp4pt 4− ∑  ∑ β 4,l  ×∆LEt −(l −1) 
+ β 4,0 ∆LEt − ∑ ∑ β 4,l =l 2 ×∆lLEt −(l −1)  
p4

+ β 4,0 ∆LE + tβ−4,0∑ ∆LE  ∑ t −l =β∑ p1×∑
2 4,l l ∆LE
β 4,tl−(l −×1)∆ LEt −(l −1) 

l = 2  l l = 2  l   
p1 − (1 − ∑ β1,i ) ECM t −1 + µt .
− (1 − ∑1β1,i ) ECM
p
p1
+ µ .
−i =∑
i =1
− (1 − ∑−β(1 1,i ) ECM 1
βt1,−i1)+ECM µt . t −t1−1 + µt t .
i =1 i =1

Where: p1 p2 p3 p4
ECM t −1 = LTB p1
p t p2
− α − ∑ βp1,2pi LTBt −i − ∑ β 2, j LGDP p3
t − pj − ∑ β 3, k LM
p4 2t −k − ∑ β 4,l LEt −l .
ECM t −1 = LTBpt1 − α − ∑1β1,i LTB t −i − ∑ β 2, j LGDPt − j − ∑ β 3, k LM 2t − k − ∑ β 4,l LEt −l .
2 p3 p3 p4
−i =∑ i −j =∑
0 t − j2, j ∑ 3, tk− j k =∑ t −l =l∑
4

− ∑t β−1,αi LTB 1t −i 1,∑

β LM− 20t −kβ−3,k∑
i =1 j =0 k =0 l =0
ECM t −1 =ECM
LTBt −−1 =α LTB β− i LTB
β 2,t j−LGDP β −LGDP LMβ 4,2ltLE
−k − .0 β 4,l LE(6)
t −l .
i =1 i =1 j =0 j =0 k =0 k =0 l =0 l =0

4. Results
Unit root test: This test makes sure that variables are not integrated of order
two (2)//there is no integration of order 2 variables//because of mentioned

Volume 1: 1-148 | No.1, September 2017 | banking technology review 33

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

spurious regression. Table 2 reports the results of ADF unit root test of Dickey
& Fuller (1979).
Table 2: Unit root test’s results
Order of
Variable T-statistic Conclusion
integration
LTB -2.731 Non-stationary process
ΔLTB -4.762 Stationary process I(1)
LGDP -1.764 Non-stationary process
ΔLGDP -11.936 Stationary process I(1)
LM2 -1.437 Non-stationary process
ΔLM2 -7.502 Stationary process I(1)
LE -9.686 Stationary process I(0)

As can be seen in Table 2, at the 5% significance level, LTB, LGDP and

LM2 are together integrated of order 1, except LE is integrated of order zero.
According to Pesaran et al. (1996), Hamuda et al. (2013), applying ARDL is the
most appropriate method for doing empirical analysis when we cannot ensure
the property of unit root and the stationary of the system, and/or all variables do
not have the same integrated order of I(1) or I(0).
Bound test: This test defines the cointegration relationship between variables.
The results are presented in Table 3.

Table 3: Bound test’s results

Number of Test
Critical Value Bounds
regressor statistic
90% 95% 97,5% 99%
k F-statistic
I(0) I(1) I(0) I(1) I(0) I(1) I(0) I(1)
3 125.100 2.711 3.800 3.219 4.378 3.727 4.898 4.385 5.615

Table 3 shows that F-statistic is larger than the critical value upper bound
at 1% significance level (provided in the appendices of Pesaran et al. (1997).
Thus, we can reject the null hypothesis H0, accepting the alternative one H1: that
the cointegration exists between variables, in other words, there is existence of
long-run relationship between variables.
Choosing the lag length for ARDL: based on SBC criteria, the optimal lags
for our ARDL is ARDL(2.0,2.0) (Table 4).
Our ARDL has R-Squared of 0.979 and R-Bar-Squared of 0.971, meaning

34 banking technology review | No.1, September 2017 | Volume 1: 1-148

 LE HOANG PHONG • DANG THI BACH VAN

Table 4: ARDL estimation with LTB as dependent variable

Variable Coefficient Standard error T-statistic P-value
LTB(-1) 0.253*** 0.080 3.168 0.005
LTB(-2) -0.112* 0.059 -1.907 0.072
LGDP 0.203** 0.076 2.676 0.015
LM2 0.168* 0.083 2.011 0.059
LM2(-1) -0.025 0.121 -0.204 0.840
LM2(-2) -0.306*** 0.081 -3.750 0.001
LE 0.628*** 0.049 12.894 0.000
INPT -6.910*** 0.726 -9.514 0.000
R-Squared 0.979 DW-statistic 1.602
R-Bar-Squared 0.971 Schwarz Bayesian Criterion 38.084
F-statistic 125.100 Pob (F-statistic) 0.000
Note: ***, ** and * are the significance levels of 1%, 5% and 10% respectively.

that the gross domestic product, exchange rate and money supply can explain
up to 97% movement of TB.
Diagnostic tests for the appropriateness of the model: we conduct regarded
tests such as: Wald test, Ramsey’s RESET test for wrong model identification,
Larange multiplier (LM) test for serial correlation, and heteroscedasticity test
(see in Table 5).
Besides, we test the stability of residuals by employing the CUSUM
(cumulative sum of recursive residuals) and CUSUMSQ (cumulative sum of the
square of recursive residuals). The results are graphed in Figure 2 and Figure 3
respectively, indicating that the CUSUM and CUMSQ move within the standard
Table 5: Diagnostic tests’ results
No Test Statistic Test-statistic P-value
CHSQ (7) 875.702 0.000
1 Wald
F (7. 19) 125.100 0.000
CHSQ (1) 0.003 0.955
2 Model identification
F (1. 18) 0.002 0.964
CHSQ (1) 0.731 0.393
3 Serial correlation
F (1. 18) 0.501 0.488
CHSQ (1) 0.841 0.359
4 Heteroskedasticity
F (1. 25) 0.804 0.379

Volume 1: 1-148 | No.1, September 2017 | banking technology review 35

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

Figure 2: Cumulative sum of recursive residuals

15
10
5
0
-5
-10
-15
1988 1993 1998 2003 2008 2013 2014

Figure 3: Cumulative sum of square of recursive residuals

1.5

1.0

0.5

0.0

-0.5
1988 1993 1998 2003 2008 2013 2014

Table 6: Long-run coefficients from ARDL(2,0,2,0)

with LTB as dependent variable
Variable Coefficient Standard error T-statistic P-value
LGDP 0.236** 0.098 2.412 0.026
LM2 -0.190*** 0.066 -2.878 0.010
LE 0.731*** 0.108 6.781 0.000
INPT -8.043*** 1.360 -5.913 0.000
Note: ***, ** and * are the significance levels of 1%, 5% and 10% respectively.

intervals of 5% significance level. These results suggest that the residuals are
stable and thus, our model is stable.
In short, the tests’ results indicate that our model is reliable and robust to
estimate the long-run and short-run coefficients.
Estimating ARDL’s long-run coefficients: Table 6 reports the estimated
coefficients of the ARDL(2.0,2.0).
The coefficients show that in the long-run, all variables have statistically

36 banking technology review | No.1, September 2017 | Volume 1: 1-148

 LE HOANG PHONG • DANG THI BACH VAN

significant effects on trade balance of Vietnam (LTB). Particularly, gross domestic

product (LGDP) and exchange rate (LE) have positive impacts whereas money
supply (LM2) has a negative one.
Estimating ARDL’s short-run coefficients: In order to analyze the influence
of short-run movement’s tendency on long-run equilibrium, we use the error
correction model (ECM). Table 7 gives the results for short-run coefficients of
our chosen ARDL.

Where:
ECM = LTB - 0,236 x LGDP + 0,190 x LM2 - 0,731 x LE + 8,043 x INPT (7)

This result indicates that in the short-run, the difference of gross domestic
product, money supply and exchange rate positively affect Vietnam’s trade
balance. Those effects are statistically significant.
The error correction provides feedback information or the adjustment speeds
of short-run coefficients that converge to their long-run equilibrium in the
model. The error correction term’s coefficient ECM(-1) is statistically significant
at 1% level, ensuring the existence of cointegration relationship as found in the
Pesaran (1997) bound test. The error correction term’s coefficient is -0.859, lying
well on the range of (-1;0) and suggesting a fast speed of adjustment. Each year,
it has to adjust up to 86% of the difference between short-run and long-run

Table 7: Short-run effects in ARDL-based ECM

with ΔLTB as dependent variable
Variable Coefficient Standard error T-statistic P-value
ΔLTB(-1) 0.112 * 0.059 1.907 0.071
ΔLGDP 0.203** 0.076 2.676 0.015
ΔLM2 0.168* 0.083 2.011 0.058
ΔLM2(-1) 0.306*** 0.082 3.750 0.001
ΔLE 0.628*** 0.049 12.894 0.000
INPT -6.910*** 0.726 -9.514 0.000
ECM(-1) -0.859*** 0.100 -8.584 0.000
R-Squared 0.971 DW-statistic 1.602
R-Bar-Squared 0.960 Schwarz Bayesian Criterion 38.084
F-statistic 105.830 Prob (F-statistic) 0.000
Note: ***, ** and * are the significance levels of 1%, 5% and 10% respectively.

Volume 1: 1-148 | No.1, September 2017 | banking technology review 37

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

values. The ECM can explain 96% of variations in the short-run of Vietnam’s
trade balance.
The impacts of GDP on trade balance valid both in the short-run and in
the long-run. This result is in line with the theory that when income (GDP)
increases, demand for money increases. To clear the money market, a country
must promote exports, generating positive effect on trade balance (Mundell,
1968). According to Solow’s growth model (Solow, 1956), in the early stage, a
low developed country can enjoy a higher growth rate when it borrows external
capitals for investment. Later, the economy is close to its steady state, thus, the
growth rate declines. Hence, a higher growth rate suggests that this country
is undergoing its initial period of developments and tends to accept current
account deficit.
The effects of money supply on trade balance can be explained within the
framework of monetary approach. In short-run, an increase in money supply
will result in a decrease in trade balance (Magee, 1976). By contrast, the
excessive money supply facilitates increasing imports and worsens trade balance
(Mundell, 1968; Magee, 1976; Johnson, 1976; Miles, 1978; Frenkel et al., 1987).
In the long-run, exchange rate has a positive impact on trade balance of
Vietnam. This result supports ones of Bickerdike (1920), Robinson (1947),
Metzler (1948): when exchange rate increases, domestic currency depreciates,
foreign goods and services are relatively more expensive than domestic ones,
imports decrease, exports increase, and trade balance improves. In Vietnam,
the fact that exchange rate has been continuously increasing in the past period
deteriorates our trade balance. This result is strengthened when it is in line with
the study of Ha Thi Thieu Dao et al. (2013).

5. Conclusion
The empirical results show that up to 97% of trade balance’s fluctuation in
Vietnam is due to the variations of the gross domestic product, exchange rate
and money supply. In the long-run, they all statistically significantly affect trade
balance (LTB).
A 1% increase of gross domestic product is associated with 0.24% increase in
trade balance at 5% significance level. This result is consistent with the study of
Duasa (2007). Based on this, Vietnam needs to focus on growth to improvethe
trade balance. According to Mundell (1968), an increase in income (GDP) causes
an increase in money demand. To balance the money, the country stimulates
exports, generating positive effects on the trade balance.

38 banking technology review | No.1, September 2017 | Volume 1: 1-148

 LE HOANG PHONG • DANG THI BACH VAN

Money supply has an adverse effect on trade balance at 1% significant level,

when money supply increases by 1%, trade balance decreases by 0,19%. This
result supports the argument of Magee (1976) that an increase of money supply
will create an adverse effect on trade balance. Thus, to stabilize trade balance (a
part of the balance of payments), a solid fulcrum for macroeconomic stability,
the government should strictly control the money supply. It is very crucial to
implement an active, flexible, and prudent monetary policy in order to stabilize
the money market, ensure the banking system’s liquidity, and meet the capital
demand for production and business of which prioritizes export-led sectors.
A key point in controlling the money supply is controlling the credit growth.
Excessive credit growth relative to nominal GDP growth will soon put pressure
on price levels and interest rates. Therefore, the control of money supply must
be appropriate with nominal GDP growth rate. In addition, since the financial
market has been stable, it is necessary to let market forces determine interest
rates. Maintaining a cap on deposit rates for under 6-month savings has led
to difficulties in attracting commercial banks' deposits, as well as increasing
consumer spending and driving savings to higher yield’s asset markets. This may
be the cause of imbalance in the capital market. Moreover, both restrictions on
maturity and the cap of government bond yields hinder the development of the
capital market and the SBV’s ability to control money supply and interest rates.
Therefore, the priority should be provided to the development of capital markets
for the development of financial markets. On the other hand, it is necessary to
regulate government expenses, in order to achieve fiscal balances in medium
and long-term.
In our study, exchange rate also has a noticeable impact on TB at 1%
significance level. For every 1% increase in exchange rate, trade balance rises
by 0.73%. For the current exchange rate management mechanism, in order to
improve the trade balance, the government should reduce the risks, especially to
increase the flexibility of the exchange rate within the permissible limits, towards
macroeconomic stability, export promotion, and import restrictions. In the
medium term, when the improved domestic financial market is matched with
effective financial regulation mechanisms, it is imperative and indispensable to
open capital and financial accounts as committed in the roadmap. For that to do,
a managed floating exchange rate regime is a reasonable choice.

Volume 1: 1-148 | No.1, September 2017 | banking technology review 39

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

References

Alexander, S. S. (1952). Effects of a devaluation on a trade balance. International

Monetary Fund Staff Papers, 2, 263-278.
Alexander, S. S. (1959). Effects of a devaluation: a simplified synthesis of elasticities and
absorption approaches. American Economic Review, 49, 21-42.
Bickerdike, C. F. (1920). The Instability of Foreign Exchange. The Economic Journal, March.
Bui Trinh, Kiyoshi Kobayashi & Vu Trung Dien (2010). Tai co cau nen kinh te va giam
tham hut thuong mai: Viec lua chon nganh trong diem. Kien nghi cua Hoi thao khoa hoc “On
dinh kinh te vi mo, duy tri da tang truong kinh te Viet Nam 2010, Trien vong nam 2011”.
Uy ban Kinh te cua Quoc hoi, 243-273 (Restructuring the economy and reducing the trade
deficit: Selecting key industries. Recommendations of the Scientific Conference named
“Stabilizing macro economic and Maintaining the economic growth in Vietnam in 2010,
prospects for 2011”. Economic Committee of the Parliament, 243-273).
Chun, H. W, Chun, H. A. L. & Chih, H. Y. (2012). Short-run and Long-run effects of
Exchange rate change on Trade balance: Evidence from China and its trading partners.
Japan and the World Economy, 24, 266-273.
Dash, A. K. (2003). Bilateral J-curve between India and her trading partners: a
quantitative perspective. Economic analysis and policy, 43(3), 315-338.
Dickey, D. & Fuller, W. (1979). Distribution of the Estimators for Autoregressive Time
Series with a Unit Root. Journal of the American Statistical Association, 74, 427-431.
Duasa, J. (2007). Determinants of Malaysia Trade Balance: An ADRL Bound Testing
Approach. Global Economic Review, 36(1).
Kremers, J. J. M., Ericsson, N. R. & Dolado, J. J. (1992). The power of cointegration
tests. Oxford Bulletin of Economics and Statistics, (54), 325-348.
Frenkel, J. A. & Razin, A. (1987). The Mundell-Fleming Model: A Quarter-Century
Later. International Monetary Fund Staff Papers, 34(4), 567-620.
Hacker, R. S. & Hatemi, A. (2004). The Effect of Exchange rate changes on Trade
balances in the short and long run: Evidence from German trade with transitional Central
European economics. Economics of transition, 12(4), 777-799.
Hamuda, A. M., Suliková, V., Gazda, V. & Horváth, D. (2013). ARDL investment
model of Tunisia. Theoretical and Applied Economics, 20(2), 57-68.
Ha Thi Thieu Dao & Pham Thi Tuyet Trinh (2013). Moi quan he ty gia hoi doai va
can can thanh toan. Tap chi Khoa hoc Dao tao ngan hang, 103, 17-24 (The relationship

40 banking technology review | No.1, September 2017 | Volume 1: 1-148

 LE HOANG PHONG • DANG THI BACH VAN

between exchange rate and the balances of payment. Banking Science and Training Review,
103, 17-24).
Harberger, A. C. (1950). Currency depreciation, income and the balance of trade.
Journal of Political Economy, 58, 47-60.
Johnson, H. G. (1976). The Monetary Theory of Balance of Payments Politics. In the
Monetary Approach to the Balance of Payments, ed. Jacob A.Frenkel and Harry G. Johnson,
Toronto.
Le Hoang Phong & Dang Thi Bach Van (2015). Kiem chung bang mo hinh ARDL tac
dong cua cac nhan to kinh te vi mo den chi so chung khoan Viet Nam. Tap chi Phat trien
va Hoi nhap, 20(30), 61-66 (Testing the impacts of macroeconomic factors to the security
index in Vietnam using ARDL model. Development and Integration Review, 20(30), 61-66).
Magee, S. (1976). The Empirical Evidence on the Monetary Approach to the Balance
of Payments and Exchange Rates. American Economic Review paper and Proceedings, 66.
Meade, J. E. (1951). The Balance of Payments (Oxford: Oxford University Press).
Metzler, J. E.(1948). A survey of Contemporary Economics. Vol.I, Richard D.Irwin, INC,
Homewood, IL.
Miles, M. A. (1978). Devaluation, the Trade Balance and the Balance of payments. New
York: Marcel Dekker.
Mundell, R. A. (1968). International Economics. New York: Macmillan.
Nguyen Huy Tuan (2011). Phan tich tac dong cua cac bien so kinh te vi mo den can can
thuong mai Viet Nam. Tap chi Cong nghe Ngan hang, 65, 13-22 (Analyzing the impacts of
macroeconomic variables on the balance of trade in Vietnam. Banking Technology Review,
65, 13-22).
Pesaran, M. H., Shin, Y. & Smith, R. J. (1996). Bounds Testing Approaches to the
Analysis of Level Relationships. DEA Working Paper 9622, Department of Applied
Economics, University of Cambridge.
Pesaran, M. H. & Pesaran B. (1997). Working with Microfit 4.0 - Interactive
Econometric Analysis. Oxford University Press, 478.
Phan Thanh Hoan & Nguyen Dang Hao (2007). Moi quan he giua ty gia hoi doai va
can can thuong mai Viet Nam thoi ky 1995-2004. Tap chi Khoa hoc, Truong Dai hoc Hue,
43 (The relationship between exchange rate and trade balance in Vietnam in the period
1995-2004. Hue University Journal of Science, 43).
Polak, J. J. (1957). Monetary analysis on income formation and payments problems.
International Monetary Fund Staff Papers, 6, 1-50.
Robinson, J. (1947). Essays in the Theory of Employment. Oxford:Basil Blackwell.
Solow, R. (1956). A contribution to the Theory of Economic Growth. The quarterly
journal of economics, 70(1), 65-94.

Volume 1: 1-148 | No.1, September 2017 | banking technology review 41

 The Impact of Macroeconomic Factors on trade Balance in Vietnam

Sugema, I. (2005). The Determinants of trade balance and adjustment to the crisis in
Indonesia. Centre For International Economic Studies Discussion Paper. 0508, 1-28.
Tran Hong Ha (2011). Ung dung mo hinh thuc hien do luong tac dong cua ty gia thuc
da phuong den can can thuong mai Viet Nam. Tap chi Thi truogn Tai chinh tien te, 16, 23-25
(Applications of models to measure the impacts of real multilateral exchange rates on the
balance of trade in Vietnam. Financial and money market Review, 16, 23-25).
Tran Nguyen Ngoc Anh Thu & Le Hoang Phong (2014). Tac dong cua dau tu cong
doi voi tang truong kinh te o Viet Nam: Goc nhin thuc nghiem tu mo hinh ARDL. Tap chi
Phat trien va Hoi nhap, 20(30), 3-10 (Impacts of public investments on economic growth
in Vietnam: An empirical view with ARDL model. Development and Integration Review,
20(30), 3-10).

42 banking technology review | No.1, September 2017 | Volume 1: 1-148

View publication stats