Ước lượng mức dự trữ ngoại hối tối ưu của việt nam tt tiếng anh

MINISTRY OF EDUCATION AND TRAINING

THE STATE BANK OF VIETNAM

BANKING UNIVERSITY HOCHIMINH CITY

*****

TRAN VUONG THINH

ESTIMATION OF VIETNAMESE OPTIMUM
RESERVES

SUMMARY OF DOCTORAL THESIS IN ECONOMICS

Major: Finance - Banking
Code: 9.34.02.01

Supervisor:
ASSOC.PROF., PH.D LE PHAN THI DIEU THAO

HOCHIMINH CITY – 2020


SUMMARY
With the important role of reserves, nations try to increase
reserves more and more and cause much opportunity cost because
of the low profit from investing reserves. Therefore, many
economic scholars research to estimate the optimum reserves of a
nation. Under this circumstance, there are three main and popular
estimation method including method based on rule of thumb,

method based on the determinants of reserves, method based on
the cost – benefit of reserves.
The research objective of the thesis is to select the method of
estimating the optimum reserves appropriate to Vietnam in three
above method and give some suitable policy implications for
government authorities. Through using Vietnamese quarterly and
yearly data in the period of 2005 – 2017 together with the
qualitative methods such as description, analysis, comparison and
the quantitative methods like ARCH, ADF, OLS, HP Filter,
ARDL, the thesis experiments upon Vietnam with three above
method. The empirical result of all three methods comes to a
conclusion that in the end of research period, the optimum
reserves approximates the actual reserves. Then Vietnam needs
continuing to increase reserves but shouldn’t push strongly the
increasing speed.
At the same time, the thesis points that the method based on the
cost – benefit of reserves is appropriate to apply in Vietnam at the
present period. Then the thesis gives some policy implications to
Vietnam such as improving the method of calculating the
variables in the estimation model, estimating in advance the
optimum reserves of next year, both controlling the optimum
reserves and increasing the actual reserves.
Keywords: reserves, optimum reserves.


1

CHAPTER 1: RESEARCH INTRODUCTION
1.1. The necessity of research
Reserves is always interested in by governments because its roles

are to manage the exchange rate towards governments’ desire and
to prevent the appearance of sudden shocks which make
vulnerable the domestic economy. Therefore, reserves in the
world and in Vietnam increase continuously during the last years.
However, the reserves assets must have the high safety and
liquidity, so their profitability is much lower than the high-risk
assets. Then, holding reserves makes the holding cost which is
called as the opportunity cost.
From the above, a nation which tries to hold much reserves
maybe is not good and it should only hold at the enough level for
nation’s needs because the extra reserves used for the necessary
activities can raise the effectiveness of economy. So how much
reserves should a nation hold so that it becomes suitable, adequate
or optimum? Many researches try to estimate the nation’s
optimum reserves and there are three outstanding main methods
including method based on rule of thumb, based on the
determinants of reserves, based on the cost – benefit of reserves.
With the methods based on rule of thumb, the optimum reserves
is estimated according to rule of thumb and applied the same to
all nations in the world. At first, it needs to list the traditional
estimation methods including the one based on import turnover
originated from the research in 1958 (Wijnholds and Kapteyn,
2001), the method based on short-term external debt proposed by
Greenspan (1999), the method based on broad money supply M2
raised by Kaminsky (1999) and Wijnholds and Kapteyn (2001),
the method based on GDP originated from Jeanne and Ranciere
(2006). In addition, many methods linked the traditional methods
also appear such as the extended Greenspan-Guidotti method or

2
the method including three above traditional methods and taking
total proposed by Shcherbakov (2002). In 2011, IMF raised the
method estimating optimum reserves for emerging countries
called as the ARA EM method and up to 2016, IMF officially
instructed to apply this method. For traditional methods, Vietnam
also applied through the researches of Nguyen Thi Xuan Phuong
(2012), Le Thi Tuan Nghia and Pham Thi Hoang Anh (2013) or
Tran Kim Anh (2018).
The another method is the method based on the determinants of
reserves to build the reserves-demand function and estimate
optimum reserves. Among the researches following this method,
the research of Edison (2003) is acknowledged and applied by
many following researches because this research builds the
reserves-demand function including five determinants covering
almost all reserves - influencing aspects with the data of 122
countries in the period of 1980 – 2002. In Vietnam, there are a
few of researches applying this method, such as the research of
student group at University of Economics Hochiminh City which
attended the scientific research contest named “The 2010 young
economist”.
The third estimation method is the method based on the cost –
benefit of reserves. The pioneer of this method is Heller (1966)
and his method is much applied by the following researches. In
these ones, there are two remarkable models including the model
of Frankel and Jovanovic (1981) and the model of Ben-Bassat and
Gottlieb (1992).
For Vietnam, the government’s regulation of optimum reserves is
completely not yet done. The Decree No. 50/2014/NĐ-CP about
reserves management only requires The State Bank establish the

reserves limit in investing suitably reserves to ensure the safety,
liquidity and profitability. This Decree has not yet mentioned the
determination of optimum reserves by The State Bank.


3
Because the optimum reserves has not yet much been interested in
Vietnam, the author finds some gaps in the space of research
about optimum reserves as follows: (i) up to now, the author has
not yet found any Vietnamese documents or researches fully
presenting the theory of optimum reserves and methods of
estimating optimum reserves. (ii) the ARA EM method for
emerging countries is appropriate to apply to Vietnam but the
author has not yet found any Vietnamese researches
experimenting with this method. (iii) the model of Ben-Bassat and
Gottlieb (1992) including sovereign risk – a characteristic of
developing countries – is also suitable for applying to Vietnam
but the author has also not yet found any Vietnamese researches
experimenting with this method.
Therefore, the research about optimum reserves in Vietnam is
very necessary. This is the reason why the author decides to
choose the research “Estimation of Vietnamese optimum
reserves” as the name of the author’s doctoral thesis. Through
this research, the author hopes that Vietnam only holds the
optimum reserves, being suitable with the Vietnamese economic
condition in each period and invest the extra resources into the
other necessary activities to foster the nation’s growth.
1.2. Research objectives
The overall research objective is to select the suitable method of
estimating optimum reserves to Vietnam and then propose some

policy implications to government authorities basing on the
experiment result of the selected method. To complete the overall
objectives, the thesis needs to obtain the detail objectives:
Firstly, the thesis clarifies three main methods of estimating
optimum reserves in the world at present.
Secondly, basing on the Vietnamese economic data and situation,
the thesis estimates Vietnamese optimum reserves according to


4
the methods based on rule of thumb, based on the determinants of
reserves, based on the cost – benefit of reserves.
Thirdly, the thesis selects the suitable method of estimating
optimum reserves to Vietnam in three methods. With the
experiment results of the selected method and comparing the
optimum reserves with the actual reserves, the thesis proposes
some policy implications to government authorities.
1.3. Research questions
With the above detail research objectives, the research questions
are raised in turn as folows:
First, how are three main methods of estimating optimum
reserves in the world at present carried out?
Second, basing on the Vietnamese economic data and situation,
when estimating optimum reserves: which the suitable measure
methods does it need to carry out for the method based on rule of
thumb? How does it need to build and carry out the model for the
method based on the determinants of reserves? How does it need
to use the way to experiment model for the method based on the
cost – benefit of reserves?
Third, which method in three main methods is suitable with

Vietnam? Which policy implications concerning the experiment
results of the suitable method should be proposed to government
authorities?
1.4. Research object and scope
Research object of the thesis is to estimate optimum reserves.
Research scope is Vietnam during the period of 2005 – 2017.
1.5. Research methods and data


5
The thesis uses qualitative methods such as listing, analyzing,
comparing, synthesizing the previous theories and researches for
reaching the first and the third detail objectives as well as a part
of the second detail objective.
The thesis also uses quantitative methods for obtaining the second
detail objective including: (i) the way of analyzing the simple
indicators for the method based on rule of thumb. (ii) the
econometric methods like ARCH, HP Filter, ADF test, OLS
regression and ARDL model are done by Stata 13.0 software for
two methods based on the determinants of reserves and based on
the cost – benefit of reserves.
The overall research process is displayed in the Figure 1.1.
The research data is collected in yearly and quarterly data during
the period of 2005 – 2017 for 11 types of data including reserves,
import turnover, export turnover, foreign portfolio investment,
short-term external debt, broad money supply M2, nominal GDP,
VND/USD exchange rate, fiscal deficit, lending rate of VND, 3month LIBOR rate of USD. These types of data are collected
from the prestigious sources such as IFS, Bloomberg, ADB,
Worldbank, CEIC Data, ICE, General Statistics Office of
Vietnam (GSO), The State Bank, The Ministry of Finance.

1.6. New findings of the thesis
In the academic aspect, (i) Basing on synthesizing the theory of
optimum reserves from the researches all over the world, the
thesis presents clearly the theory about three main methods of
estimating optimum reserves which has not yet systematized in
Vietnam by the previous researches. Three main methods include
method based on rule of thumb, based on the determinants of
reserves, based on the cost – benefit of reserves. Therefore, thesis
contributes to enrich the theory of optimum reserves for
Vietnamese economic academic basis.


6
ESTIMATION OF VIETNAMESE OPTIMUM RESERVES

The previous involving researches and the theory about estimating optimum reserves

The method based on rule
of thumb

The method based on the
determinants of reserves

The method based on the
cost – benefit of reserves

The real situation and Vietnamese economic data

Experiment of three
traditional methods and

ARA EM method

Build reserves-demand
function, the independent
variables are the determinants

Build the research model
based on the model of BenBassat and Gottlieb (1992)

Using the yearly data and
optimum standards based on
rule of thumb

Using the quarterly data and
the ARCH method, ADF test,
OLS regression

Using the quarterly data and
the methods of HP Filter,
ARCH, ADF, ARDL

Estimating Vietnamese optimum reserves and comparing actual reserves

Discussion of the research results

Selecting the suitable method to Vietnam

Policy implications for Vietnam basing on the selected method

Figure 1.1. The research process of the thesis

Source: The author’s synthesis


7
(ii) the ARA EM method proposed by IMF and the method of
Ben-Bassat and Gottlieb (1992) are the methods of estimating
optimum reserves which the author has not yet found at any
previous researches and experiments in Vietnam. By carrying out
these two methods in Vietnam, the thesis supplies the Vietnamese
experiment evidences of two methods for the field of optimum
reserves in the world academic basis.
In the practical aspect, (i) the ARA EM method proposed on the
data of emerging countries. The method of Ben-Bassat and
Gottlieb (1992) designs the model based on a characteristic of
developing countries – sovereign risk. Therefore, these two
methods are suitable for experimenting in Vietnam. Besides, the
method based on the determinants of reserves shows the linkage
with nation’s particular situation in each period, so it is also
suitable for experimenting in Vietnam. Through these
experiments, the thesis supplies some methods of estimating
optimum reserves and help Vietnam have more selections in
determining optimum reserves. (ii) With the optimum reserves
from the experiment results of three methods and compared with
the actual reserves, all of them come to the general conclusion
that Vietnam should continue to increase reserves but needs to
follow the plan and is not necessary to push strongly the speed of
cumulating reserves. Basing on this conclusion, the thesis
proposes the appropriate policy implications to government
authorities in order to use effectively the nation’s resources.
1.7. The structure of the thesis: The thesis includes 5 chapters:

Chapter 1: Research introduction.
Chapter 2: Theory of estimating optimum reserves.
Chapter 3: Model of estimating Vietnamese optimum reserves.
Chapter 4: Results of estimating Vietnamese optimum reserves.
Chapter 5: Conclusions and policy implications for Vietnam.


8

CHAPTER 2: THEORY OF ESTIMATING OPTIMUM
RESERVES
2.1. Foreign exchange reserves (Reserves)
2.1.1.Concept of foreign exchange
Foreign exchange is the facilities of international payment
including foreign currencies, valuable papers in foreign currency,
international - standard gold and domestic currency in
international payment.
2.1.2. Concept of reserves
According to IMF (2009), reserves is the external assets which are
available and controlled by nation’s monetary authorities for
meeting the financing needs and equilibrating the balance of
payments, for intervening in forex market to impact on the
exchange rate and for other related purposes.
2.1.3. Sources of reserves
Including kinds of resources: receiving the SDR allocation from
IMF basing on the nation’s level of contributing to IMF; drawing
reserves from IMF if there is reserves position in IMF; making
increase of purchasing foreign currencies, gold and valuable
papers by money inflows of current account and financial account
2.1.4. Reason for the necessity of holding reserves

The Impossible Trinity originated from Frankel, J.A. in 1999 is
the reason. Frankel (1999) says that according to the global trend
of integration, nations can choose the neutral model. Then,
reserves can help nations control the fluctuation of exchange rate
in the determined limitation.
2.1.5. Roles of reserves
Two main roles is to affect the exchange rate for stabilizing
current account (mercantile motivation) and to finance for


9
stabilizing financial account (precautionary motivation). Besides,
the other roles of reserves include raising the belief of people and
investors in domestic currency, helping nations borrow external
debts more easily…
2.2. Optimum reserves
2.2.1. The necessity to determine optimum reserves
Basing on the following foundations: (i) Triffin’s Dilemma
presents that when USD is the main global reserves assets, the
more reserves raises, the more USA balance of payments gets
deficit. This makes unbalance the global economy and destroy the
stability of the international monetary system. Therefore, nations
should not hold reserves too much. (ii) Because the reserves
assets must ensure safety and liquidity, the profitability is very
low and so , it creates the big opportunity cost when holding too
much reserves. Then, nations only need to hold reserves at the
adequate or optimum level.
2.2.2. Concept of optimum reserves
The optimum reserves is the reserves level which is “adequate to
safety”, ensures dealing with the sudden shocks to nation, mainly

the shocks influencing on two important components of balance
of payments - current account and financial account.
2.3. The method of estimating optimum reserves based on rule
of thumb
2.3.1. Traditional methods
Including the method based on import turnover with the optimum
reserves level of import turnover in 3 months which implies that
reserves can be enough for financing balance of trade or current
account; the method based on short-term external debt with the
optimum reserves level of 100% of short-term external debt so
that reserves is adequate to finance the volatility of financial


10
account; the method based on broad money supply M2 with the
optimum reserves level of 20% of M2 for ensuring that reserves
can finance the outflow of domestic capital due to nation’s
instability; the method based on GDP with the optimum reserves
level of 10% of GDP which is rarely used in fact. Conversely,
three above popular methods are much used around the world.
2.3.2. Methods based on linking traditional methods
These methods link the above traditional methods with the
popular ways of linking such as linking short-term external debt
with deficit of current account; comparing the traditional methods
each other and selecting the highest reserves; linking three above
popular traditional methods and taking total amount.
2.3.3. IMF’s ARA EM method
The ARA EM method evaluates the adequate reserves level of
emerging countries, proposed by IMF in 2011 and officially
instructed to apply in 2016. IMF finds that there are four sources

causing the depletion of reserves in emerging countries, including
export turnover (X), short-term external debt (STED), other
portfolio loan which is represented by foreign portfolio
investment (OPL), broad money supply (M2). With this basis, the
optimum reserves level is displayed as follows:
• For nations without controlling capital flow
Fixed rate regime: R* = 30% STED + 20% OPL + 10% M2 + 10% X
Floating rate regime: R* = 30% STED + 15% OPL + 5% M2 + 5% X

• For nations with controlling capital flow
Fixed rate regime: R* = 30% STED + 20% OPL + 5% M2 + 10% X
Floating rate regime: R* = 30% STED + 15% OPL + 2.5% M2 + 5% X

In Vietnam, the author has not yet found any researches on the
ARA EM method, so the author carries out this method for


11
Vietnam to raise the way of determining Vietnamese optimum
reserves.
2.4. The method of estimating optimum reserves based on the
determinants of reserves
This method is based on building the linear regression function in
which reserves is dependent variable and determinants of reserves
are independent variables. After estimating the parameters,
reserves-demand function is founded and it is used for estimating
optimum reserves.
2.4.1. Review of related researches
Edison (2003) presents comprehensively five determinants of
reserves to build reserves-demand function and it is much applied

in many following researches. These researches include the
research of Gosselin and Parent (2005) which build reservesdemand function for eight Asian emerging countries; the research
of Khan and Ahmed (2005) which experiments for Pakistan; the
research of Prabheesh et al (2007), Sehgal and Sharma (2008),
Nainwal et al (2013) which experiments for India; the research of
Afrin et al (2014), Chowdhury et al (2014) which experiments for
Bangladesh. In Vietnam, there are the research of student group at
University of Economics Hochiminh City which attended the
scientific research contest named “The 2010 young economist”
and the research experiments for Vietnam basing on the model of
Edison (2003).
2.4.2. Determinants of reserves
Through review of the above related researches and theory of
reserves, there are six determinants of reserves.
The economy scale: the larger it is, the higher reserves is.


12
The vulnerability of current account: is influenced by the
volatility of balance of trade (through trade openness and export
volatility) and scale of remittances.
The vulnerability of financial account: is dependent on financial
openness and broad money supply M2.
The flexibility of exchange rate: is measured by the volatility of
exchange rate and it has the negative or positive relationship with
reserves.
The opportunity cost: the higher it is, the lower reserves is
because nations are afraid of being costly when holding much
reserves.
The stability of nation: includes the economic stability and the

political stability, help to raise the position of domestic currency.
2.4.3. Remarks to build the model in Vietnam
Firstly, the author experiments for Vietnam basing on the model
of Edison (2003) including five determinants due to its credibility.
Secondly, the author uses ARCH model for calculating the
standard deviation as proxy for export volatility and exchange rate
volatility.
Thirdly, to show the highest opportunity cost so that Vietnam has
to be more careful when cumulating more reserves, the author
chooses the lending rate of VND as proxy for opportunity cost.
Fourthly, the vulnerability of financial account is only measured
by financial openness (foreign portfolio investment/ GDP) and the
author eliminates M2 because Vietnam always controls seriously
the capital inflows and outflows.
2.5. The method of estimating optimum reserves based on the
cost – benefit of reserves
2.5.1. Approach of “cost–benefit” of reserves from Heller (1966)


13
Heller (1966) is the first person who proposes this approach when
he states that benefit of reserves is the adjusting cost from deficit
of balance of payments and cost of reserves is the opportunity
cost. The optimum reserves level is the one at which marginal
cost and marginal benefit of reserves are equal.
2.5.2. The model of Frankel and Jovanovic (1981)
According to Frankel and Jovanovic (1981), the model has two
independent variables, including the random volatility of reserves
(benefit of reserves) and the lost income (cost of reserves). Some
researches applying this model belong to Ramachandran (2004),

Silva and Da Silva (2004), Hee-Ryang Ra (2007), Shijaku (2012),
Sinem and Nebiye (2014).
2.5.3. The model of Ben-Bassat and Gottlieb (1992)
Ben-Bassat and Gottlieb (1992) take sovereign risk into the model
and present that central bank always tries to minimize the total
cost of holding reserves through the following function:
=

0

+ (1 - π) C1

(2.1)

with:
is the expected total cost of reserves; 0 is the loss due
to country’s default when reserves amount is equal to zero; C1 is
the loss of income (opportunity cost) when reserves amount is
higher than zero; is the probability of default (probability of
being zero for reserves and (1– π) is the probability of being
higher than zero for reserves.
To minimize the total cost, the derivative of equation (2.1) under
reserves must be zero and then the optimum reserves is displayed:
*

=
R

+

0

(2.14)

with C0, r, và R in turn are the loss due to country’s default,
opportunity cost, probability of default and marginal probability
of default.


14
The researches applying the model of Ben-Bassat and Gottlieb
(1992) include the research of Ozyildirim and Yaman (2005)
estimating optimum reserves for Turkey; the research of Tecnica
(2012) experimenting for Colombia; the research of Prabheesh
(2013) experimenting for India; the research of Tule et al (2015)
experimenting for Nigeria. At present, Vietnam has not yet any
researches on this method.
2.5.4. Remarks to build the model in Vietnam
Firstly, the model of Ben-Bassat and Gottlieb (1992) adds the
factor of sovereign risk and it has not yet experimented in
Vietnam, so it is suitable for carrying out this model in Vietnam.
Secondly, to be united with the model based on the determinants
of reserves, in this model, opportunity cost is represented by the
lending rate of VND.
Thirdly, basing on the crisis in 2008 which decreases the speed of
GDP growth in Vietnam during following years, the author
calculates the loss of GDP as proxy for the loss due to country’s
default
Fourthly, risk premium is calculated by the interest rate including
risk (the lending rate of VND) and the interest rate without risk

(3-month LIBOR rate of USD).
Fifthly, through review of related researches, the risk premium
function includes the independent variables of trade openness,
volatility of foreign portfolio investment, the ratio of short-term
external debt/ reserves and the ratio of fiscal deficit/ GDP.
Sixthly, the thesis applies the modern ways of calculation such as
calculating volatility of foreign portfolio investment by ARCH
model, calculating the loss due to country’s default by HP Filter
and using ARDL regression for the risk premium function.


15

CHAPTER 3: MODEL OF ESTIMATING
VIETNAMESE OPTIMUM RESERVES
3.1. For the method based on rule of thumb
3.1.1. Traditional methods
The author applies three popular estimation methods in the world
for Vietnam, including the method based on import turnover,
based on short-term external debt and based on broad money
supply M2 with the optimum reserves standard acknowledged by
IMF in turn as import turnover in 3 months, 100% of short-term
external debt and 20% of M2 (IMF, 2011). For the method based
on GDP, its optimum standard has not yet acknowledged by IMF,
so the author does not apply it for Vietnam.
3.1.2. Methods based on linking traditional methods
The author does not apply them for Vietnam because they are not
reasonable or calculation is done coincidentally when estimating
optimum reserves.
3.1.3. IMF’s ARA EM method

Vietnam is the nation under the fixed exchange rate regime (soft
pegs) (IMF, 2018). At the same time, Vietnam also controls
seriously capital inflows and outflows. Therefore, in the ARA EM
method, the model of estimating optimum reserves for Vietnam:
R* = 30% STED + 20% OPL + 5% M2 + 10% X
3.1.4. Research data
Yearly data is collected during the period of 2005 – 2017 with six
types of data, including reserves, import turnover, short-term
external debt, broad money supply M2, export turnover and
foreign portfolio investment from the sources like IFS,
Worldbank, CEIC Data, Bloomberg.


16
3.2. For the method based on the determinants of reserves
3.2.1. Building the empirical model for Vietnam
According to the model of Edison (2003) and related researches,
the empirical model for Vietnam is displayed as follows:
lnrest = β1 + β2lngdpt + β3opent + β4expvt + β5fpit + β6ervt +
β7costt + ut
in which: β1: intercept; β2,..., β7: the coefficients of independent
variables; u: error and t: denotes time (quarterly).
3.2.2. Method of determining variables in the empirical model
lnres: the natural logarithm of reserves as proxy for reserves
demand or optimum reserves.
lngdp: the natural logarithm of GDP as proxy for economy scale,
expected to have the positive relationship with reserves.
open: trade openness is measured by the ratio of import turnover/
GDP, as proxy for vulnerability of current account, expected to
have the positive relationship with reserves.

expv: export volatility is calculated by the standard deviation of
export change ratio with the formula of (Yt – Yt-4)/Yt-4 and
application of ARCH model, as proxy for vulnerability of current
account, expected to have the negative relationship with reserves.
fpi: financial openness is measured by the ratio of foreign
portfolio investment, as proxy for vulnerability of financial
account, expected to have the positive relationship with reserves.
erv: exchange rate volatility is calculated by the standard
deviation of natural logarithm of exchange rate with applying
ARCH model, as proxy for flexibility of exchange rate, expected
to have the positive relationship with reserves.
cost: the lending rate of VND as proxy for opportunity cost,
expected to have the negative relationship with reserves.


17
3.2.3. Process of carrying out the empirical model
Step 1: Calculating variables in the empirical model: export
volatility and exchange rate volatility are calculated by ARCH
model.
Step 2: Descriptive statistic of variables in the empirical model.
Step 3: Stationarity test for all variables in the empirical model:
find the optimum lag and take stationarity test at the optimum lag
by ADF method.
Step 4: Choosing regression method in determining reservesdemand function: after taking stationarity test in fact, all variables
are stationary series, so it is suitable for applying OLS regression
to determine reserves-demand function.
Step 5: Testing all coefficients and the suitability of the empirical
model: use three significant level of 1%, 5% and 10% for testing
as well as analyze R2.

Step 6: Diagnostic tests for the empirical model: Test of
multicollinearity, heteroscedasticity and autocorrelation.
Step 7: Estimating the optimum reserves and comparing with the
actual reserves.
3.2.4. Research data
Data is collected during the period of 2005 – 2017 with quarterly
frequency to raise the accuracy of the model. Kinds of data
include reserves, nominal GDP, import turnover, export turnover,
foreign portfolio investment, exchange rate of VND/USD, lending
rate of VND from the sources like IFS, GSO, Bloomberg, The
State Bank.
3.3. For the method based on the cost – benefit of reserves
3.3.1. The empirical model for Vietnam


18
According to sub-section 2.5.4, the model of Ben-Bassat and
Gottlieb (1992) is chosen for experimenting in Vietnam and it is
the equation (2.14) with four variables.
3.3.2. Method of determining variables in the empirical model
3.3.2.1. Determining the opportunity cost
According to sub-section 2.5.4, the opportunity cost is represented
by the lending rate of VND.
3.3.2.2. Determining the loss due to country’s default
Basing on the speed of Vietnamese GDP growth decreased after
the crisis in 2008, the author finds that the period of decreased
speed of GDP growth is the period of 2008 - 2012. Then, HP
Filter method is used for estimating the potential GDP amount,
considered as GDP which Vietnam can obtain if there is no crisis
in 2008. At last, the author calculates the difference between

actual GDP and potential GDP in the period of 2008 – 2012 with
the decreased speed of GDP growth. In fact, when calculating
potential GDP, it shows that the decreasing effect lasts up to
Q.2/2013. Therefore, the decreasing period used for calculating
the difference is the period of Q.1/ 2008 – Q.2/ 2013. This
difference is considered as the loss due to country’s default for
Vietnam.
3.3.2.3. Determining the risk premium model to calculate
probability of default and marginal probability of default
The risk premium model includes:
• The dependent variable is the natural logarithm of risk
*

premium, calculated by the formula:

*

(3.3)

with i is the interest rate for lending risky countries, measured by
the lending rate of VND and i* is the non-risk lending rate,
represented by 3-month LIBOR rate of USD.


19
• The independent variables include trade openness measured by
import turnover/ GDP, the volatility of foreign portfolio
investment calculated by the standard deviation in the ARCH
model, the ratio of short-term external debt/ reserves and the ratio
of fiscal deficit/ GDP.

The risk premium equation is displayed as follows:
=

t

=

0

+

1

t

+

2

t

+

3

t

+

4

t

+

t

(3.4)

in which:
t;
t ;
t ;
t ;
t and t are the
natural logarithm of risk premium, trade openness, volatility of
foreign portfolio investment, natural logarithm of the ratio of
short-term external debt/ reserves, the ratio of fiscal deficit/ GDP
and the error of regression equation respectively.
3.3.2.4. Process of carrying out the risk premium model to
calculate probability of default and marginal probability of
default
Step 1: Calculating variables in the risk premium model
Step 2: Descriptive statistic of variables in the risk premium
model
Step 3: Stationarity test for all variables in the risk premium model
Step 4: Selecting the regression method to determine the model:
After taking stationarity test in fact, variables are not stationary at
the same order, so it is suitable for applying ARDL regression to
determine the risk premium model.

Step 5: Testing the coefficients of the risk premium model
Step 6: Tests for ensuring the credibility of the model
including bounds test, heteroscedasticity, autocorrelation, white
noise of residuals, cumulative sum of square of recursive residuals.
Step 7: Calculating probability of default and marginal
probability of default.


20
Probability of default (π) is calculated by the equation:
f

=

1+

f

Marginal probability of default (πR) is calculated by taking
derivative under reserves of the equation f – equation (3.4).
3.3.3. Estimation of Vietnamese optimum reserves
Four determined variables are taken into the equation (2.14) to
estimate Vietnamese optimum reserves and compare with the
actual reserves.
3.3.4. Research data
Data is collected during the period of 2005 – 2017 with the
quarterly frequency, including reserves, nominal GDP, import
turnover, foreign portfolio investment, short-term external debt,
fiscal deficit, lending rate of VND, 3-month LIBOR rate of USD
from the sources such as IFS, GSO, Bloomberg, ADB,

Worldbank, ICE, The Ministry of Finance, However, for shortterm external debt, data is converted yearly series into quarterly
series by using the interpolation method.


21

CHAPTER 4: RESULTS OF ESTIMATING
VIETNAMESE OPTIMUM RESERVES
4.1. Real situation of Vietnamese reserves
4.1.1. Scale of Vietnamese reserves
In general, during the period of 2005 – 2017, Vietnamese reserves
increased continuously, except in the period of 2009 – 2011,
reserves decreased sharply because The State Bank had to use the
reserves for stabilizing the forex market under the influence of the
crisis in 2008. Until the end of 2017, Vietnamese reserves was
over 49 billion USD.
4.1.2. Structure of Vietnamese reserves
The structure of Vietnamese reserves includes four elements but it
can eliminate the element of reserves position at IMF because its
proportion to total reserves is similar to zero. Therefore,
Vietnamese reserves includes foreign currencies and kinds of
valuable papers take the much high proportion of approximately
97%; gold and SDR take the low proportion of about 3% and in
which the proportion of gold is higher than SDR. The advantage
of reserves assets in strong foreign currencies is that it can
immediately meet the necessary economic activities which have
to use reserves. Accordingly, foreign currencies and kinds of
valuable papers should continue to become the priority element in
Vietnamese reserves.
4.2. Results of the method based on rule of thumb

4.2.1. Traditional methods
4.2.1.1. The method based on import turnover
Chart 4.3. indicates that in the period of 2005 – 2008, Vietnamese
actual reserves is higher than the optimum reserves but in the
following period of 2009 – 2017, with the influence of the crisis
in 2008, reserves is always lower than the optimum reserves.


22
From 2012, Vietnamese economy starts to recover again and
although the yearly average speed of reserves growth in this
period is fairly high at 18.69%, it surpasses a little the yearly
average speed of import growth at 15.56%. Hence, reserves has
not yet caught import. In the end of research period – 2017, the
actual reserves is at 2.7 months of import turnover and nearly
reaches the optimum level.
60,000,000,000

USD

50,000,000,000
40,000,000,000
30,000,000,000
20,000,000,000
10,000,000,000
0
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

ACTUAL RESERVES

OPTIMUM RESERVES

Chart 4.3. Vietnamese actual reserves and optimum reserves
based on import turnover in the period of 2005 – 2017
Source : IFS (2018) and author’s calculation
4.2.1.2. The method based on short-term external debt
60,000,000,000

USD

50,000,000,000
40,000,000,000
30,000,000,000
20,000,000,000
10,000,000,000
0

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

ACTUAL RESERVES

OPTIMUM RESERVES

Chart 4.5. Vietnamese actual reserves and optimum reserves
based on short-term external debt in the period of 2005 – 2017
Source : IFS, World Bank (2018) and author’s calculation


23

Chart 4.5 shows that Vietnamese reserves always surpasses the
optimum level during all period of 2005 – 2017. It happens
because Vietnamese short-term external debt in this period is not
much and reserves is enough ability to finance 100% short-term
debt. However, this debt is increasing sharply at present and in the
future. Specifically, in 2017, the speed of short-tern external debt
growth is 56.34% to last year. Therefore, although actual reserves
is higher than the optimum level, it has not yet completely safe
and it should continue to increase reserves for being enough
ability to finance short-term external debt in the future.
4.2.1.3. The method based on broad money supply M2
80,000,000,000
70,000,000,000

USD

60,000,000,000
50,000,000,000
40,000,000,000
30,000,000,000
20,000,000,000
10,000,000,000
0

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

ACTUAL RESERVES

OPTIMUM RESERVES

Chart 4.7. Vietnamese actual reserves and optimum reserves
based on broad money supply M2 in the period of 2005 – 2017
Source : IFS, CEIC Data (2018) and author’s calculation
Chart 4.7. indicates that in the period of 2005 – 2008, the actual
reserves exceeds the optimum level. However, in the period of
2009 – 2017, the actual reserves is much lower than the optimum
reserves. Although reserves recovers with the strong yearly
average speed of growth at 18.69% during this period but this
speed is not much higher than the average speed of M2 growth at
16.10% , so reserves cannot catch up M2. This makes during the
time of after crisis in 2008, the actual reserves is always lower